X-FAB (XFAB.PA) Exposed to growing markets, but fairly valued. 16 May 2017 Europe/France Equity Research Semiconductor Equipment

Transcription

1 Europe/France Equity Research Semiconductor Equipment Rating NEUTRAL [V] Price (11 May 17, ) 8.11 Target price ( ) 8.15 Market Cap ( m) 1,060.6 Enterprise value ( m) Target price is for 12 months. [V] = Stock Considered Volatile (see Disclosure Appendix) Research Analysts Share price performance Achal Sultania achal.sultania@credit-suisse.com Quang Tung Le quangtung.le@credit-suisse.com A p r A p r A p r A p r M ay XFAB.PA CAC 4 0 INDEX The price relative chart measures performance against the CAC 40 INDEX which closed at on 11/05/17 On 11/05/17 the spot exchange rate was 1/Eu 1.- Eu.92/US$1 Performance 1M 3M 12M Absolute (%) -2.9 Relative (%) -8.4 X-FAB (XFAB.PA) INITIATION Exposed to growing markets, but fairly valued Initiating with a Neutral rating and a TP of X-FAB is a pure-play foundry specialising in mixed-signal analog chips, where it is the No. 5 player in the analog space. We expect X-FAB's earnings to rise from US$0.37 in 2017E to US$0.73 in 2019E, driven by its exposure to Auto/Industrial markets and margin expansion. However, we see risk-reward as fairly balanced at current levels, and hence initiate with a Neutral rating. Exposed to growing end markets. 1) X-FAB has ~50% of sales exposed to the Auto/Industrial markets, where we see potential for a high-single-digit semis-related revenue CAGR driven by rising semis content. 2) We expect foundry revenues will continue to outgrow the semis market, driven by an ongoing shift to fabless (foundry as a percentage of semis up from 10% in 2011 to 16% in 2016). 3) We expect analog foundries to record a ~10% CAGR, driven by increasing semis content and limited pricing pressure. We model X-FAB's sales to grow to US$580mn (+13% yoy; +10% organic) in 2017E, followed by 9% pa, reaching $690mn in 2019E. This is lower than X- FAB's target of over $750mn sales by 2019 due to a sharper-than-expected decline in an outsourced contract (we assume the end customer to be Goodix in China; $107mn revenue in 2016 to decline to c$20mn in 2017E). Utilisation should drive margin expansion given fixed cost structure. Our capacity analysis suggests that X-FAB could have enough fab capacity to deliver ~US$755mn of sales by 2019E. With management targeting ~90% utilisation by then (up from 80% in Q416), and with two-thirds of total costs being fixed, we think gross margins could rise 600bp over the next three years, driving EBITDA margin from 20% in 2016 to 26% in 2019E. Risks and catalysts ahead. 1) Semis slowdown could lead to downside to our 10%+ CAGR estimate for specialty analog foundries; 2) Customer concentration risk with Melexis accounting for one-third of sales; 3) USD depreciation vs. EUR and MYR is a potential headwind. In terms of catalysts, Q217 results are likely to be around the third week of August. TP of We value X-FAB using an average of i) P/E of 11.0x and ii) EV/Sales of 1.50x on our 2019 estimates, discounted back by one year at 8% and adding back net cash per share of US$1.5 or 1.4 (ending 2017). Financial and valuation metrics Year 12/16A 12/17E 12/18E 12/19E Revenue (US$ m) EBITDA (US$ m) Adjusted net income (US$ m) CS EPS (adj.) (US$) Prev. EPS (US$) ROIC (%) P/E (adj.) (x) P/E rel. (%) EV/EBITDA (x) Dividend (12/17E, US$) 0.00 Net debt/equity (12/17E,%) Dividend yield (12/17E,%) 0.0 Net debt (12/17E, US$ m) BV/share (12/17E, US$) 4.8 IC (12/17E, US$ m) Free float (%) 40.8 EV/IC (12/17E, (x) 2.2 Source: Company data, Thomson Reuters, Credit Suisse estimates DISCLOSURE APPENDIX AT THE BACK OF THIS REPORT CONTAINS IMPORTANT DISCLOSURES, ANALYST CERTIFICATIONS, LEGAL ENTITY DISCLOSURE AND THE STATUS OF NON-US ANALYSTS. US Disclosure: Credit Suisse does and seeks to do business with companies covered in its research reports. As a result, investors should be aware that the Firm may have a conflict of interest that could affect the objectivity of this report. Investors should consider this report as only a single factor in making their investment decision.

2 X-FAB (XFAB.PA) Price (11 May 2017): 8.11; Rating: NEUTRAL; Target Price: 8.15; Analyst: Achal Sultania Income statement (US$ m) 12/16A 12/17E 12/18E 12/19E Revenue EBITDA Depr. & amort. (50) (56) (66) (77) EBIT Net interest exp. (8) (7) (5) (5) Associates PBT Income taxes 4 0 (3) (4) Profit after tax Minorities Preferred dividends Associates & other Net profit Other NPAT adjustments Reported net income Cash flow (US$ m) 12/16A 12/17E 12/18E 12/19E EBIT Net interest (8) (7) (5) (5) Cash taxes paid Change in working capital (16) (10) (7) (8) Other cash and non-cash items Cash flow from operations CAPEX (72) (102) (128) (122) Free cashflow to the firm (0) (9) 1 42 Acquisitions (0) Divestments Other investment/(outflows) (10) Cash flow from investments (82) (100) (126) (120) Net share issue/(repurchase) Dividends paid Issuance (retirement) of debt Cashflow from financing Changes in net cash/debt (60) Net debt at start - 60 (195) (201) Change in net debt 60 (254) (7) (47) Net debt at end 60 (195) (201) (249) Balance sheet (US$ m) 12/16A 12/17E 12/18E 12/19E Assets Total current assets Total assets ,083 Liabilities Total current liabilities Total liabilities Total equity and liabilities ,083 Per share 12/16A 12/17E 12/18E 12/19E No. of shares (wtd avg.) (mn) CS EPS (adj.) (US$) Dividend (US$) Free cash flow per share (US$) (0.00) (0.07) Key ratios and valuation 12/16A 12/17E 12/18E 12/19E Growth/Margin (%) Sales growth (%) EBIT growth (%) Net income growth (%) EPS growth (%) - (20.0) EBITDA margin (%) EBIT margin (%) Pretax profit margin (%) Net income margin (%) Valuation 12/16A 12/17E 12/18E 12/19E EV/Sales (x) EV/EBITDA (x) EV/EBIT (x) Dividend yield (%) P/E (x) Credit ratios (%) 12/16A 12/17E 12/18E 12/19E Net debt/equity (%) 18.5 (30.7) (28.4) (30.7) Net debt to EBITDA (x) 0.6 (1.8) (1.4) (1.4) Interest coverage ratio (x) Source: FTI, Company data, Thomson Reuters, Credit Suisse Securities (EUROPE) LTD. Estimates Company Background X-FAB is a pure-play foundry specialising in mixed-signal analog chips used in the semiconductor market across the Auto, Industrial, communications, Consumer and Medical segments. Blue/Grey Sky Scenario Our Blue Sky Scenario ( ) In our blue sky scenario, we assume X-Fab sees robust sales growth driving 25% potential upside to our 2019 earnings estimate. Using an average of 12x P/E and 2.0x EV/sales on 2019 estimates, then discounting it back by 1 year and adding net cash per share (ending 2017E), we arrive at a valuation of Our Grey Sky Scenario ( ) 5.30 In our grey sky scenario, we assume X-Fab sees slowdown in sales resulting in -30% potential downside to our 2019 earnings estimate. Using an average of 8.5x P/E and 1.0x EV/sales on 2019 estimates, then discounting it back by 1 year and adding net cash per share (ending 2017E), we arrive at a valuation of Share price performance Apr- 17 Apr- 17 Apr- 17 Apr- 17 May- 17 XFAB.PA CAC 40 INDEX The price relative chart measures performance against the CAC 40 INDEX which closed at on 11/05/17 On 11/05/17 the spot exchange rate was 1/Eu 1.- Eu.92/US$1 X-FAB (XFAB.PA) 2

3 Table of contents Exposed to growing markets, but fairly valued 1 Key charts 4 Investment thesis 6 Key positives 8 Key risks 10 Valuation TP of Demand trends for the Semiconductor and Foundry market 21 Growth drivers for X-FAB's end markets 25 Foundry market structure Digital vs. Specialty Analog vs. IDM 39 X-FAB company background 45 X-FAB financials focus on revenue growth and margin expansion 56 Financial models 70 Credit Suisse PEERs 73 X-FAB (XFAB.PA) 3

4 Key charts Key charts foundry market Figure 1: Scenario analysis: If specialty analog foundry revenues as a % of industry revenues continue to rise at the same pace as in the past, we expect a revenue CAGR of 11-12% over E Revenues (US$ mn) E 2018E 2019E 2020E CAGR CAGR E Global semiconductor industry revenues 306, , , , , , , , , , % 3.7% % change yoy 1.8% -2.3% 5.2% 8.6% -2.3% 1.5% 7.2% 3.8% 0.8% 3.2% Semi TAM served by specialty analog foundries 67,465 64,711 63,265 69,339 69,463 72,623 76,242 79,455 82,250 86, % 4.5% % change yoy 1.1% -4.1% -2.2% 9.6% 0.2% 4.5% 5.0% 4.2% 3.5% 5.4% Top5 speciality analog foundries 2,250 2,697 2,544 3,154 3,271 3,956 4,554 5,053 5,422 5, % 10.7% % change yoy 6.0% 19.8% -5.7% 24.0% 3.7% 20.9% 15.1% 11.0% 7.3% 9.5% as % of semi industry revenues 0.73% 0.90% 0.81% 0.92% 0.98% 1.16% 1.25% 1.34% 1.42% 1.51% Top5 speciality analog foundries 2,250 2,697 2,544 3,154 3,271 3,956 4,475 4,999 5,523 6, % 11.8% % change yoy 6.0% 19.8% -5.7% 24.0% 3.7% 20.9% 13.1% 11.7% 10.5% 12.0% as % of specialty analog foundries core TAM 3.34% 4.17% 4.02% 4.55% 4.71% 5.45% 5.87% 6.29% 6.71% 7.14% Source: Gartner, Company data, Credit Suisse estimates *Core TAM served by specialty analog foundries includes analog, discrete, ASIC and Non-Optical Sensor products in the overall semiconductor market Figure 2: Pure-play foundries have a significant share in the foundry market over IDMs Foundry (US$ mn) TSMC (Taiwan) 13,307 14,299 16,951 19,935 25,138 26,574 29,488 GlobalFoundries (U.S.) 3,510 3,195 4,013 4,122 4,355 5,019 5,545 UMC Group (Taiwan) 3,965 3,760 3,730 3,959 4,331 4,464 4,582 SMIC (China) 1,555 1,320 1,542 1,962 1,970 2,236 2,921 Powerchip (Taiwan) ,175 1,291 1,268 1,275 TowerJazz (Israel) ,250 Vanguard (Taiwan) Hua Hong Semi (China) Dongbu HiTek (S.Korea) X-FAB (Europe) Others (Pure-Play) 1,802 1,543 2,490 2,280 2,405 2,238 Total Pure-Play Foundries 27,000 31,100 36,186 42,519 45,237 50,005 % change yoy 15% 16% 18% 6% 11% Total IDM Foundries 4,910 6,520 6,614 5,100 5,200 5,595 % change yoy 33% 1% -23% 2% 8% Total Pure-Play + IDM Foundries 31,910 37,620 42,800 47,619 50,437 55,600 % change yoy 18% 14% 11% 6% 10% Source: Company data, IC Insights, Credit Suisse research Figure 3: Pure-play foundry by end market ($50bn in 2016) Figure 4: Pure-Play foundry by feature size ($50bn in 2016) ( ) = 2010 share Source: X-FAB data (IC Insights, McClean Report 2017) Source: X-FAB data (IC Insights, McClean Report 2017) X-FAB (XFAB.PA) 4

5 Key charts X-FAB Figure 5: Our capacity utilisation analysis suggests that at a 90% utilisation rate, X-FAB has the potential to deliver sales of around $690m in 2019 Capacity (WSPM in '000 of 200mm equivalent) E 2018E 2019E Erfurt (Germany) since Lubbock (USA) since Kuching-Sarawak (Malaysia) since Dresden (Germany) since Corbeil-Essonnes (France) since Itzehoe (Germany) since 2011 (fully owned since 2015) - MEMS Total WSPM (200 mm equivalent) capacity at the end of the year ( 000) Utilization rate (%) 47% 55% 66% 68% 79% 84% 87% 90% Total WSPM (200mm equivalent) produced at target utilization rate % change yoy 17% 20% 13% 36% 48% 7% 6% Total X-FAB sales (US$ mn) Total X-FAB sales excl. Outsourcing (US$ mn) % change yoy 12% 14% 0% 27% 34% 13% 9% Source: Company data, Credit Suisse estimates Figure 6: Around two-thirds of X-FAB's COS is fixed, with one-third variable Figure 7: X-FAB has scope for GM improvement as capacity utilization in fab goes up 100% 80% 60% 40% 20% 0% 28% 27% 25% 47% 46% 49% 25% 26% 26% Total fixed costs Total variable costs Fixed COS (Facility costs + D&A) Fixed & Partly Variable COS (Employee-related + Fixed asset costs) Variable COS (excl. Outsourcing) (Material + Other costs) Gross Margin (%) 45% 40% 35% 30% 25% 20% 15% 10% 5% 0% 26.5% 20.5% 8.5% E 2018E 2019E TowerJazz Vanguard Dongbu HiTek Hua Hong Semi Average for X-FAB peers X-FAB Source: Company data, Credit Suisse estimates Source: Company data, Credit Suisse estimates, Thomson Reuters (for 2017/2018 estimates for TowerJazz and Dongbu HiTek), Credit Suisse estimates (for 2017/2018 estimates for Vanguard and Hua Hong Semi, which are covered by CS analyst Randy Abrams in Taiwan) Figure 8: We apply 11x P/E to 2019 estimates for X- FAB, discount it back for 1 year and add net cash Implied valuation 2019E grey sky blue sky Sales (US$ mn) Net profit (US$ mn) EPS ($) P/E (x) Implied fair value for X-Fab (US$) discounting back by 1 year at 8% 8% 8% 8% Implied TP for X-Fab (US$) (excl. net cash) Net cash (ending 2017E) (US$ mn) Net cash per share (US$) Implied TP for X-Fab (US$) (incl. net cash) FX (EUR / USD) rate Implied TP for X-Fab ( ) (incl. net cash) Source: Company data, Credit Suisse estimates Figure 9: We apply 1.50x EV/sales to 2019 estimates for X-FAB and discount it back for 1 year Implied valuation 2019E grey sky blue sky Sales (US$ mn) EBIT EBIT margin (%) 15.1% 10.8% 18.8% EV / Sales (x) Implied EV (US$ mn) 1, ,474 discounting back by 1 year at 8% 8% 8% 8% Implied EV discounted to 2018E (US$ mn) ,365 Net cash (ending 2017E) (US$ mn) Implied market cap (US$ mn) 1, ,560 Share count (mn) Implied TP for X-Fab (US$) FX (EUR / USD) rate Implied TP for X-Fab ( ) Source: Company data, Credit Suisse estimates X-FAB (XFAB.PA) 5

6 Investment thesis Initiating with a Neutral rating and a TP of 8.15 X-FAB is a pure-play foundry specialising in mixed-signal analog chips used in the semiconductor market across the Auto, Industrial, Communications, Consumer and Medical segments. While X-FAB has only a 1% revenue share in the overall foundry market (given the size of large digital foundries), we note it is the No. 5 player in the analog space with a 13% share within the top five analog foundries as of 2016 (TowerJazz at 32%, Vanguard at 20%, Hua Hong Semi at 18% and Dongbu HiTek at 17%). X-FAB generated $513mn in sales in 2016 (including $32mn from the Altis fab acquisition in Q416 out of full-year Altis sales of $122mn in 2016) with 10% EBIT margins. We expect X-FAB's EBIT margins to rise to 15% by 2019E, driving EBITDA margins of 26% over time. This would result in EPS rising from US$0.37 in 2017E to US$0.73 in 2019E. However, we see risk-reward as fairly balanced at current levels, and hence initiate with a Neutral rating and a TP of Figure 10: X-FAB financials: We expect EBITDA margins to improve to 26% by 2019E US$ ' E 2018E 2019E X-FAB sales 258, , , , , , , ,004 X-FAB sales (excl. Outsourcing) 258, , , , , , , ,004 Gross Profit (7,200) 24,700 51,518 57, , , , ,851 GM (%) -2.8% 8.5% 15.6% 17.5% 20.5% 21.7% 24.2% 26.5% Gross Profit (excl. Outsourcing) (7,200) 24,700 51,518 57,842 88, , , ,851 GM (%) -2.8% 8.5% 15.6% 17.5% 21.1% 21.9% 24.2% 26.5% Total opex (43,200) (44,400) (45,746) (45,705) (57,003) (72,450) (75,992) (79,641) as % of sales 16.7% 15.3% 13.8% 13.8% 11.1% 12.5% 12.0% 11.5% as % of sales (excl. Outsourcing) 16.7% 15.3% 13.8% 13.8% 13.6% 12.9% 12.0% 11.5% Other income (1,100) 4,000 2,918 4,134 2, EBIT (51,500) (15,700) 8,690 16,271 50,456 54,500 78, ,111 EBIT margin (%) -19.9% -5.4% 2.6% 4.9% 9.8% 9.4% 12.4% 15.1% EBIT (excl. Outsourcing) (51,500) (15,700) 8,690 16,271 33,762 51,100 78, ,111 EBIT margin (excl. Outsourcing) (%) -19.9% -5.4% 2.6% 4.9% 8.1% 9.1% 12.4% 15.1% D&A 73,000 57,300 46,465 47,165 50,158 55,680 66,381 76,590 as % of sales 28.2% 19.8% 14.1% 14.2% 9.8% 9.6% 10.5% 11.1% as % of sales (excl. Outsourcing) 28.2% 19.8% 14.1% 14.2% 12.0% 9.9% 10.5% 11.1% EBITDA 21,500 41,600 55,155 63, , , , ,701 EBITDA margin (%) 8.3% 14.3% 16.7% 19.2% 19.6% 19.0% 22.9% 26.2% EBITDA (excl. Outsourcing) 21,500 41,600 55,155 63,436 83, , , ,701 EBITDA margin (excl. Outsourcing) (%) 8.3% 14.3% 16.7% 19.2% 20.0% 19.1% 22.9% 26.2% Source: Company data, Credit Suisse estimates Exposed to growing end markets Exposed to growing end markets. X-FAB has ~50% of sales exposed to the Auto/Industrial markets, where we see potential for a high-single-digit semis-related revenue CAGR driven by rising semis content due to increasing focus around energy efficiency, automation, connectivity and security. Foundry market has been growing faster than the overall semiconductor market. We believe that foundry revenues will continue to outgrow the broader semis market, driven by an ongoing shift to fabless (foundry as a percentage of semis up from 10% in 2011 to 16% in 2016). Within the foundry market, we expect analog foundries to record a ~10% CAGR, driven by increasing semis content and limited pricing pressure. We model X-FAB's sales to grow to US$580mn (+13% yoy; +10% organic) in 2017E, followed by 9% pa, reaching $690mn in 2019E. This is lower than X-FAB's target of over $750mn sales by 2019 given the decline in an outsourcing contract for fingerprint sensors (we assume the end- X-FAB (XFAB.PA) 6

7 customer is Goodix) for which we estimate the revenue to decrease from $107mn (of which $94mn was outsourced) in 2016 to around $20mn in 2017; the decline appears to be sharper than expected, based on comments during the Q1 earnings call (see page 57 for further details). Utilisation should drive margin expansion given fixed cost structure Our capacity analysis suggests that X-FAB could have enough fab capacity to deliver ~US$755mn of sales by 2019E. With management targeting ~90% utilisation by then (up from 80% in Q416), and with two-thirds of total costs being fixed, we think gross margins could rise 600bp over the next three years from 20.5% in 2016 to 26.5% by 2019E. This would help EBITDA margins improve from 20% in 2016 to 26% in 2019E. Figure 11: X-FAB aiming to get to 90% capacity utilisation by 2019E Figure 12: X-FAB's GMs are well below peer group levels due to lower utilisation levels Capacity Utilization (%) 120% 100% 80% 60% 40% 20% 75% 83% 79% 81% 47% 100% 84% 55% 102% 94% 66% 82% 90% 68% 89% 98% 79% 84% 87% 90% Gross Margin (%) 45% 40% 35% 30% 25% 20% 15% 10% 5% 8.5% 20.5% 26.5% 0% E 2018E 2019E Vanguard Hua Hong Semi X-FAB 0% E 2018E 2019E TowerJazz Vanguard Dongbu HiTek Hua Hong Semi Average for X-FAB peers X-FAB Source: Company data, Credit Suisse research Source: Company data, Credit Suisse estimates, Thomson Reuters (for 2017/2018 estimates for TowerJazz and Dongbu HiTek), Credit Suisse estimates (for 2017/2018 estimates for Vanguard and Hua Hong Semi, which are covered by CS analyst Randy Abrams in Taiwan) Catalyst Q2 earnings around the third week of August Post recent Q1 earnings on 11 th May, X-FAB is scheduled to report its Q2 results around the third week of August. For Q117, the company delivered sales of $147.9mn (-7% qoq/+39% yoy/+12% yoy adjusted for X-FAB France acquisition). Within that, outsourcing revenues relating to Goodix were $15.5mn in Q1, which the company expects to decline to only $3-5mn in Q217. For Q2, X-FAB is guiding for $ mn of sales. We model $140mn for Q2, which implies 5% down qoq. However, when adjusted for the decline in outsourcing revenues, we forecast 3% qoq growth in Q2 sales. Valuation looks fairly balanced at current levels; TP of 8.15 We value X-FAB using an average of i) P/E of 11.0x and ii) EV/Sales of 1.50x on our 2019 estimates, discounting it back by one year at 8% and adding back net cash per share of US$1.5 or 1.4 (ending 2017). Figure 13: Our TP of 8.15 is based on an average of P/E and EV/sales multiples applied to our 2019 estimates Implied valuation (avg. of P/E and EV/Sales) 2019E grey sky blue sky Fair value based on P/E analysis Fair value based on EV/Sales analysis Implied TP for X-Fab ( ) Current share price ( ) % potential upside / (downside) 1% -35% 33% Potential upside/downside based on prices as of 11 th May 2017; Source: Company data, Credit Suisse estimates X-FAB (XFAB.PA) 7

8 Key positives Exposed to end market growing at 10%+ CAGR trend likely to continue Foundries growing much faster than the broader semis market. The emergence of fabless business models means that OEMs (original equipment manufacturers) like Apple, Google and several Chinese handset brands invest mainly in proprietary IC design without investing in manufacturing capabilities. Meanwhile, fabless companies such as Qualcomm, MediaTek and Melexis focus on IC design and testing while relying on foundries for manufacturing. As a result, the foundry market has been growing faster than the overall semiconductor market for the past few years. Data from IC Insights show that total foundry revenues grew from $32bn in 2011 to $56bn in 2016, implying a CAGR of 12%, which is well above the 2% revenue CAGR recorded by the overall semiconductor market over the same period. 10%+ CAGR possible for specialty analog foundries. The share of specialty analog foundry revenues (for the top five players) as a percentage of total semiconductor industry revenues has been increasing over time rising from 0.73% in 2011 to 1.16% in 2016 (Figure 37), which is close to a 10bp increase per annum. If specialty analog foundry revenues as a percentage of global semis revenues continue growing at 10bp pa, they would reach around 1.5% by 2020E, which would imply specialty analog foundry revenues of $5.9bn for the top five players in 2020E, compared with $4.0bn in 2016 (implying a CAGR of 11%). High Automotive and Industrial exposure bodes well for the long term. With X-FAB generating over 50% of its revenues from the Auto and Industrial segments, we see this as a positive for its end-market exposure given our view around rising silicon content in both these industries driven by energy efficiency, automation, connectivity and security. Over the next five years, we expect the Auto semis market to record a 7% CAGR (with potential upside to a 10% CAGR) and Industrial semis to see a high-single digit CAGR. Investing in upcoming growth areas. X-FAB has been increasing its capex investments in areas around SiC (silicon carbide), MEMS (sensors) and emerging applications in Medical, where it plans to continue investing over the next three years. This is also reflected in its healthy prototyping pipeline, which we see as a positive indicator for long-term revenue growth. Scope for material margin improvement Two-thirds of costs are fixed and one-third are variable, indicating significant leverage potential if the top-line grows. X-FAB has seen gross margin (GM) expansion in every year since 2012 having delivered 20.5% GMs in Over the same period, capacity utilisation at its fabs on a blended basis rose from 47% in 2012 to 79% in 2016 (80% for Q416 including the France fab). With capacity utilisation targeted to rise further to around 90% by 2019, we believe GMs will continue to expand to 26.5% by 2019E. X-FAB notes that around two-thirds of its total cost of sales (COS) is fixed (relating to personnel, D&A and facility costs) with the remaining third being variable (material expenses and some other items). Our detailed cost analysis suggests that total COS as a percentage of revenue could decline from 79% in 2016 to 73% by 2019E, thereby driving 600bps of improvement in group GMs. Even at 26.5% in 2019E, X-FAB's GM would be the lowest in its peer group. When we compare the GM structure at X-FAB to other specialty analog peers such as TowerJazz and Vanguard, we note that at a 20.5% GM in 2016, X-FAB was operating at well below the average of its peer group, at 32%. If X-FAB were to reach our forecast of 26.5% GMs for 2019E, it would still be below this peer group average. This shows the potential for GM improvement at X-FAB if capacity utilisation rises to the targeted level of ~90% by X-FAB (XFAB.PA) 8

9 EBIT and EBITDA margins should continue to improve. X-FAB's EBIT improved from a loss of $51.5mn in 2012 to $50.5mn in 2016 driven by revenue growth, which led to a GM improvement. For 2016, it delivered 10% EBIT margins, or 8% if we were to exclude the outsourcing contract. Driven by our view that GMs can rise by around 600bps along with a small improvement in the opex-to-sales ratio (as revenues grow), we see EBIT margins rising further to 15% by 2019E. Driven by higher capex, we model D&A rising to $77mn by 2019E (11% of sales), resulting in EBITDA margins of 26% in 2019E (vs. 20% in 2016). X-FAB (XFAB.PA) 9

10 Key risks A slowdown in the semiconductor industry would be a downside risk to our medium-term 10%+ CAGR estimate for the specialty analog foundry market We assume that the overall semiconductor market will see a revenue CAGR of 4% over E. Within this, we believe core markets served by specialty analog foundries (like X-FAB) will grow at 5% pa. With specialty analog foundries having grown at a significantly higher rate than the overall semis market, driven by an ongoing shift to fabless models, we believe X-FAB can record a 10% CAGR in the medium term. If there is any potential slowdown in the global macro environment, it would lead to a slowdown in the semiconductor market, which in turn would have an impact on growth rates for the analog foundry market. High revenue exposure to one customer (Melexis) Melexis accounted for 34% of X-FAB's sales in 2016, which could be a source of potential risk in case of socket loss. While X-FAB's revenues from Melexis have largely grown in line with the growth in Melexis' own revenues over the past few years (although the trend has not been linear in all years), we acknowledge the risk that X-FAB could lose new chip manufacturing contracts from Melexis, or Melexis could lose contracts from its key customers such as auto suppliers. In this case, X-FAB may not be able to offset such a loss with wins from other automotive customers. FX impact A USD depreciation versus the EUR and the MYR is a potential headwind given 40% of the company's cost base is in EUR and 10% in MYR, with only 11% of sales coming from EUR, as of Q (including X-FAB France). Potential integration issues with the recently acquired fab in France X-FAB (in Q416) acquired a fab in France from Altis Semiconductor for 9.2m. The company plans to transform this fab from mainly a communications & consumer foundry to increase its exposure to the automotive and industrial markets. As part of this, it plans to invest up to $60mn in capex over 2017 to In addition, for the French fab, management is targeting a capacity utilisation rate of around the group average by Q317 (up from a 70% rate in Q416). We see this as a potential risk given the integration issues associated with such programmes. X-FAB (XFAB.PA) 10

11 Valuation TP of 8.15 In this section, we look at the comparable valuation of peer group companies for X-FAB to arrive at a fair value for X-FAB shares. A wide valuation range for the peer group. If we look at the valuation metrics for these foundry companies over the past three years, we can see that the range is quite wide, with Vanguard trading at a significant premium to the average, and Dongbu HiTek trading at a material discount to the average if we look at P/E metrics. Figure 14: The valuation range for X-FAB's peer group has been quite wide P/E (x) EV/Sales (x) EV/EBITDA (x) FY1 FY2 FY3 FY1 FY2 FY3 FY1 FY2 FY3 Vanguard (current) TowerJazz (current) Hua Hong Semi (current) DongbuHiTek (current) Average (current) X-FAB (current, on CS estimates) Vanguard (avg. since 2014) TowerJazz (avg. since 2014) Hua Hong Semi (avg. since 2014) Dongbu HiTek (avg. since 2015) Average (last 3 year history) Source: Thomson Reuters (for Consensus), Credit Suisse research (Current valuation is based on closing price as of 10 th May 2017). Note X-FAB comps are based on CS estimates, whereas Vanguard, Tower Jazz, Hua Hong Semi and DongbuHiTek comps are based on Thomson Reuters consensus. We value X-FAB using an average of P/E and EV/Sales analyses. To value X-FAB, we use an 11.0x P/E multiple and a 1.50x EV/sales multiple to our 2019 estimates, discount it back by 1 year at 8% (given we are using 2019 estimates as opposed to 2018 or FY2 estimates) and adjust for the net cash position we estimate ending Note X-FAB's financials are reported in USD whereas its share price is quoted in EUR. This gives us a fair value of 8.2 for X-FAB using a P/E analysis and 8.1 using an EV/sales multiple, yielding an average of Figure 15: We apply 11x P/E to 2019 estimates for X- FAB, discount it back for 1 year and add net cash Implied valuation 2019E grey sky blue sky Sales (US$ mn) Net profit (US$ mn) EPS ($) P/E (x) Implied fair value for X-Fab (US$) discounting back by 1 year at 8% 8% 8% 8% Implied TP for X-Fab (US$) (excl. net cash) Net cash (ending 2017E) (US$ mn) Net cash per share (US$) Implied TP for X-Fab (US$) (incl. net cash) FX (EUR / USD) rate Implied TP for X-Fab ( ) (incl. net cash) Source: Credit Suisse estimates Figure 16: We apply 1.50x EV/sales to 2019 estimates for X-FAB and discount it back for 1 year Implied valuation 2019E grey sky blue sky Sales (US$ mn) EBIT EBIT margin (%) 15.1% 10.8% 18.8% EV / Sales (x) Implied EV (US$ mn) 1, ,474 discounting back by 1 year at 8% 8% 8% 8% Implied EV discounted to 2018E (US$ mn) ,365 Net cash (ending 2017E) (US$ mn) Implied market cap (US$ mn) 1, ,560 Share count (mn) Implied TP for X-Fab (US$) FX (EUR / USD) rate Implied TP for X-Fab ( ) Source: Credit Suisse estimates X-FAB (XFAB.PA) 11

12 Figure 17: Our TP of 8.15 is based on average of P/E and EV/sales multiple applied to our 2019 estimates Implied valuation (avg. of P/E and EV/Sales) 2019E grey sky blue sky Fair value based on P/E analysis Fair value based on EV/Sales analysis Implied TP for X-Fab ( ) Current share price ( ) % potential upside / (downside) 1% -35% 33% Source: Company data, Credit Suisse estimates (based on share price as of 11 th May 2017) Peer multiple comparison There are a number of international peers in the specialty analog pure-play foundry market that could be direct comparables for X-FAB. These include Vanguard (Taiwan-based), TowerJazz (Israeli foundry group, also listed in the US), Hua Hong Semiconductor (Chinabased) and Dongbu HiTek (Korea-based). While we acknowledge that these foundries are of different sizes (Figure 18) and some may specialise more in communication products as opposed to automotive, industrial and medical (which are X-FAB's main focus areas), we see them as direct comparables given that they manufacture analog mixed-signal chips. Revenue exposure by end-market for peer group for 2016 is as follows: Vanguard: Computing (30%), Communications (21%), Consumer (22%), Others (27%). TowerJazz: Power (28%), RF+Analog for Mobile (22%), RF+Analog for Infrastructure (9%), CMOS Image Sensors (16%), Others (25%). Hua Hong Semi: Consumer (66%), Communications (16%), Computing (6%), Industrial+Automotive (12%). X-FAB: Automotive (41%), Industrial (11%), Medical (3%), Communications & Consumer (42% including Outsourced part), Others (4%). Figure 18: Comparables for X-FAB: Vanguard, TowerJazz, Hua Hong Semi and Dongbu HiTek Company name Currency Share Price (local ccy) Market Cap (local ccy mn) Market Cap ($ mn) EV (local ccy mn) EV ($ mn) Vanguard Taiwan New $ ,894 2,971 71,913 2,377 TowerJazz US$ ,295 2,295 2,277 2,277 Hua Hong Semi Hong Kong $ ,166 1,434 9,118 1,171 Dongbu HiTek South Korean Won 21, , ,218,501 1,073 X-FAB Euro 8.1 1,068 1, ,083 Source: Thomson Reuters, Company data, Credit Suisse research (based on share price as of 10 th May 2017) X-FAB (XFAB.PA) 12

13 Vanguard International Semiconductor (VIS) is a Taiwan-based foundry currently trading on 15x P/E and 2.5x EV/sales on 2018 consensus estimates. The company offers wafer foundry services, such as CMOS, various types of ICs (including general logic ICs, highvoltage ICs, analog ICs, radio frequency ICs), as well as multiple project wafer (MPW) services, mask services and design services. VIS currently has three 8-inch fabs with a monthly capacity of approximately 187k wafers in Besides the display-related ICs, analog ICs (particularly power management ICs) and mixed signal ICs are key product areas for VIS. Figure 19: Vanguard is currently trading on 15x P/E and 2.5x EV/sales on 2018 consensus estimates Vanguard P/E Multiple EV/Sales Multiple EV/EBITDA Multiple Year FY1 FY2 FY3 FY1 FY2 FY3 FY1 FY2 FY YTD Current Average since Average since Average since Average since Source: Thomson Reuters, Company data, Credit Suisse research (based on share price as of 10 th May 2017) TowerJazz is another foundry, listed in both Israel and the US. Its shares are currently trading at 10x P/E and 1.5x EV/sales on 2018 consensus estimates. The company offers wafers of 150mm, 200mm and 300mm and its ICs are incorporated in various markets, such as communication and consumer electronics, PCs, automotive, industrial and medical. The company's manufacturing sites are located in Israel, the US and Japan, with Maxim (US semiconductor company) being its biggest customer. Figure 20: TowerJazz is currently trading on 10x P/E and 1.5x EV/sales on 2018 consensus estimates TowerJazz P/E Multiple EV/Sales Multiple EV/EBITDA Multiple Year FY1 FY2 FY3 FY1 FY2 FY3 FY1 FY2 FY YTD Current Average since Average since Source: Thomson Reuters, Company data, Credit Suisse research (based on share price as of 10 th May 2017) X-FAB (XFAB.PA) 13

14 Hua Hong Semiconductor is a China-based foundry group listed on the Hong Kong stock exchange. The company focuses mainly on manufacturing 200mm wafers for specialty applications, e-non-volatile memory (envm) and power discrete. It is currently trading on 10x P/E and 1.4x EV/Sales on consensus estimates for Figure 21: Hua Hong Semi is currently trading on 10x P/E and 1.4x EV/sales on 2018 consensus estimates Hua Hong Semiconductor P/E Multiple EV/Sales Multiple EV/EBITDA Multiple Year FY1 FY2 FY3 FY1 FY2 FY3 FY1 FY2 FY YTD Current Average since Average since Source: Thomson Reuters, Company data, Credit Suisse research (based on share price as of 10 th May 2017) Dongbu HiTek is based in South Korea and manufactures mainly fab system large-scale integration (LSI) wafers, which are used in signal-processing chips. It also produces LCD driver IC devices, CMOS image sensors and ASICs (application specific integrated circuits), among others. The shares are currently trading at 6x P/E and 1.4x EV/sales on consensus estimates for Figure 22: Dongbu HiTek is currently trading on 6x P/E and 1.4x EV/sales on 2018 consensus estimates Dongbu HiTek P/E Multiple EV/Sales Multiple EV/EBITDA Multiple Year FY1 FY2 FY3 FY1 FY2 FY3 FY1 FY2 FY YTD Current Average since Source: Thomson Reuters, Company data, Credit Suisse research (based on share price as of 10 th May 2017) X-FAB (XFAB.PA) 14

15 How does X-FAB compare with its peer group? Comparing X-FAB to the four analog foundries mentioned above, we note a few interesting observations around revenue and profitability stats. Production capacity at foundry and utilisation levels. TowerJazz and X-FAB have the highest number of fabs (7 and 6 respectively) vs. other players having 2 or 3 fabs. While we understand that capacity utilisation remains key to GM, we equally believe that having fabs close to one another has cost advantages, which is also evident from the lower GM structure at both TowerJazz and X-FAB versus other players such as Vanguard, Hua Hong Semi and Dongbu HiTek. In terms of wafer capacity, X-FAB's size was roughly a third of TowerJazz and Vanguard (around 190k WSPM for 200mm equivalent), but after the acquisition of the French fab, it is close to 100k WSPM of capacity. Figure 23: Production capacity at X-FAB and its key peer group over time Wafer starts per month ('000) Figure 24: X-FAB aiming to get to 90% capacity utilisation by 2019E Capacity Utilization (%) 120% 100% 80% 60% 40% 20% 75% 83% 79% 81% 47% 100% 84% 55% 102% 94% 66% 82% 90% 68% 89% 98% 79% 84% 87% 90% TowerJazz Vanguard Dongbu HiTek Hua Hong Semi X-FAB 0% E 2018E 2019E Vanguard Hua Hong Semi X-FAB Source: Company data, Credit Suisse research Source: Company data, Credit Suisse estimates (for X-FAB for 2017/2018/2019) Revenue per fab and Revenue per employee trends. We also look at metrics such as revenue per fab and revenue per employee. We notice that both these trends at X-FAB remain well below some of its peers. But we also believe that X-FAB should be able to close some of the gap if the company delivers on its mid-term objective of over $750mn of sales in In that scenario, revenue per fab at X-FAB would rise from $85mn in 2016 to $115mn in 2019E, but still be below the average of $255mn of sales per fab for its four peers for Figure 25: Revenue per fab at peer group foundries 400 Figure 26: Revenue per employee at peer group 350 Revenue per fab ($ mn) TowerJazz Vanguard Dongbu HiTek Hua Hong Semi X-FAB Revenue per employee ($ mn) TowerJazz Vanguard Dongbu HiTek Hua Hong Semi X-FAB Source: Company data, Credit Suisse research Source: Company data, Credit Suisse research X-FAB (XFAB.PA) 15

16 Margin profile for different players. Looking at gross margins and EBIT margins, again we see that X-FAB historicals look underwhelming compared with its peers, with X-FAB operating at a lower profitability than its four comparable peers. However, we expect X- FAB's GMs to rise from 20.5% in 2016 to 26.5% in 2019E, driven by higher sales and improved capacity utilisation. Figure 27: GM trends at analog foundries Figure 28: EBIT trends at analog foundries 45% 30% Gross Margin (%) 40% 35% 30% 25% 20% 15% 10% 5% 0% 26.5% 20.5% 8.5% E 2018E 2019E TowerJazz Vanguard Dongbu HiTek Hua Hong Semi Average for X-FAB peers X-FAB EBIT Margin (%) 25% 20% 15% 10% 5% 0% -5% -10% -15% 15.1% 9.8% E 2018E 2019E -5.4% TowerJazz Vanguard Dongbu HiTek Hua Hong Semi Average for X-FAB peers X-FAB Source: Company data, Credit Suisse estimates, Thomson Reuters (for 2017/2018 estimates for TowerJazz and Dongbu HiTek), Credit Suisse estimates (for 2017/2018 estimates for Vanguard and Hua Hong Semi, which are covered by CS analyst Randy Abrams in Taiwan) Source: Company data, Credit Suisse estimates, Thomson Reuters (for 2017/2018 estimates for TowerJazz and Dongbu HiTek), Credit Suisse estimates (for 2017/2018 estimates for Vanguard and Hua Hong Semi, which are covered by CS analyst Randy Abrams in Taiwan) X-FAB (XFAB.PA) 16

17 DCF analysis suggests a fair value of 8.3 per share DCF as a cross check valuation. In this section, we discuss our DCF analysis as a crosscheck for our X-FAB valuation. However, based on available peer data and short-cycle visibility on customer orders in this industry, especially in the communications segment, we prefer to base our TP on comps analysis. Mid-term sales growth of high single digits in 2018/2019 before moderating. Based on midterm financial objectives from X-FAB, we believe the company should deliver a c9% revenue CAGR over 2018/2019, before moderating to mid-single-digit growth until EBITDA margins of around 26% long term. The company is targeting EBITDA margins of 19-20% in 2017 and over 27% in We assume the company's margin will reach 26% in 2019 and then stay around that level as fab utilisation reaches around 90% from the current level of 75-80%. For our DCF, we use a beta of 1.25, a risk-free rate of 1.5%, an equity risk premium of 7% and a long-term tax rate of 25% (although the current tax rate is low at less than 5% owing to the significant amount of deferred tax assets in the company's multiple geographies). In addition, we assume the company is targeting a capital structure longer term in which equity is at 66% of the enterprise value (note it was 66% ending Dec-16). This gives us a WACC of 7.3%. For terminal or perpetual growth rate, we assume 1.0%. The figure below shows the sensitivity of X-FAB shares to different WACC and perpetual growth rate assumptions. Figure 29: Sensitivity analysis for fair value of X-FAB's share price using DCF sensitivities around WACC and perpetual growth rate assumptions in per share, unless otherwise stated WACC (%) Perpetual growth rate (%) 0-2.0% -1.0% 0.0% 1.0% 1.5% 2.0% 2.5% 5.3% % % % % % % % % Source: Credit Suisse estimates X-FAB (XFAB.PA) 17

18 Figure 30: Discounted cash flow analysis for X-FAB in US$ '000s, unless otherwise stated X-FAB (year ending) Dec 2015 Dec 2016 Dec 2017 Dec 2018 Dec 2019 Dec 2020 Dec 2021 Dec 2022 Dec 2023 Dec 2024 Dec 2025 DCF E 2018E 2019E 2020E 2021E 2022E 2023E 2024E 2025E Revenue 331, , , , , , , , , , ,246 % change yoy 0.2% 54.9% 13.1% 9.0% 9.1% 7.6% 6.1% 4.6% 3.1% 3.1% 3.1% COGS (273,277) (407,831) (453,850) (478,892) (507,153) (542,203) (571,570) (598,110) (616,911) (636,302) (656,303) % of sales 82.5% 79.5% 78.3% 75.8% 73.5% 73.0% 72.5% 72.5% 72.5% 72.5% 72.5% Gross Profit 57, , , , , , , , , , ,943 Gross margin (%) 17.5% 20.5% 21.8% 24.3% 26.5% 27.0% 27.5% 27.5% 27.5% 27.5% 27.5% Research and development expenses (21,326) (26,847) (31,270) (32,845) (35,135) (38,679) (41,485) (43,736) (45,331) (46,976) (48,673) as % of sales 6.4% 5.2% 5.4% 5.2% 5.1% 5.2% 5.3% 5.3% 5.3% 5.4% 5.4% R&D expenses (27,691) (34,066) (38,570) (40,145) (42,435) (45,679) (48,485) (50,736) (52,331) (53,976) (55,673) as % of sales 8.4% 6.6% 6.7% 6.4% 6.2% 6.2% 6.2% 6.2% 6.2% 6.2% 6.2% Grants 6,365 7,219 7,300 7,300 7,300 7,000 7,000 7,000 7,000 7,000 7,000 Selling expenses (6,219) (7,369) (9,715) (10,273) (10,695) (11,327) (11,826) (12,168) (12,338) (12,507) (12,673) as % of sales 1.9% 1.4% 1.7% 1.6% 1.6% 1.5% 1.5% 1.5% 1.5% 1.4% 1.4% G&A expenses (18,160) (22,787) (31,465) (32,874) (33,810) (36,023) (38,236) (40,011) (41,269) (42,566) (43,904) as % of sales 5.5% 4.4% 5.4% 5.2% 4.9% 4.9% 4.9% 4.9% 4.9% 4.9% 4.9% Total opex (45,705) (57,003) (72,450) (75,992) (79,641) (86,029) (91,547) (95,916) (98,938) (102,049) (105,250) opex as % of sales 13.8% 11.1% 12.5% 12.0% 11.5% 11.6% 11.6% 11.6% 11.6% 11.6% 11.6% Rental income from investment properties (net) 1,231 1,897 1,000 1,000 1,000 1,000 1,000 1,000 1,000 1,000 1,000 Other income (net) 2, (200) (100) (100) (100) (100) (100) (100) (100) (100) EBIT 16,271 50,456 54,500 78, , , , , , , ,592 EBIT margin (%) 4.9% 9.8% 9.4% 12.4% 15.1% 15.5% 16.0% 16.0% 16.0% 16.0% 16.0% D&A 47,165 50,158 55,680 66,381 76,590 80,588 83,567 85,385 85,942 86,449 86,904 as % of sales 14.2% 9.8% 9.6% 10.5% 11.1% 10.9% 10.6% 10.4% 10.1% 9.9% 9.6% EBITDA 63, , , , , , , , , , ,496 EBITDA margin (%) 19.2% 19.6% 19.0% 22.9% 26.2% 26.4% 26.6% 26.3% 26.1% 25.8% 25.6% Finance income 14,104 11,119 5,000 5,000 5,000 5,000 5,000 5,000 5,000 5,000 5,000 Finance expenses (19,316) (19,123) (11,500) (10,000) (10,000) (10,000) (10,000) (10,000) (10,000) (10,000) (10,000) Net financial result (5,212) (8,004) (6,500) (5,000) (5,000) (5,000) (5,000) (5,000) (5,000) (5,000) (5,000) Profit/(loss) before taxes 11,059 42,452 48,000 73,216 99, , , , , , ,592 Tax expenses 2,315 3, (2,929) (3,964) (9,937) (16,962) (24,102) (31,431) (32,450) (33,502) Tax rate (%) -20.9% -8.2% -0.5% 4.0% 4.0% 9.0% 14.0% 19.0% 24.0% 24.0% 24.0% Inventories 61,857 88,972 88,450 93,250 98, , , , , , ,944 as % of sales 18.7% 17.3% 15.3% 14.8% 14.3% 18.0% 18.0% 18.0% 18.0% 18.0% 18.0% Trade and other receivables 44,964 77,292 87,000 91,669 96, , , , , , ,787 as % of sales 13.6% 15.1% 15.0% 14.5% 14.0% 15.0% 15.0% 15.0% 15.0% 15.0% 15.0% Trade payables 12,256 49,032 49,300 50,576 51,750 74,274 78,837 82,498 85,091 87,766 90,525 as % of sales 3.7% 9.6% 8.5% 8.0% 7.5% 10.0% 10.0% 10.0% 10.0% 10.0% 10.0% Net working capital 94, , , , , , , , , , ,207 Change in net working capital (19,504) (22,667) (8,918) (8,193) (8,833) (27,655) (10,495) (8,420) (5,964) (6,152) (6,345) Change in other working capital items 5,026 6,308 (1,130) Capex (71,700) (74,400) (102,080) (128,021) (122,131) (98,785) (96,970) (97,348) (100,408) (103,564) (106,819) as % of sales 21.7% 14.5% 17.6% 20.3% 17.7% 13.3% 12.3% 11.8% 11.8% 11.8% 11.8% Maintenance Capex (22,800) (31,600) (38,280) (42,674) (46,230) (50,507) (53,609) (56,099) (57,862) (59,681) (61,557) as % of sales 6.9% 6.2% 6.6% 6.8% 6.7% 6.8% 6.8% 6.8% 6.8% 6.8% 6.8% Expansion capex (48,900) (42,800) (63,800) (85,347) (75,900) (48,278) (43,360) (41,249) (42,546) (43,883) (45,262) as % of sales 14.8% 8.3% 11.0% 13.5% 11.0% 6.5% 5.5% 5.0% 5.0% 5.0% 5.0% Free cash flow (25,639) 5,351 (8,208) 1,210 41,369 54,623 80,297 82,369 79,101 79,491 79,829 PV of Free Cash Flow 314,081 Assumptions Debt capital structure Terminal Value 1,276,157 Beta end 33.7% PV of Terminal Value 675,851 Risk free rate 1.5% Target 34.0% Enterprise Value 989,931 Equity Risk Premium 7.0% Net cash (end of 2017) 194,502 Cost of Equity 10.3% Equity Value (US$ '000) 1,184,434 Avg cost of debt (pre tax) 2.2% Equity Value (US$ mn) 1,184 Avg Cost of debt (post tax) 1.6% WACC 7.32% Share count (mn) LT Tax rate 25.0% Perpetual growth rate 1.00% FX (USD/EUR) 1.09 Equity Value ( per share) 8.3 Source: Company data, Credit Suisse estimates X-FAB (XFAB.PA) 18

19 Credit Suisse HOLT analysis Credit Suisse HOLT is a proprietary valuation tool that aims to convert income statement and balance sheet information into a CFROI return. The goal is to help assess a company's underlying economics. As per our forecasts, returns for X-FAB are expected to decline from 4.9% in 2016 to 4.0% in 2018 due to increasing capex, before starting to improve to 5.4% by 2019, embedding the cyclical and capex intensive nature of the industry. This is driven by EBITDA margin expansion from around 19% in 2016 to 26% by 2019E, with asset efficiency declining from 0.38x to 0.30x over this period. We highlight the HOLT valuation sensitivities based on various assumptions around sales growth and EBITDA margins over time. Note our HOLT analysis suggests a warranted price of US$9.68 (or 8.9) per share, higher than our TP of This is due to lower discount rate of 4.3% being used by HOLT. X-FAB (XFAB.PA) 19

20 Figure 31: Credit Suisse HOLT analysis X FAB SILICON FOUNDRIES EV (XFAB) HOLT - Credit Suisse Analyst Scenario Data Current Price: USD 8.84 Warranted Price: USD 9.68 Valuation date: 12-May-17 EBITDA Margin (parallel % point change to forecasts) Sales Growth (parallel % point change to forecasts) Dec-15A Dec-16A Dec-17E Dec-18E Dec-19E -2.0% -1.0% 0.0% 1.0% 2.0% Sales Growth, % % -28% -19% -8% -1.0% -20% -10% 0.0% -13% -2% 1.0% -5% 6% 18% 32% 2.0% 3% 14% 27% 42% More than 10% downside 1% 13% 26% 9% 22% 37% Within 10% CFROI & Discount Rate (in %) Asset Growth (in %) % 16% Historical CFROI Historic al Transaction CFROI Forecast CFROI Forecast Transaction CFROI Discount Rate Historical Asset Growth Rate Historic al Growth Incl Intang Forecast Growth Forecast Growth Incl Intang Normalis ed Growth Rate 48% 58% More than 10% upside EBITDA Mgn, % Asset Turns, x CFROI, % Disc Rate, % Asset Grth, % Value/Cost, x Economic PE, x Leverage, % Sales Growth (in %) Operating Margin and EBITDA (in %) - see note* Asset Turns (x) Source: Credit Suisse HOLT. CFROI, HOLT, and ValueSearch are trademarks or registered trademarks of Credit Suisse Group AG or its affiliates in the United States and other countries. * Operating margin (yellow) is EBITDA (grey) plus rental expense and R&D expense For display purposes, these charts have been capped as follows:- Sales growth capped at 20% Source: Company data, Credit Suisse estimates, Credit Suisse HOLT X-FAB (XFAB.PA) 20

21 Demand trends for the Semiconductor and Foundry market X-FAB is one of the leading pure-play foundry companies in the area of specialty analog/mixed-signal chips. It provides manufacturing and design support services for semiconductor companies that either design these analog/mixed-signal ICs for use in their own products (companies like Huawei, Samsung, Hitachi and ABB) or for their customers' products (Melexis, Sensata and AMS). These chips can be used across multiple endmarkets ranging from communications to automotive to industrial. X-FAB specialises in design and processes around CMOS (complementary metal-oxide semiconductor), SOI (silicon on insulator), SiC (silicon carbide) and MEMS (micro-electro-mechanical systems). These chips are then used mainly in Automotive, Industrial and Medical applications, along with specific-use cases in the Communications and Consumer segments such as RF (radio frequency) and MEMS. Therefore, in this section of the report, we look at a detailed breakdown of the global semiconductor market by both end-market and product type. Given that X-FAB is a pure-play foundry and sells into the wider semiconductor players, as a next step we then look at the correlation between foundry revenues and semiconductor industry revenues over time. 4% revenue growth expected for the overall semiconductor industry Gartner expects a 4% revenue CAGR for the overall semiconductor industry in According to Gartner data, Communications and Automotive have been the fastest-growing end markets for the semiconductor industry, both recording 7% CAGR over However, with smartphone unit growth slowing, Gartner expects Automotive and Industrial (both benefiting from rising semiconductor content) to lead in terms of growth, with a 7% and 10% CAGR, respectively, over With growth in Communications slowing, Gartner expects an overall semiconductor industry revenue CAGR of 4% over , from a base of $340bn in 2016 (Figure 32). We provide more detail around the key drivers for Automotive, Industrial and Communications (MEMS), which are key markets for X-FAB, later in this section. Figure 32: Gartner expects a semiconductor industry revenue CAGR of 4% over Semis Industry revenues (US$ mn) E 2018E 2019E 2020E CAGR CAGR E Automotive 15,849 22,032 24,201 24,992 27,094 29,985 30,298 32,320 34,289 36,811 39,529 42, % 7.1% % change yoy -20.7% 39.0% 9.8% 3.3% 8.4% 10.7% 1.0% 6.7% 6.1% 7.4% 7.4% 7.4% Industrial 18,260 26,622 27,962 26,546 29,279 32,719 31,936 33,740 37,676 41,720 45,263 49, % 10.0% % change yoy -17.5% 45.8% 5.0% -5.1% 10.3% 11.7% -2.4% 5.7% 11.7% 10.7% 8.5% 9.2% Communications 59,750 75,220 81,444 84,943 93, , , , , , , , % 2.1% % change yoy -9.7% 25.9% 8.3% 4.3% 10.3% 9.6% 4.7% 8.5% 9.2% 1.6% -2.1% 0.1% Consumer 40,389 49,429 46,005 42,552 38,127 38,524 36,708 36,190 38,185 39,820 40,016 42, % 4.2% % change yoy -12.8% 22.4% -6.9% -7.5% -10.4% 1.0% -4.7% -1.4% 5.5% 4.3% 0.5% 6.7% Data Processing 91, , , , , , , , , , , , % 2.2% % change yoy -6.0% 35.8% -0.8% -4.7% 5.3% 8.8% -7.8% -6.0% 4.8% 2.9% -0.5% 1.9% Others (Military / Aerospace / Medical) 2,825 3,791 3,818 3,255 3,375 3,931 3,977 3,989 4,185 4,266 4,252 4, % 1.8% % change yoy -5.4% 34.2% 0.7% -14.7% 3.7% 16.5% 1.2% 0.3% 4.9% 1.9% -0.3% 0.9% Total Semiconductor market 228, , , , , , , , , , , , % 3.7% % change yoy -10.4% 31.8% 1.8% -2.3% 5.2% 8.6% -2.3% 1.5% 7.2% 3.8% 0.8% 3.2% Source: Gartner data and estimates, Credit Suisse research and a 5% CAGR for X-FAB's core markets. Taking a more detailed look at the technology and design offerings from X-FAB, we note that the key products manufactured by X-FAB relate to analog, application-specific IC, discretes and non-optical sensors. These four product categories accounted for $73bn of semiconductor sales in 2016 (or ~20% of total industry revenues). Gartner expects this part of the market to record a 5% CAGR to $87bn by 2020 (Figure 33). X-FAB (XFAB.PA) 21

22 Figure 33: Gartner expects a 5% CAGR for X-FAB's core markets within the overall semiconductor industry ( E) Semis Industry revenues (US$ mn) E 2018E 2019E 2020E CAGR CAGR E Total Semiconductor market 228, , , , , , , , , , , , % 3.7% % change yoy -10.4% 31.8% 1.8% -2.3% 5.2% 8.6% -2.3% 1.5% 7.2% 3.8% 0.8% 3.2% of which Analog (General Purpose) 15,301 21,160 20,312 19,109 19,597 21,015 21,046 21,486 22,284 23,153 23,798 24, % 3.8% % change yoy -13.2% 38.3% -4.0% -5.9% 2.6% 7.2% 0.2% 2.1% 3.7% 3.9% 2.8% 4.9% Application Specific IC 19,066 21,661 21,824 21,320 19,835 21,227 22,338 22,920 24,138 25,370 26,009 27, % 4.8% % change yoy -18.6% 13.6% 0.8% -2.3% -7.0% 7.0% 5.2% 2.6% 5.3% 5.1% 2.5% 6.2% Discretes 14,371 19,608 20,428 18,517 17,726 19,225 17,696 18,809 19,429 19,661 20,111 20, % 2.6% % change yoy -17.1% 36.4% 4.2% -9.4% -4.3% 8.5% -8.0% 6.3% 3.3% 1.2% 2.3% 3.8% Non-Optical Sensors 2,782 4,289 4,901 5,766 6,107 7,872 8,383 9,408 10,391 11,271 12,333 13, % 8.9% % change yoy -7.3% 54.1% 14.3% 17.7% 5.9% 28.9% 6.5% 12.2% 10.5% 8.5% 9.4% 7.3% Core markets served by X-FAB 51,520 66,718 67,465 64,711 63,265 69,339 69,463 72,623 76,242 79,455 82,250 86, % 4.5% % change yoy -16.1% 29.5% 1.1% -4.1% -2.2% 9.6% 0.2% 4.5% 5.0% 4.2% 3.5% 5.4% Optoelectronics (not core for X-FAB) 15,707 21,889 23,510 24,790 25,616 27,430 26,919 27,152 28,736 31,442 34,018 36, % 7.8% % change yoy -5.8% 39.4% 7.4% 5.4% 3.3% 7.1% -1.9% 0.9% 5.8% 9.4% 8.2% 7.7% Markets served by X-FAB 67,228 88,607 90,975 89,501 88,881 96,769 96,382 99, , , , , % 5.4% % change yoy -13.9% 31.8% 2.7% -1.6% -0.7% 8.9% -0.4% 3.5% 5.2% 5.6% 4.8% 6.1% Source: Gartner data and estimates, Credit Suisse research Comparing revenue growth forecasts from different market research firms. In Figure 34, we compare revenue growth forecasts for X-FAB's different end markets within the overall semiconductor market. This confirms that Automotive and Industrial (which together account for 52% of X-FAB's 2016 revenues) are widely expected to be the fastest-growing segments within the semiconductor industry. Figure 34: Comparison of growth forecasts for the semiconductor market (also by end market) from different third-party market research firms Forecasts at Analyst CAGR E TAM CAGR forecasts Presentation Market research used by X-FAB IHS Markit Gartner Automotive (41% of X-FAB 2016 sales) 9.8% CAGR ; IC Insights (McClean Report, 2017) 5.8% 7.0% Industrial (11% of X-FAB 2016 sales) 7.5% CAGR ; IHS Press Release (22 March 2016) 8.2% 9.1% Communications (30% of X-FAB 2016 sales) 6.3% CAGR ; IC Insights (McClean Report, May 2016) 1.4% 3.4% Consumer (12% of X-FAB 2016 sales) NA 3.4% 3.1% Medical (3% of X-FAB 2016 sales) 12.3% CAGR ; IC Insights as quoted by zacks.com (May 2016) NA NA Others (4% of X-FAB 2016 sales) NA NA NA Total Semiconductor Market 5.2% CAGR ; IC Insights (McClean Report, 2017) 3.3% Source: X-FAB Analyst Presentation, Gartner estimates, IHS Markit estimates, Credit Suisse research Foundry outgrowing the broader semiconductor market a trend that is likely to continue The emergence of OEMs (original equipment manufacturers) and fabless business models means that OEMs like Apple, Google and several Chinese handset brands invest mainly in proprietary IC design without investing in manufacturing capabilities. Meanwhile, fabless companies like Qualcomm, MediaTek and Melexis focus on IC design and testing while relying on foundries for manufacturing. As a result, the foundry market has been growing faster than the overall semiconductor market for the past few years. For example, data from IC Insights show that total foundry revenues grew from $32bn in 2011 to $56bn in This would imply a CAGR of 12% well above the 2% CAGR for overall semiconductor market revenues over the same period. As a result, foundry revenues as a percentage of total semiconductor industry revenues have risen from 10% in 2011 to 16% in 2016 (Figure 35). X-FAB (XFAB.PA) 22

23 Figure 35: Foundry revenues have significantly outgrown semiconductor industry revenues, rising from 10% of industry sales in 2011 to 16% in 2016 Revenues (US$ mn) CAGR Global semiconductor industry revenues 306, , , , , , % % change yoy -2.3% 5.2% 8.6% -2.3% 1.5% Pure-Play Foundry revenues 27,000 31,100 36,186 42,519 45,237 50, % % change yoy 15.2% 16.4% 17.5% 6.4% 10.5% as % of semi industry revenues 8.8% 10.4% 11.5% 12.4% 13.5% 14.7% IDM Foundry revenues 4,910 6,520 6,614 5,100 5,200 5, % % change yoy 32.8% 1.4% -22.9% 2.0% 7.6% as % of semi industry revenues 1.6% 2.2% 2.1% 1.5% 1.6% 1.6% Total Foundry revenues 31,910 37,620 42,800 47,619 50,437 55, % % change yoy 17.9% 13.8% 11.3% 5.9% 10.2% as % of semi industry revenues 10.4% 12.5% 13.6% 13.9% 15.1% 16.4% Source: Gartner, IC Insights, X-FAB Pure-Play foundry CAGR has been significantly higher than IDM foundry growth. Within this foundry revenue growth, we can clearly see that pure-play foundry revenue growth at a 13% CAGR has been materially higher than the 3% CAGR seen at IDM foundries. X-FAB is a specialty analog foundry, which is a small part of the total foundry market... There are two types of pure-play foundry. The first is a digital deep sub-micron foundry, which focuses on leading node technology, required in the memory and logic markets and used mainly in Communications, Computing or Data Processing and Consumer end markets. Examples of such companies are TSMC, GlobalFoundries, UMC and SMIC. These digital-focused foundries account for ~88% of the total pure-play foundry market. The second type of foundry is specialty analog, where the key focus is around discrete and analog chips, which are used mainly in Automotive, Industrial and Medical, along with certain specific-use cases in Communications and Consumer. Examples of such foundries are TowerJazz, Vanguard, Hua Hong Semi and X-FAB. These specialty analog foundries account for around 10% of the total pure-play foundries market. but more importantly, specialty analog foundry revenues have been growing in excess of 10%... Although specialty analog is only a small part of the total foundry market, what is interesting is that revenues here have recorded a 12% CAGR over , which is in line with the growth seen in the overall foundry market (also 12% CAGR) but again materially higher than the growth seen in the semiconductor market (only 2% CAGR). and also rising as a percentage of global semis industry revenues. It is also noteworthy that specialty analog foundry revenues (for the top five players) grew from 0.73% of total semiconductor industry revenues in 2011 to 1.16% in Figure 36: The top 5 specialty analog foundries* have also seen a 12% revenue CAGR in the past 5 years Revenues (US$ mn) CAGR Global semiconductor industry revenues 301, , , , , , , % % change yoy 1.8% -2.3% 5.2% 8.6% -2.3% 1.5% Top5 speciality analog foundries 2,123 2,250 2,697 2,544 3,154 3,271 3, % % change yoy 6.0% 19.8% -5.7% 24.0% 3.7% 20.9% as % of semi industry revenues 0.70% 0.73% 0.90% 0.81% 0.92% 0.98% 1.16% Source: Gartner, Company data (Revenues), Credit Suisse research *Top 5 specialty analog foundries include TowerJazz, Vanguard, Hua Hong Semi, Dongbu HiTek and X-FAB. X-FAB (XFAB.PA) 23

24 We believe a 10% or higher CAGR is reasonable for Specialty Analog foundries While we acknowledge the challenges in trying to forecast growth in specialty foundry revenues at an industry level, we look at two trends to get a sense of potential growth over the mid-to-long term for such companies. Figure 37: We expect Specialty Analog foundry revenues (top 5 players) as a % of global semis industry revenues to keep rising Top5 specialty foundry as % of global semi revenues 1.60% 1.40% 1.20% 1.00% 0.80% 0.60% 0.40% 0.20% 0.00% 0.73% 0.90% 0.81% 0.92% 0.98% 1.16% 1.25% 1.34% 1.42% 1.51% E 2018E 2019E 2020E Figure 38: We expect Specialty Analog foundry revenues (top 5 players) as a % of core TAM* within the semis industry to continue rising at the same rate Top5 specialty foundry as % of core semi revenues 8.00% 7.00% 6.00% 5.00% 4.00% 3.00% 2.00% 1.00% 0.00% 3.34% 4.17% 4.02% 4.55% 4.71% 5.45% 5.87% 6.29% 6.71% 7.14% E 2018E 2019E 2020E Source: Gartner, Company data (Revenues), Credit Suisse estimates Source: Gartner, Company data (Revenues), Credit Suisse estimates *Core TAM includes analog, discrete, ASIC and Non-Optical Sensor products in the overall semiconductor market. First, we look at the trend for specialty analog foundry revenues (for the top five players) as a percentage of total semiconductor industry revenues over time. As shown in Figure 37, this rose from 0.73% in 2011 to 1.16% in 2016, close to a 10bp increase per annum. If we assume that specialty analog foundry revenues as a percentage of global semis revenues keep growing at 10bp pa, the share would be around 1.5% by 2020E, which would in turn imply specialty analog foundry revenues of $5.9bn for the top five players in 2020E compared with $4.0bn in 2016 (implying a CAGR of 11%). We show this in Figure 39. Figure 39: Scenario Analysis: If specialty analog foundry revenues as a % of total industry revenues continue to rise at the same rate seen since 2011, we expect a revenue CAGR of 11-12% in E Revenues (US$ mn) E 2018E 2019E 2020E CAGR CAGR E Global semiconductor industry revenues 306, , , , , , , , , , % 3.7% % change yoy 1.8% -2.3% 5.2% 8.6% -2.3% 1.5% 7.2% 3.8% 0.8% 3.2% Semi TAM served by specialty analog foundries 67,465 64,711 63,265 69,339 69,463 72,623 76,242 79,455 82,250 86, % 4.5% % change yoy 1.1% -4.1% -2.2% 9.6% 0.2% 4.5% 5.0% 4.2% 3.5% 5.4% Top5 speciality analog foundries 2,250 2,697 2,544 3,154 3,271 3,956 4,554 5,053 5,422 5, % 10.7% % change yoy 6.0% 19.8% -5.7% 24.0% 3.7% 20.9% 15.1% 11.0% 7.3% 9.5% as % of semi industry revenues 0.73% 0.90% 0.81% 0.92% 0.98% 1.16% 1.25% 1.34% 1.42% 1.51% Top5 speciality analog foundries 2,250 2,697 2,544 3,154 3,271 3,956 4,475 4,999 5,523 6, % 11.8% % change yoy 6.0% 19.8% -5.7% 24.0% 3.7% 20.9% 13.1% 11.7% 10.5% 12.0% as % of specialty analog foundries core TAM 3.34% 4.17% 4.02% 4.55% 4.71% 5.45% 5.87% 6.29% 6.71% 7.14% Source: Gartner, Company data (Revenues), Credit Suisse estimates *Core TAM served by specialty analog foundries includes analog, discrete, ASIC and Non-Optical Sensor products in the overall semiconductor market Second, we look at the trend for specialty analog foundry revenues (for the top five players) as a percentage of semis industry revenues for core products over time. Core products (or TAM) in this case would include analog, ASIC, discrete and non-optical sensor chips. As shown in Figure 38, this increased from 3.34% in 2011 to 5.45% in 2016, a c40bp increase per annum. If we assume that specialty analog foundry revenues as a percentage of semis revenues for core TAM keep growing at 40bp pa, it would be around 7.1% by 2020E, which would then imply specialty analog foundry revenues of $6.2bn for the top five players in 2020E compared with $4.0bn in 2016 (implying a CAGR of 12%). We also show this in Figure 39. X-FAB (XFAB.PA) 24

25 Growth drivers for X-FAB's end markets Automotive Semis should see a 7% CAGR; content upside can drive a 10% CAGR The two important growth drivers for the automotive-related semiconductor market are car production volumes and the average semiconductor content in a car. According to Gartner, the size of the semiconductor market relating to the automotive segment was $22bn in 2010, and recorded a c7% CAGR over (Figure 40). Gartner expects this market to continue to grow at 7% pa over We believe this is achievable, driven by a combination of growth in car production volumes (~2% CAGR) and rising semiconductor content in cars (~5% CAGR). Below we discuss these growth drivers in detail. Figure 40: Gartner expects semiconductor revenues for the Automotive industry to grow at 7% long term Semiconductor revenues in Automotive (US$ mn) E 2018E 2019E 2020E CAGR 16-20E ASIC 1,722 1,750 1,454 1,440 1,518 1,543 1,671 1,780 1,897 2,053 2, % ASSP 5,551 6,269 6,462 7,126 7,970 7,911 8,190 8,360 8,562 8,844 9, % Analog (total) 1,198 1,299 1,419 1,928 2,139 2,416 2,677 2,921 3,196 3,463 3, % Discrete (total) 3,363 3,795 4,361 4,442 4,691 4,380 4,769 4,961 5,066 5,259 5, % General Purpose Logic , % Memory 1,130 1,103 1,237 1,418 1,681 1,769 1,579 1,767 1,692 1,604 1, % Microcomponents 4,753 5,391 5,247 5,267 5,598 5,466 5,986 6,424 6,945 7,596 8, % Opto-electronics 1,123 1,190 1,297 1,633 2,175 2,595 2,940 3,279 4,337 5,181 5, % Non-optical sensors 2,687 2,884 3,012 3,249 3,564 3,492 3,729 3,963 4,230 4,581 4, % Total (by component type) ($ mn) 22,032 24,201 24,992 27,094 29,985 30,298 32,320 34,289 36,811 39,529 42, % % change 39.0% 9.8% 3.3% 8.4% 10.7% 1.0% 6.7% 6.1% 7.4% 7.4% 7.4% Semiconductor revenues in Automotive (US$ mn) E 2018E 2019E 2020E CAGR 16-20E Powertrain 4,136 4,218 4,459 4,432 4,591 4,680 4,788 5,042 5, % Chassis 4,083 4,270 4,633 4,603 5,025 5,378 5,737 6,096 6, % Safety 3,595 3,954 4,505 4,524 4,732 5,012 5,232 5,479 5, % Infotainment 3,937 4,465 5,086 5,281 5,556 5,914 6,270 6,663 7, % Body 3,909 4,242 4,773 4,750 4,964 5,155 5,949 6,535 6, % Aftermarket 2,111 2,132 2,169 2,190 2,112 2,176 2,170 2,230 2, % Instrument Cluster 1,572 1,698 1,853 1,923 2,075 2,183 2,292 2,475 2, % ADAS * 896 1,281 1,539 1,647 2,084 2,461 2,869 3,286 3, % EV / HEV * ,182 1,329 1,502 1,724 1, % Total (by application) ($ mn) 24,992 27,094 29,985 30,299 32,320 34,289 36,811 39,529 42, % % change 8.4% 10.7% 1.0% 6.7% 6.1% 7.4% 7.4% 7.4% Source: Gartner data and estimates, Credit Suisse research Car production units should grow at 2% pa over the next 4 years slower than the 5% CAGR seen in the past five years According to industry data, total car production in 2016 was close to 100mn, rising from 77mn in 2011 (implying a CAGR of 5% over ). We forecast this to rise to 106mn units by 2020E (Figure 43), which implies a CAGR of 2%. China slowdown ahead but still likely to grow. China (~25% of global auto units) has been driving the bulk of growth in global units, and our China auto analyst (Bin Wang) has been flagging a likely deceleration in unit growth as we progress through 2017 as the State Council starts to increase the purchase tax on smaller vehicles from 5.0% to 7.5% in 2017 (similar to what happened a year after the 2008 financial crisis). As such, we believe that car unit growth in China will slow from +18% yoy in 2016 to +7% in 2017 before stabilising at 3% growth in the longer term. Europe and NA unit growth likely to flatten out beyond Europe (~25% of global car units) and NA (~20% of global units) are the other two key regions accounting for a cumulative 45% of car sales. After seeing robust growth in the EU and the US, we are now seeing some signs of saturation and hence expect units to largely flatten out beyond APAC (excl. China) should grow at 2-3% pa. This region accounts for roughly 15% of global car sales, and we expect it to see 2-3% pa growth in car unit sales over time. X-FAB (XFAB.PA) 25

26 Figure 41: YoY growth in auto units in key markets Figure 42: China is ~25% of global auto units (2016) Auto unit growth (yoy) in Key Markets 12% 10% 8% 6% 4% 2% 0% -2% 6% 6% 9% 4% 3% 2% 1% 6% 8% 6% 8% 5% 5% 4% 4% 4% 3% 2% 2% -1% 6% 5% 5% 0% 3% 8% 0% 5% 10% 5% 6% 5% 5% Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 8% 4% 10% 12% RoW, 16% WE, 16% 8% 5% JPN + KOR, 7% APAC (ex CHN), 16% CHN, 25% NA, 19% Source: MarkLines, Credit Suisse research Source: MarkLines, Credit Suisse research Figure 43: Car volumes to grow at close to 2% pa Car production units (mn) E 2018E 2019E 2020E RoW Japan APAC (excl China) China North America Europe Source: Infineon data (based on IHS), Credit Suisse estimates Figure 44: Avg semis content in cars to rise to $430 LT 307 GLOBAL RoW JPN KOR CHN NA EU E Semiconductor content per car (US$) Source: Infineon data (based on Strategy Analytics), Credit Suisse estimates Rising semis content in cars a secular accelerating trend; at least 5% content CAGR ahead While car production volumes remain dependent on the overall macro environment, one of the secular trends in the automotive market in the past few years has been an increase in semiconductor components going into cars. Figure 45: Some examples of the use of semiconductors in cars Source: Infineon Automotive brochure X-FAB (XFAB.PA) 26

27 For example, semiconductors are being used in cars in applications ranging from power controls to voltage regulators, microcontrollers, intelligent sensors and radio-frequency. With a growing focus on improving energy efficiency, increasing safety and improving vehicle connectivity, we see a clear trend towards a continued rise in semiconductor content in cars. Figure 46: Auto semis market by component type Figure 47: Auto semis market by application Non-optical sensors 12% Optoelectronics 9% ASIC 5% ASSP 25% Instrument Cluster 6% Aftermarket 7% ADAS * 6% EV / HEV * 4% Powertrain 14% Chassis 16% Microcompone nts 19% Memory 5% General Purpose Logic 2% Source: Gartner, Credit Suisse research Discrete (total) 15% Analog (total) 8% Body 15% Infotainment 17% Safety 15% * ADAS --- Adanced Driver Assistance Systems; * EV / HEV --- Electric / Hybrid Electric Vehicle Source: Gartner, Credit Suisse research Improving energy efficiency. One of the key themes over the past few years has been a growing focus on ways to reduce CO2 emissions. For cars, one of the targets is to enable technology to bring down emission levels from around grams of CO2/km globally to around 100 grams by 2020 (Figure 48). To achieve this, car manufacturers are looking at two options: Electrification of mechanical parts. Increasing electrical control of existing mechanical units, for example the use of electronic power steering (EPS), replacing hydraulic pumps with electric pumps, and electronically controlled fans all of which should help reduce emission levels. Adoption of EV/HEV cars. A key enabler for lower emissions would be the adoption of electric and hybrid vehicles (EV and HEV). According to industry research, the average semiconductor content for a typical combustion engine control in a mid-sized car is around $340, whereas it increases to around $710 for HEV due to some $370 of semis content add-on. Of this $370 increase, around 75% is due to power-related semiconductor products. This implies total power-related semis content goes up from around $70 in a normal car to around $350 in an HEV car. Similarly, for an EV midsized car, the average semi content is around $705 with power accounting for $390 (or 55% of the total), as shown in Figure 49. X-FAB (XFAB.PA) 27

28 Figure 48: Vehicle emission levels long-term target to reduce it to ~100 grams of CO2 per km Figure 49: Semi content in an HEV / EV car is typically over 2x that of a normal petrol/diesel car Source: The International Council On Clean Transportation (Sep 2015) Source: Infineon presentation (Aug 2016) Increasing safety or ADAS features. With passive safety systems used in automotive already reaching their limit in terms of effectiveness (to avoid accidents and road fatalities), there is rising focus on active safety systems, which are currently used mainly in premium segment cars. Most of these systems revolve around providing Advanced Driver Assistance Systems (ADAS) through more cameras, radar, light radar (Lidar) and actuators. But clearly there is an increasing focus on introducing these safety mechanisms even to lower-segment cars. So features such as autonomous braking systems for pedestrian protection and automatic distance/speed control will likely be included in almost all medium-sized cars in the next few years, not only in developed regions but also across emerging markets. Figure 50: Safety features becoming increasingly important even in lower-segment cars Figure 51: Increasing connectivity features in a car such as e-call, v2v connection and navigation E-Call Internet Navigation Local-based service advertisement Automatic toll collection Maintenance updates Traffic alerts / V2V Power on demand Source: Infineon Automotive brochures Source: Credit Suisse research Better vehicle connectivity. Another visible trend is increasing connectivity within cars, be it internet, navigation, automated tolls or traffic alerts (Figure 51). This will also result in the need for higher levels of security against fraud in areas of critical safety functions (e.g. brakes and steering systems) and theft (e.g. immobilisers and central locking). These connectivity and security features will be enabled by microcontrollers used in safety and powertrain, and also discrete security controllers in the areas of infotainment and car-tocar communication. X-FAB (XFAB.PA) 28

29 Figure 52: Inclusion of more ADAS features can drive significant growth in semis content per car Source: Infineon presentation (Aug 2016) 7% CAGR for the global automotive semiconductor market in E Based on a combination of these two drivers car production volumes and average semis content per car we estimate that global semiconductor revenues relating to the automotive market will grow from $35bn in 2016 to $45bn by 2020E, implying a CAGR of 7% (Figure 53). 2% unit CAGR for cars globally. According to industry data, total car production in 2016 was close to 100mn units, rising from 77mn in 2011 (implying a CAGR of 5% over ). We forecast this to rise to 105mn units by 2020 (Figure 43), which implies a CAGR of 2%. This is a slowdown compared with a 5% unit CAGR seen over as we expect growth in China to slow to a more sustainable 4% CAGR versus the double-digit growth seen since Global average for semi content value per car to rise from ~$350 in 2015 to ~$430 by Driven by the trends mentioned above, the average semiconductor content per car rose from roughly $300 in 2011 to close to $350 in 2015, a trend that seems to be secular. Based on our analysis of car production data and semiconductor content per car by region, we see semis content per car on a global basis rising towards $430 by 2020E (Figure 44), implying a 7% CAGR over E. X-FAB (XFAB.PA) 29

30 Figure 53: We forecast Automotive-related semiconductor industry revenues to record a 7% CAGR (2% unit growth + 5% content growth = 7% CAGR) Car production, semiconductor content per car and automotive-related semiconductor revenues by region over time E 2018E 2019E 2020E CAGR 16-20E Europe Car production (mn) % % change -4.5% 1.0% 3.1% 9.2% 6.3% 4.0% 1.5% 1.0% 1.0% Semiconductor value per car ($) % % change 1.3% 5.0% 3.6% 3.5% 3.6% 3.6% 3.7% 4.0% 3.8% Semiconductor revenue ($ mn) 7,701 8,171 8,723 9,855 10,855 11,696 12,317 12,937 13, % % change -3.2% 6.1% 6.8% 13.0% 10.2% 7.7% 5.3% 5.0% 4.9% North America Car production (mn) % % change 17.6% 5.2% 4.9% 5.2% 0.2% -2.0% 0.0% 0.5% 0.5% Semiconductor value per car ($) % % change -0.5% 4.6% 3.0% 2.9% 3.6% 3.6% 4.3% 3.9% 3.8% Semiconductor revenue ($ mn) 6,314 6,950 7,514 8,140 8,454 8,583 8,956 9,354 9, % % change 17.0% 10.1% 8.1% 8.3% 3.9% 1.5% 4.3% 4.4% 4.3% China Car production (mn) % % change 5.7% 14.5% 6.1% 8.9% 17.8% 7.0% 3.0% 3.0% 3.0% Semiconductor value per car ($) % % change 6.7% 5.8% 6.4% 6.0% 7.1% 7.1% 8.1% 7.4% 6.9% Semiconductor revenue ($ mn) 3,832 4,643 5,243 6,056 7,638 8,752 9,747 10,780 11, % % change 12.8% 21.2% 12.9% 15.5% 26.1% 14.6% 11.4% 10.6% 10.1% S Korea Car production (mn) % % change -2.4% -0.6% 6.8% 11.3% -0.4% 1.0% 2.0% 2.0% 2.0% Semiconductor value per car ($) % % change 0.3% 3.9% 4.3% 4.2% 5.0% 5.0% 5.0% 7.6% 7.1% Semiconductor revenue ($ mn) % % change -2.1% 3.3% 11.4% 16.0% 4.6% 6.1% 7.1% 9.7% 9.2% Japan Car production (mn) % % change 19.0% -4.3% 2.6% -9.0% -2.1% 5.0% 0.0% 0.0% 0.0% Semiconductor value per car ($) % % change 3.0% 3.2% 3.6% 3.5% 2.2% 2.2% 1.0% 3.4% 3.3% Semiconductor revenue ($ mn) 4,145 4,095 4,352 4,101 4,103 4,405 4,448 4,599 4, % % change 22.6% -1.2% 6.3% -5.8% 0.1% 7.3% 1.0% 3.4% 3.3% Rest of the world Car production (mn) % % change 4.5% -0.5% -2.2% -1.0% 0.9% 3.1% 3.6% 2.2% 2.2% Semiconductor value per car ($) % % change -6.0% 7.1% 11.0% 1.4% 1.3% 24.1% 4.5% 4.5% 4.5% Semiconductor revenue ($ mn) 2,540 2,706 2,939 2,952 3,018 3,860 4,180 4,464 4, % % change -1.8% 6.5% 8.6% 0.5% 2.2% 27.9% 8.3% 6.8% 6.8% E 2018E 2019E 2020E CAGR 16-20E Global Car production (mn) % % change 6.0% 3.9% 3.1% 4.5% 6.3% 3.6% 1.9% 1.7% 1.7% Semiconductor value per car ($) % % change 1.7% 4.1% 5.1% 3.6% 2.9% 5.5% 4.3% 4.6% 4.4% Semiconductor revenue ($ mn) 25,044 27,093 29,360 31,787 34,783 38,053 40,460 43,025 45, % % change 7.7% 8.2% 8.4% 8.3% 9.4% 9.4% 6.3% 6.3% 6.2% Source: Gartner, data from Infineon Automotive presentations, Credit Suisse estimates X-FAB (XFAB.PA) 30

31 Scope for potential upside for content growth Potential source of upside #1 Rising content in smartphones could potentially add another 3ppt of CAGR to content Growth in smartphone semis content over time is an interesting example. Semis content in iphones recorded an 8% CAGR from the iphone 4 to iphone 7 Plus (from ~US$200 to ~$320 over six years), which we think can also happen in autos, where average semis content is ~$350 currently, but above $1,000 in high-end EV/HEV cars. This is even more interesting given that semis account for a large proportion of a smartphone's price (for example, semis content in the iphone 7 Plus 256GB model is ~$320 and it sells at a retail price of $969, which implies semis content is close to 35% of the retail price). However, semis content as a percentage of the retail price for an average car is still quite low (we estimate around %). This increase in smartphone semis content was driven by the need for bigger displays, faster apps processors, higher memory, more sensors, better camera quality, increased battery life and advanced connectivity options. Figure 54: Semis content in iphones has recorded an 8% CAGR in the past 6 years Final Assembly & Test Semis content ($) Non-Electronic parts Cameras Display/Touchscreen Battery Pack Main Electronics 50 0 iphone4 (launched in Jun- 10) iphone7 Plus (launched in Sep-16) Source: TechInsights (Portelligent), Company data, Credit Suisse research What would an 8% semis content CAGR in cars imply in absolute terms? Is it even plausible? Many of these trends we believe are applicable to autos as we move towards advanced safety, semi-autonomous driving, electrification and increased connectivity, which would suggest that content growth in cars could even accelerate beyond our 5% estimate in an optimistic scenario. While we realise that affordability and purchasing power remain an issue, which this comparison does not take into account, it indicates potential upside to content growth. At an 8% CAGR for semis content in cars, it would imply that average content in cars rises from ~US$350 in 2015 to slightly over US$500 by We note that average content in cars sold in Japan is close to US$500 already; so, again, plausible, in our view (although as we said it remains a theoretical exercise). If this were to happen, it could potentially add another 3ppt of CAGR to the overall automotive semiconductor market, implying a 10% CAGR for industry revenues versus the 7% we forecast over X-FAB (XFAB.PA) 31

32 Figure 55: Scenario analysis: If auto content sees an 8% CAGR, the global average would rise from around $350 in 2015 to just over $500 by 2020E; while theoretically we think this is plausible, affordability of premium cars in EMs may prove to be a bottleneck Avg. semi content in car by region ($) If content grows at 8% CAGR (like we saw with iphone), then avg. content in cars at global level will be over $500 by Japan is already at close to $500 in Europe NA China Korea Japan Brazil Global avg. Avg. semi content in car (2015) 514 Global avg. (scenario analysis) Source: Infineon, Credit Suisse estimates (for scenario analysis) Potential source of upside #2 What if (H)EV car adoption were to increase? 12% (H)EV penetration by 2020E could potentially add another 3ppt of growth to the content CAGR Gartner expects (H)EV-related semis revenues to grow from $1.2bn in 2016 to only $2.0bn by EV/HEV-related semis revenues were around $750mn in 2012, rising to $1.2bn by 2016, implying a 12% CAGR. Gartner expects a CAGR of 14% for this segment within Automotive semis. So looking for only a gradual adoption of H(EV) cars. So far the adoption of electric and hybrid cars has been only gradual, given limited choice around cars, higher prices for such models, mileage delivered per charge and availability of charging stations on both highways and within cities, and some of these issues will take time to be resolved. But equally, with increasing focus on reducing emission levels and a cleaner environment, we are likely to see a gradual adoption of electric cars. For example, there were around 2mn (H)EV cars sold in 2012, which rose to 3-4mn in 2016, implying 3-4% of total units sold last year. X-FAB (XFAB.PA) 32

33 Figure 56: Sensitivity of the auto semis market to the adoption of (H)EV scope for potential upside in the long term Total # of cars sold globally in 2020E (mn) Semi revenues from Auto segment in 2020E ($ mn) 45,681 Incremental semi content for (H)EV car Number of EV/HEV cars sold in 2020 (mn) Number of EV/HEV cars sold in 2020 (mn) as % of total cars sold globally as % of total cars sold globally 0.3% 2.6% 5.0% 7.4% 9.7% 12.1% 0.3% 2.6% 5.0% 7.4% 9.7% 12.1% ,365 1,803 2,240-4% -3% -2% -1% 0% 1% ,193 1,755 2,318 2,880-4% -3% -2% -1% 1% 2% ,458 2,145 2,833 3,520-4% -3% -1% 0% 2% 3% ,723 2,535 3,348 4,160-4% -2% -1% 1% 3% 5% ,050 1,988 2,925 3,863 4,800-4% -2% 0% 2% 4% 6% ,190 2,253 3,315 4,378 5,440-4% -2% 1% 3% 5% 8% ,330 2,518 3,705 4,893 6,080-4% -1% 1% 4% 6% 9% ,470 2,783 4,095 5,408 6,720-4% -1% 2% 5% 7% 10% ,610 3,048 4,485 5,923 7,360-4% -1% 2% 5% 9% 12% Source: Credit Suisse estimates (H)EV has the potential to drive upside to semis revenues in the automotive segment. Depending on the adoption level of hybrid and electric cars, we have also looked at the sensitivity of the semiconductor industry revenues relating to the automotive market. For example, assuming by 2020E there are some 5mn (H)EV cars implying 5% of total shipments (up from 2mn in 2012), the semiconductor market from (H)EV cars would be around $2.0bn based on incremental semis content in electric vehicles of around $375 per car. However, if the penetration number rises to 10%+ by 2020, then let's say assuming 12% penetration, it would imply 13mn (H)EV cars sold in 2020E. In this case, related semis revenues would be around $7.4bn (an incremental $5.4bn, assuming $575 of incremental semis content per car, which is higher than $375 of EV/HEV-related incremental content for a 'normal' car as many of these cars such as Tesla and BMW i3 will be aimed at the premium segment, suggesting incremental content may be higher than for 'normal' cars), indicating 12% potential upside to the automotive semiconductor market globally in 2020E (Figure 56). In this scenario, the total industry revenues could reach around $51bn in 2020E, which would then imply a 10% CAGR. Figure 57: Drivers of content growth if Autonomous Driving takes off Source: Infineon presentation (Oct 2016) X-FAB (XFAB.PA) 33

34 Potential source of upside #3 What about Autonomous driving? Still early days but significant potential Another potential source of upside long term will depend on the adoption trends for Autonomous Driving (AD). While it is still early days, AD as a theme has been increasingly talked about by companies across the automotive supply chain, and increased adoption would lead to content growth around areas such as cameras, radar sensing, actuators/sensors, etc. Industrial Semis another area of robust growth Figure 58: Key application areas that we believe will drive growth in the Industrial Semiconductor market Source: Infineon presentation (Aug 2016) Given a wide range of application areas for Industrial semiconductor chips, which ranges from factory automation, smart grids, renewable energy (wind and solar), rail/metro traction, home appliances, data centres, and oil & gas exploration, we acknowledge that trying to forecast growth in the Industrial semiconductor chip market can be challenging. However, we also note that trends such as energy efficiency, industrial automation and the gradual shift to renewable energy should help drive growth in this end market. Gartner data show that the Industrial semiconductor chip market grew from $27bn in 2012 to $34bn in 2016 (a CAGR of 6%). Gartner forecasts a 10% CAGR to $49bn in revenues by 2020 (Figure 59). X-FAB (XFAB.PA) 34

35 Figure 59: Gartner expects Semiconductor revenues from Industrial markets to record a 10% CAGR Semiconductor revenues in Industrial (US$ mn) E 2018E 2019E 2020E CAGR 16-20E ASIC 1,161 1,363 1,310 1,217 1,255 1,031 1,037 1,190 1,319 1,454 1, % ASSP 2,165 2,587 2,792 2,959 3,235 3,310 3,526 3,855 4,236 4,673 5, % Analog (total) 4,474 4,507 4,174 4,931 5,437 5,558 5,683 6,019 6,268 6,411 6, % Discrete (total) 4,345 4,796 4,202 4,388 4,916 4,634 5,184 5,540 5,756 6,141 6, % General Purpose Logic 1,427 1,450 1,389 1,473 1,556 1,628 1,687 1,811 1,957 2,076 2, % Memory 3,326 3,307 2,423 2,983 3,414 2,970 2,410 2,890 3,321 3,226 3, % Microcomponents 4,494 4,361 4,039 4,325 4,809 4,542 4,832 5,401 6,034 6,578 7, % Opto-electronics 4,873 5,230 5,842 6,602 7,612 7,816 8,897 10,400 12,151 13,924 15, % Non-optical sensors % Total (by component type) (US$ mn) 26,622 27,962 26,546 29,279 32,719 31,936 33,740 37,676 41,720 45,263 49, % % change 45.8% 5.0% -5.1% 10.3% 11.7% -2.4% 5.7% 11.7% 10.7% 8.5% 9.2% Semiconductor revenues in Industrial (US$ mn) E 2018E 2019E 2020E CAGR 16-20E Americas 6,297 6,483 6,134 7,129 7,785 7,393 7,568 8,144 8,672 8,991 9, % % change 47.7% 3.0% -5.4% 16.2% 9.2% -5.0% 2.4% 7.6% 6.5% 3.7% 4.2% Asia-Pacific 6,260 6,941 6,491 8,151 9,695 10,090 11,448 13,555 15,905 18,302 21, % % change 38.0% 10.9% -6.5% 25.6% 18.9% 4.1% 13.5% 18.4% 17.3% 15.1% 15.4% EMEA 7,525 7,796 7,518 6,804 7,560 7,315 7,450 8,114 8,729 9,213 9, % % change 46.8% 3.6% -3.6% -9.5% 11.1% -3.2% 1.8% 8.9% 7.6% 5.5% 6.2% Japan 6,540 6,742 6,402 7,196 7,680 7,137 7,275 7,863 8,414 8,758 9, % % change 50.9% 3.1% -5.0% 12.4% 6.7% -7.1% 1.9% 8.1% 7.0% 4.1% 4.5% Total (by region) (US$ mn) 26,622 27,962 26,546 29,279 32,719 31,936 33,740 37,676 41,720 45,263 49, % % change 45.8% 5.0% -5.1% 10.3% 11.7% -2.4% 5.7% 11.7% 10.7% 8.5% 9.2% Source: Gartner data and estimates, Credit Suisse research Factory automation Industrial motors are indispensable to the functioning of much of the equipment around us, be it cranes, conveyor belts, escalators or robots. Not only that, these motors also find their way into a number of home appliances like washing machines, air conditioning units, refrigerators and induction cooking ranges. Industrial automation is a key to changing the production landscape where requirements around product quality, energy efficiency and mobility are becoming increasingly significant. In addition, rising labour costs even in emerging markets are compelling companies to increase their efforts towards more automation. Figure 60: 10% of global electricity can be saved by variable-speed drives in factories Savings potential --- Variable Speed Drive 100% 100% 42% 28% 25% ~10% 80% 60% 40% 100% 20% 42% 66% 90% 30-50% 0% total electricity demand thereof energy needs of industry thereof energy needs of electric motors thereof energy needs of uncontrolled motors thereof savings potential Source: Infineon (based on ABB study) X-FAB (XFAB.PA) 35

36 Infineon has noted in the past that according to a study by ABB, 42% of power consumed worldwide is attributable to industries/factories, of which 66% is used by electric motors (so 28% of total electricity consumed). Within this, 90% of the power is consumed by motors that do not allow the flexibility to increase or decrease the speed of the motor depending on the requirement. This accounts for around 25% of power consumed globally. Within this, 30-50% of motors would benefit from the introduction of variablespeed drives, which could result in savings of 10% of the total electricity consumption worldwide. One of the focus areas has been using some sort of electronic control system to regulate the speed of these industrial motors, which otherwise run at constant speed. Adoption of speed-regulating motor controls (depending on changing requirements) can thus help in reducing energy consumption. This is becoming increasingly important as Infineon expects that approximately 300mn electric motors have been installed around the world in industrial applications, and these machines account for two-thirds of commercially consumed electricity. Energy efficiency in home appliances Similar to factories, home appliance makers are also leaning towards controlled or variable motors that would make these appliances more energy efficient. A simple example here is that applications where a motor could only either be switched off or on in the past are now being replaced by systems where speed can be made variable depending on the load requirement. Figure 61: Home Appliances could be a key driver for Industrial Semis 2,500 Figure 62: Increasing penetration of Variable Speed Drive (VSD) in Home Appliances VSD Penetration development Power Semis in Home Appliances ($ mn) 2,000 1,500 1, ,196 2,080 Units (mn) E CAGR washing machines % fridges/freezers % dish washer % E room air conditioners % Source: Infineon presentation (May 2016), Credit Suisse research Source: Infineon presentation (May 2016), Credit Suisse research Communications semis growth likely to be weak, but sensors should see healthy growth Gartner data show that the Communications semiconductor chip market grew from $85bn in 2012 to $117bn in 2016 (CAGR of 8%), albeit driven largely by robust smartphone growth where industry volumes saw a 20% CAGR. However, Gartner expects this market to record a CAGR of only 2% reaching $127bn in revenues by 2020E (Figure 63). But what is worth noting here is that within the Communications segment, X-FAB's exposure is predominantly in MEMS/sensors and RF (through the recent acquisition of Altis, now called X-FAB France). We believe that both these areas should grow significantly faster than the overall Communications segment, driven by increasing content. X-FAB (XFAB.PA) 36

37 Sensors for communication devices should see strong growth. Gartner expects a revenue CAGR of 6% for non-optical sensors used in Communication applications (Figure 63), driven by evidence of an increasing number of sensors used in mobile devices for various applications, be it security (facial recognition, fingerprint), navigation/gaming (compass, gyroscope, accelerometer) or imaging (time of flight, colour sensing). Figure 63: Gartner expects a semis revenue CAGR of 2% for the Communications market Semi revenues in Communications (US$ mn) E 2018E 2019E 2020E CAGR 16-20E ASIC 7,668 7,099 7,115 6,616 7,821 9,918 10,721 11,326 12,095 12,443 12, % ASSP 33,720 35,960 38,703 42,668 45,173 43,485 44,554 46,340 47,726 49,903 51, % Analog (total) 5,558 5,310 4,866 5,428 5,899 5,854 6,078 6,258 6,453 6,623 6, % Discrete (total) 3,483 3,657 3,277 3,168 3,604 3,329 3,321 3,342 3,391 3,395 3, % General Purpose Logic 3,876 4,026 4,272 4,672 5,198 4,836 4,676 4,812 4,899 4,747 4, % Memory 11,857 14,680 15,355 19,363 21,457 26,109 32,522 39,750 38,641 32,694 30, % Microcomponents 2,693 3,112 2,804 2,903 3,015 2,888 3,007 3,104 3,211 3,310 3, % Opto-electronics 5,728 6,663 7,137 7,092 7,865 7,957 7,859 8,010 8,314 8,608 8, % Non-optical sensors ,414 1,760 2,640 3,152 3,880 4,399 4,653 4,901 4, % Total (by component type) (US$ mn) 75,220 81,444 84,943 93, , , , , , , , % % change 25.9% 8.3% 4.3% 10.3% 9.6% 4.7% 8.5% 9.2% 1.6% -2.1% 0.1% Semi revenues in Communications (US$ mn) E 2018E 2019E 2020E CAGR 16-20E Americas 9,725 9,548 8,524 8,503 8,794 8,449 8,525 8,707 8,452 7,989 7, % % change 23.7% -1.8% -10.7% -0.3% 3.4% -3.9% 0.9% 2.1% -2.9% -5.5% -4.9% Asia-Pacific 49,245 55,963 62,290 74,032 84,316 90,651 99, , , , , % % change 25.7% 13.6% 11.3% 18.9% 13.9% 7.5% 10.1% 10.7% 2.5% -1.5% 0.9% EMEA 12,527 12,578 11,202 8,433 6,958 6,087 6,087 6,130 5,943 5,617 5, % % change 29.9% 0.4% -10.9% -24.7% -17.5% -12.5% 0.0% 0.7% -3.0% -5.5% -4.3% Japan 3,723 3,354 2,927 2,701 2,603 2,341 2,210 2,040 1,760 1,529 1, % % change 21.3% -9.9% -12.7% -7.7% -3.6% -10.1% -5.6% -7.7% -13.7% -13.1% -14.0% Total (by region) (US$ mn) 75,220 81,443 84,943 93, , , , , , , , % % change 25.9% 8.3% 4.3% 10.3% 9.6% 4.7% 8.5% 9.2% 1.6% -2.1% 0.1% Source: Gartner data and estimates, Credit Suisse research RF remains another growth area. Similarly, for RF products, we are seeing similar trends as there is demand for higher RF content per device driven by rising data traffic on mobile networks and a wide variety of spectrums used for 3G/4G networks (and potentially 5G networks when it is rolled out in the next couple of years). Figure 64: Size of RF market for mobile phones Figure 65: Average RF content / handset is rising Source: Qorvo presentation (Mar 2016) Source: Qorvo presentation (Mar 2016) X-FAB (XFAB.PA) 37

38 Consumer Semis to record a 4% CAGR Again, looking at Gartner data, we see that the Consumer semiconductor chip market declined from $42.5bn in 2012 to $36bn in 2016 (CAGR of -4%), and should reach $43bn in revenues by 2020E, increasing at a 4% CAGR (Figure 66). Figure 66: Gartner expects semiconductor revenues from the Consumer market to record a 4% CAGR Semi revenues in Consumer (US$ mn) E 2018E 2019E 2020E CAGR 16-20E ASIC 6,538 6,146 5,520 4,817 4,917 4,883 4,849 5,249 5,417 5,333 6, % ASSP 13,280 12,463 12,241 9,819 9,653 9,711 10,109 10,531 11,120 11,679 12, % Analog (total) 4,690 4,156 3,930 3,046 3,090 2,951 2,942 2,972 3,045 3,071 3, % Discrete (total) 4,171 4,087 3,173 2,852 2,894 2,539 2,740 2,781 2,721 2,648 2, % General Purpose Logic 2,816 2,665 2,726 2,631 2,628 2,419 2,463 2,616 2,793 2,972 3, % Memory 8,360 7,052 5,705 6,553 7,193 6,670 5,757 6,675 7,029 6,233 6, % Microcomponents 2,991 2,640 2,363 2,360 2,398 2,361 2,503 2,772 3,127 3,423 3, % Opto-electronics 6,160 6,328 6,439 5,688 5,373 4,672 4,262 3,889 3,650 3,421 3, % Non-optical sensors ,236 1, % Total (by component type) (US$ mn) 49,429 46,005 42,552 38,127 38,524 36,708 36,190 38,185 39,820 40,016 42, % % change 22.4% -6.9% -7.5% -10.4% 1.0% -4.7% -1.4% 5.5% 4.3% 0.5% 6.7% Semi revenues in Consumer (US$ mn) E 2018E 2019E 2020E CAGR 16-20E Americas 4,406 4,557 4,303 4,031 4,203 4,046 4,014 4,018 4,041 3,898 3, % % change 25.8% 3.4% -5.6% -6.3% 4.3% -3.7% -0.8% 0.1% 0.6% -3.5% -3.3% Asia-Pacific 32,717 31,016 28,774 25,927 26,540 25,580 25,386 27,454 29,094 29,476 32, % % change 20.7% -5.2% -7.2% -9.9% 2.4% -3.6% -0.8% 8.1% 6.0% 1.3% 8.9% EMEA 2,787 2,331 2,019 1,849 1,879 1,737 1,523 1,340 1,181 1, % % change 15.4% -16.4% -13.4% -8.4% 1.6% -7.5% -12.3% -12.0% -11.9% -15.0% -18.0% Japan 9,520 8,101 7,456 6,320 5,902 5,345 5,268 5,372 5,504 5,639 5, % % change 29.2% -14.9% -8.0% -15.2% -6.6% -9.4% -1.4% 2.0% 2.5% 2.4% 6.3% Total (by region) (US$ mn) 49,429 46,005 42,552 38,127 38,524 36,709 36,190 38,185 39,820 40,016 42, % % change 22.4% -6.9% -7.5% -10.4% 1.0% -4.7% -1.4% 5.5% 4.3% 0.5% 6.7% Source: Gartner data and estimates, Credit Suisse research X-FAB (XFAB.PA) 38

39 Foundry market structure Digital vs. Specialty Analog vs. IDM The semiconductor foundry market is significantly smaller than the semiconductor market overall. According to Gartner, the foundry market in 2015 was c.$50bn (or 15%) of the total semiconductor industry (Figure 67). Not surprisingly, the two are highly correlated and grow (decline) as the demand for semiconductors increases (decreases). However, what is also evident is that foundry revenues as a percentage of semis industry revenues have been rising over time (from 9% in 2009 to 15% in 2015), driven by the ongoing shift to fabless and OEM models. Figure 67: Foundry revenues are only about 15% of total semi industry revenues, but they have been growing faster than industry revenues % % % US$ bn % 10% % change yoy 100 0% 50-10% 0-20% Foundry Revenues Semiconductor Market Revenues % change yoy (Foundry) % change yoy (Semiconductor Market) Source: Gartner (May 2016) A look at the different types of foundries The semiconductor foundry industry is made up of two types of players Pure-Play foundries (TSMC, GlobalFoundries, UMC, TowerJazz, X-FAB, etc.), and IDMs (Integrated Device Manufacturers, such as Samsung, Infineon, and STMicroelectronics). Pure-Play Foundries manufacture only ICs (integrated circuits) and do not design them. Instead, they take design orders from their customers (typically fabless companies and in some cases also IDMs) and produce IC chips according to their customers' specifications. Integrated Device Manufacturers (IDMs), in contrast, engage in both the design and manufacture of the ICs, but sometimes may outsource the manufacturing part to pureplay foundries to avoid having to invest in capex internally if they need extra production capacity. In terms of market share, the foundry industry is dominated by pure-play foundries, whose revenue in 2016 was $50bn, compared with $6bn generated by IDM foundries (Figure 68). X-FAB (XFAB.PA) 39

40 Figure 68: Pure-play foundries have a significant share in the Foundry market over IDMs Foundry (US$ mn) TSMC (Taiwan) 13,307 14,299 16,951 19,935 25,138 26,574 29,488 GlobalFoundries (U.S.) 3,510 3,195 4,013 4,122 4,355 5,019 5,545 UMC Group (Taiwan) 3,965 3,760 3,730 3,959 4,331 4,464 4,582 SMIC (China) 1,555 1,320 1,542 1,962 1,970 2,236 2,921 Powerchip (Taiwan) ,175 1,291 1,268 1,275 TowerJazz (Israel) ,250 Vanguard (Taiwan) Hua Hong Semi (China) Dongbu HiTek (S.Korea) X-FAB (Europe) Others (Pure-Play) 1,802 1,543 2,490 2,280 2,405 2,238 Total Pure-Play Foundries 27,000 31,100 36,186 42,519 45,237 50,005 % change yoy 15% 16% 18% 6% 11% Total IDM Foundries 4,910 6,520 6,614 5,100 5,200 5,595 % change yoy 33% 1% -23% 2% 8% Total Pure-Play + IDM Foundries 31,910 37,620 42,800 47,619 50,437 55,600 % change yoy 18% 14% 11% 6% 10% Source: Company data, IC Insights, Credit Suisse research Within pure-play foundries, there are two types of companies: The first is the digital deep sub-micron foundry, which focuses on leading node technology that is required in the memory and logic markets, which are used mainly in Communications, Computing and Data Processing, and Consumer end markets. Examples of such companies are TSMC, GlobalFoundries, UMC and SMIC. Given the size of smartphones, tablets and PCs in the overall semis market, these digital-focused foundries account for ~88% of the total pure-play foundry market. Second is the specialty analog foundry, where the key focus is around discrete and analog chips that are used mainly in Automotive and Industrial, along with certain specific-use cases in Communications and Consumer. Examples of such foundries are TowerJazz, Vanguard, Hua Hong Semi and X-FAB. These specialty analog foundries account for ~8% of the total pure-play foundries market. Figure 69: Pure-play foundry sales by end markets ($50bn in 2016) Figure 70: Pure-play foundry sales by feature size ($50bn in 2016) ( ) = 2010 share Source: X-FAB analyst presentation (IC Insights, McClean Report 2017) Source: X-FAB analyst presentation (IC Insights, McClean Report 2017) X-FAB (XFAB.PA) 40

41 Market share structure in foundries Here we look at the market shares of different foundry players over time. Given the different end markets and customer exposures for a digital foundry (like TSMC, UMC or GlobalFoundries) versus a specialty analog foundry (like TowerJazz, Vanguard or X-FAB), we believe it is more useful to look at two metrics one is the market share of specialty analog foundries in the total foundry market and second is the share shifts (if any) within the specialty analog foundry market. Specialty Analog foundries have maintained their share within the overall growing foundry market. As we see in Figure 71, digital foundries account for ~88% of total revenues in the foundry market, with specialty analog foundries having an ~8% share. These market shares have been largely stable over time, which shows that both these foundry submarkets have been growing at a roughly similar pace (at the same time materially outgrowing the overall semiconductor industry). Figure 71: Digital and Specialty Analog Foundries have largely maintained their market shares in the 85-88% and 7-9% range respectively over time Foundry revenue share (%) TSMC (Taiwan) 53.0% 54.5% 55.1% 59.1% 58.7% 59.0% GlobalFoundries (US) 11.8% 12.9% 11.4% 10.2% 11.1% 11.1% UMC Group (Taiwan) 13.9% 12.0% 10.9% 10.2% 9.9% 9.2% SMIC (China) 4.9% 5.0% 5.4% 4.6% 4.9% 5.8% Powerchip (Taiwan) 1.4% 2.0% 3.2% 3.0% 2.8% 2.5% Digital Foundries (sub-total) 85.0% 86.4% 86.1% 87.2% 87.5% 87.6% TowerJazz (Israel) 2.3% 2.1% 1.4% 1.9% 2.1% 2.5% Vanguard (Taiwan) 1.9% 1.9% 2.0% 1.9% 1.6% 1.6% Hua Hong Semi (China) 2.3% 2.2% 1.6% 1.6% 1.4% 1.4% Dongbu HiTek (S.Korea) 1.9% 1.7% 1.2% 1.3% 1.3% 1.3% X-FAB (Europe) 0.8% 0.8% 0.8% 0.7% 1.0% Specialty Analog Foundries (sub-total) 8.3% 8.7% 7.0% 7.4% 7.2% 7.9% Others (Pure-Play) 6.7% 5.0% 6.9% 5.4% 5.3% 4.5% Total 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% Source: Company data (Revenues), IC Insights, Credit Suisse research X-FAB has maintained its market share within the top five specialty analog players at ~10%. Within the specialty analog market, TowerJazz has been consolidating its leading market position, helped by its Panasonic JV in 2014 and the acquisition of Maxim's fab in Texas in X-FAB is the #5 player in the specialty foundry market with a ~10% share. In 2016, X-FAB's share was boosted by the acquisition of Altis' France fab (and its consolidation in X-FAB's Q416 financials) along with a significant contract, of which X-FAB outsourced the production to a partner foundry. Figure 72: TowerJazz leads the pack in Specialty Analog foundry with X-FAB holding the #5 position in terms of market share Specialty Analog Foundry TowerJazz (Israel) 23.7% 19.9% 26.2% 29.4% 31.6% Vanguard (Taiwan) 21.6% 28.0% 25.0% 22.5% 20.2% Hua Hong Semi (China) 25.1% 23.0% 21.1% 19.9% 18.2% Dongbu HiTek (S.Korea) 20.0% 17.7% 17.2% 18.1% 17.0% X-FAB (Europe) 9.6% 11.4% 10.5% 10.1% 13.0% Top5 players 100.0% 100.0% 100.0% 100.0% 100.0% Source: Company data (Revenues), IC Insights, Credit Suisse research X-FAB (XFAB.PA) 41

42 Key differences between digital and specialty analog foundries We expect digital foundries to continue to dominate, accounting for 85-90% of total foundry industry revenues given their exposure to leading node technologies, which are used for chips relating to CPU, logic and memory, and then in devices such as smartphones, tablets, PCs and servers. However, we believe analog foundries will also benefit from the growth trends we have seen for chips being used in industrial and automotive applications. As such, we see some key differences between the two types of foundries, with the factors below favouring specialty analog foundry players. Figure 73: Illustrative lifecycle for an Automotive product Source: Company data Long product lifecycle and tough qualification process. Specialty analog foundries that have exposure to Automotive and Industrial products benefit from the long product lifecycle in both of these industries. Some of the automotive products can have a shelf life of as high as 10 years with minimal changes over time. Compare this with digital chips used in smartphones or PCs; the product lifecycle in these categories is typically months (for smartphones) and may be a couple of years (for PCs). Figure 74: Capex-to-sales requirements are much lower in a specialty analog foundry than in a digital foundry Capex to sales (%) 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 82% 54% 59% 50% 53% 52% 41% 16% 16% 13% 9% 14% 15% 12% Digital foundries average Analog/Mixed-signal foundries average Digital Foundries include TSMC, UMC, GlobalFoundries, SMIC Analog Foundries include TowerJazz, Vanguard, X-FAB Source: X-FAB, IC Insights, The McClean Report 2017, FactSet X-FAB (XFAB.PA) 42

43 Lower capex intensity requirement. Driven by the need to have a smaller feature size along with increasing demand for faster and higher computing power, digital foundries have to continue to invest in capex in a race to move to lower manufacturing nodes. This has led to a very high capex-to-sales ratio for digital foundries (an average of 55% over the past six years for TSMC, UMC, SMIC and GlobalFoundries). In comparison, the majority of the mixed-signal analog, discrete, sensor chips being made by specialty analog foundries like X-FAB are being manufactured at node sizes of 130nm or above (with limited requirement to move to lower nodes in the future). This has led to a much lower capex intensity for these analog foundries, where the average capex-to-sales ratio has been around 14% for TowerJazz, Vanguard and X-FAB over , as we show in Figure 74. Figure 75: X-FAB's view of its mixed-signal analog foundry business model vs. digital foundry Source: Company data Lower ASP pressures in specialty analog chips. Another key element in the analog market is a lower level of pricing pressure vs. the digital market. Our analysis around pricing data shows that analog chip ASP fell at the rate of 3% pa over , which is well below the c10% pa declines we have seen in the digital chip market. Figure 76: ASP trends for mixed-signal analog chips have been resilient 140, , ,000 80,000 60,000 40,000 20, Chip ASP (US$) Analog semis revenues ($ mn) Chip units (mn) Chip ASP (US$) Source: SIA data X-FAB (XFAB.PA) 43

44 Large portfolio of process technologies at analog foundries. Another difference lies in the level of chip customisation. Analog ICs are usually customised, whereas digital chips are mostly standardised. This allows analog foundry players like X-FAB to benefit from higher customer retention rates and lower pricing pressure than digital foundry players. Figure 77: Broad portfolio of process technologies at X-FAB M/S = Mixed-Signal, NVM = Non Volatile Memory, RF = Radio Frequency, SOI = Silicon On Insulator Source: Company data X-FAB (XFAB.PA) 44

45 X-FAB company background X-FAB is a pure-play analog/mixed-signal and specialty foundry group for semiconductor applications. The company claims to be the largest pure-play foundry within the Auto/Industrial/Medical semiconductor segments and among the top 5 global specialty analog foundries. The group is under the umbrella of X-FAB Silicon Foundries SE (the holding company) in Belgium, and is headquartered in Erfurt, Germany. The company focuses on manufacturing chips for customers in the Automotive, Industrial and Medical space, but also has a presence in the Communications and Consumer segments. X-FAB operates six manufacturing sites across three continents Europe, North America and Asia and has approximately 3,800 employees globally (including full-time and part-time personnel). Figure 78: X-FAB operates 6 foundries, across 4 countries and 3 continents Source: Company data, Credit Suisse research With a current capacity of 94,000 8-inch (or 200mm) equivalent wafer starts per month, X- FAB works with both fabless semiconductor companies and IDMs to manufacture chip products of their design. As a pure-play foundry, X-FAB mainly manufactures and does not design chips. The company does have its own library of IP-protected designs (including Digital and Analog); however, this is used only to support (and help retain) clients. X-FAB has over 250 patents in more than 190 active patent families, ranging from process design kits to manufacturing process libraries. X-FAB (XFAB.PA) 45

46 Figure 79: History of X-FAB History of X-FAB Year Details: 1992 Current majority shareholders of X-FAB Silicon Foundries SE acquire a fab in Erfurt, Germany 1999 Acquisition of a fab from Texas Instruments in Lubbock, Texas, USA 1999 Acquisition of Thesys in Erfurt, Germany; disposal of its non-foundry business 2002 Acquisition of a Zarlink fab in Plymouth, UK 2006 Merger with 1st Silicon in Kuching, Malaysia - state of Sarawak becomes a shareholder 2007 Acquisition of Zfoundry in Dresden, Germany 2009 The manufacturing site in Plymouth, UK, is sold to Plessey 2011 X-FAB expands its MEMS capabilities and becomes shareholder of MEMS Foundry Itzehoe, Germany, a Fraunhofer spin-off (fully owned since 2015) 2014 X-FAB builds a new MEMS cleanroom in Erfurt, Germany, to further strengthen its MEMS capabilities and establishes a new entity named X-FAB MEMS Foundry GmbH 2016 Acquisition of Altis assets, renamed into X-FAB France Source: Company data, Credit Suisse research X-FAB (XFAB.PA) 46

47 Figure 80: Key management positions at X-FAB Management experienced industry veterans X-FAB's management team members have strong semiconductor industry experience and most have a long employment history with X-FAB. Rudi De Winter, the CEO of X-FAB since 2011, previously served as the CEO of Melexis (an automotive semiconductor player), which is also X-FAB's biggest customer (34% of revenues in 2016). He is also one of the key shareholders of X-FAB through his share ownership of XTRION, which now owns 47% of X-FAB foundry group (it was 61% before the IPO). XTRION also has a 54% stake in Melexis. Chris Förster, who joined the company in 2001, has been the CFO of X-FAB since The company's CTO, COO and the majority of its VPs have also been with X-FAB for around 10 years or more. Name Current Role Joining year Work experience Rudi De Winter CEO 2011 CEO of X-FAB since 2011 CEO of Melexis from 1997 to 2011 Major shareholder of X-FAB through Xtrion Engineering degree from Ghent University - from 1978 to 1983 Chris Förster CFO 2001 CFO of X-FAB since 2003 Managing Director of X-FAB Dresden since 2007 Financial Controller and Finance Manager at Sachsenring Automobiltechnik AG, previously Dr. Jens Kosch CTO 1999 CTO of X-FAB since 2001 Dr. Manfred Riemer COO 2006 COO of X-FAB since 2006 Master's degree (Dipl. Kfm.) in business economics from Georg August University. Director of Design Support of X-FAB from 1999 to 2001 Director of Design Engineering of Thesys Mikroelektronik GmbH from 1995 to 1999 Dipl.-Ing. and Dr.-Ing. degrees from Ilmenau Technical University, Ilmenau, Germany - from 1980 to 1987 Waferfab Manager, Germany, previously Operations manager at Thesys Gesellschaft für Mikroelektronik mbh in Erfurt, previously Technology development engineer at ERMIC mbh in Erfurt, previously Microelectronic development engineer at Funkwerk in Erfurt, previously Ph.D. degree in chemistry from the Martin-Luther-University Halle - from 1981 to 1984 Diploma in analytical and organic chemistry from the Humboldt University in Berlin - from 1976 to 1981 Norbert Meyer VP Strategic Projects 2007 VP Strategic Projects More than 25 years of experience in the semiconductor industry Master s degree in electrical engineering from the Technical University of Karlsruhe, Germany MBA from INSEAD, France Thomas Hartung VP Sales & Marketing 1990s VP Sales & Marketing Communication since 1990s Communication Director of Product Marketing for ASICs and ASSPs at Thesys Gesellschaft für Mikroelektronik mbh, previously Process engineer for Ion implantation and metallization, previously Diploma in physics from the Friedrich Schiller University in Jena, Germany Frank Lorenz VP Quality 2007 VP Quality More than 35 years experience in semiconductor industry Quality Manager at ZMDI in Dresden, Germany Volker Herbig VP Business Unit MEMS 2004 VP Business Unit MEMS since 2016 Diploma in physics/semiconductor components from Technical University in Chemnitz, Germany Director Product Marketing at X-FAB group since 2014 Director Technical Marketing at X-FAB from 2004 to 2013 Project Manager at Carl Zeiss SMS from 2000 to 2004 Engineering Manager at Inkjet Technologies from 1996 to 1999 Master s degree in physics from Humboldt University, Berlin, Germany - from 1985 to 1990 Henryk Schoder VP Human Resources 2014 VP Human Resources HR Manager at Infineon s Dresden site, as Recruitment Consultant in Brighton (UK), Singapore and Dubai, previously HR and IT Manager for Masdar PV, previously Source: Company data, Credit Suisse research Masters in Psychology X-FAB (XFAB.PA) 47

48 X-FAB (XFAB.PA) 48 Shareholding structure of X-FAB (post IPO) Looking at the legal structure of X-FAB, we see that the company and its biggest customer, Melexis, are both majority owned by the same entity, XTRION. XTRION owns 53.6% of Melexis and 46.75% of X-FAB (it was 61.4% of X-FAB before the IPO). XTRION is owned by Roland Duchâtelet (co-founder of Melexis), Françoise Chombar (current CEO of Melexis and wife of Rudi De Winter, also former COO of Melexis from 1994 to 2011) and Rudi De Winter (current CEO of X- FAB, also former CEO of Melexis). Sarawak Technology Holdings (Malaysia), which used to own 35.2% stake in X-FAB, has now reduced its stake to 11.4% post the IPO and partial exercise of the over-allotment option. CEO Rudi De Winter also served as Melexis CEO for 14 years (from 1997 to 2011). Roland Duchâtelet is the co-founder of Melexis and currently serves as Chairman of the Melexis Board and Non-Executive Director on the Board of X-FAB. Post the IPO, XTRION has agreed to a lock-up period of three years, while Sarawak and X-FAB management to a lock-up period of one year. Figure 81: Overview of current legal and shareholding structure X-FAB and Melexis (which is X-FAB's largest customer) are both majority owned by XTRION; Post the IPO, X-TRION owns 46.75% while Sarawak Technology's holding has reduced from 35.2% to 11.4% in X-FAB Roland Duchatelet Francoise Chombar Rudi De Winter Sarawak State (Non-Executive Director, X-FAB Board) (Melexis CEO), Belgium (X-FAB CEO), Belgium Financial Secretary Belgium 100% 100% XTRION N.V. Sarawak Technology TDK-Micronas GmbH Belgium Holdings Sdn. Bhd. Germany Others Malaysia 53.60% 46.75% 11.43% 0.00% 41.82% (61.40% pre IPO) (35.20% pre IPO) (2.00% pre IPO) (1.40% pre IPO) Melexis N.V. X-FAB Silicon Foundries SE Belgium (X-FAB biggest customer) Belgium 100% 100% 100% 97.20% 100% X-FAB MEMS X-FAB Japan K.K. X-FAB France X-FAB Semicond. X-FAB Sarawak Foundry GmbH Japan SAS Foundries AG Sdn. Bhd. Germany Germany, (2.8% treasury stock*) Malaysia Branch Office UK 94.90% 100% 100% 100% 100% X-FAB Dresden X-FAB MEMS MicroDesign OOO X-FAB Dresden X-FAB Texas Inc. Grundstucks- Foundry Itzehoe (a limited company) GmbH & Co KG USA Vermietungsgesell. GmbH Russia Germany mbh & Co KG Germany Germany Source: Company data, Credit Suisse research 16 May 2017

49 X-FAB's key customers and product offerings In 2016, X-FAB served 333 customers worldwide (of which 41 were in automotive) and offered over 1,500 products in the production stage and 900 products in the prototyping stage. Of the products offered, in over 90% of cases X-FAB was the only sourcing foundry used by its clients, which shows a high level of customer stickiness and close client relationships. Of its 333 customers, 170 were small customers (generated less than $100k of revenues in 2016), 123 were medium-sized (between $100k and $1mn) and 40 were larger customers (contributing over $1mn of annual revenues). Figure 82: List of key customers for X-FAB (in no particular order) Selected customers of X-FAB Top 5 customers (60% / 61% / 68% of 2014 / 2015 / 2016 sales) Melexis (42% / 42% / 34% of 2014/15/16 sales) Lite-On Goodix (21% of 2016 sales) Power Integrations Sensata Microsemi Integrated Device Technology (IDT) Senodia Knowles Electronics Austria Micro Systems (ams) Microchip ON Semiconductor ic-haus Micronas Source: Company data, Credit Suisse research. *Note we estimate the Goodix outsourcing contract to decline to only 4% of sales for 2017E; (see page 57 for further details). High revenue exposure to Melexis. Melexis is X-FAB's biggest customer by far, accounting for 34% of revenues in 2016, with the top 5 customers totalling 68% of the company's sales (Figure 82). We note that the exposure to Melexis has declined from 42% in 2014 and 2015, driven partly by the significant outsourcing contract in Communications during We expect Melexis' exposure to decline further towards 30% post the revenue contribution from the full year of consolidation of X-FAB France, acquired from Altis. Figure 83: Growth comparison trends between X- FAB's revenues from Melexis and Melexis' own sales have not looked linear in the past Figure 84: but over time, X-FAB revenues from Melexis have grown in line with growth in Melexis' own revenues 25% 500, ,285 yoy revenue growth (%) 20% 15% 10% 5% 400, , , , ,041 93, ,792 0% E 2018E 2019E Melexis X-FAB revenues from Melexis Melexis sales ( '000) X-FAB sales from Melexis (US$ '000) Source: Company data, Credit Suisse research, Thomson Reuters (for consensus growth forecasts at Melexis) Source: Company data, Credit Suisse research X-FAB (XFAB.PA) 49

50 Figure 85 shows X-FAB's selected end markets, along with its key customers and the types of chips those customers manufacture at X-FAB's foundries. Figure 85: X-FAB selected markets, customers and applications, and X-FAB's ambition and objectives Market Automotive Industrial Medical Communications & Consumer Selected X-FAB customers - Melexis (biggest customer, 34% of 2016 revenue) - Microsemi - Sensata - Top 5 customers = 68% of Semikron - AMS - Hitachi - Thermo Fisher Scientific - Knowles Electronics - IDT (Integrated Device Technology) Selected brands served by Selected applications X-FAB customers Ambition and objectives - Mercedes-Benz - Barometric Pressure Sensor, Inertial Sensor - BMW and Compass for navigation - Volkswagen - IR Temperature Sensor for Cabin - Tesla - Occupant Detection - ABB - 3D Vision - Opto Coupler for Galvanic Decoupling - Inductive Coupler for Galvanic Decoupling - Hall position sensors - Hitachi - Thermo Fisher Scientific - Samsung - Huawei - Ultrasonic Transducers and Receivers - Blood-Pressure Sensors - X-Ray Scintillators - Hearing aid devices - Dental X-Ray - Ingestible Smart Pills - Auto-focus actuator for camera - IR Temperature sensor - Microphones - Compass for Navigation - Fingerprint reader - Maintain technology and market leadership position in automotive foundry - Invest into technologies to serve innovative applications, incl. MEMS and SiC - Early positioning of SiC for Power applications - Further exploit potential for sensors, actuators and sensor conditioning through MEMS and CMOS capabilities (incl. integration into smart sensors) - Exploit growing market of lab-on-chip applications with MEMS and CMOS capabilities, incl. disposable chips for blood, DNA and similar analyses - Leverage high reliability experience from automotive applications for sensitive / mission-critical devices (e.g. pacemakers) - Opportunistic development of technologies/ applications for which X-FAB can create sufficient differentiation - Exploit MEMS volume opportunities with discrete and integrated MEMS Source: Company data, Credit Suisse research High exposure to Auto & Industrial, but Communications also rising now. X-FAB has historically focused primarily on the Autos and Industrial segments, with these two markets accounting for nearly 70% of group revenues in In the past 2-3 years, the company has also expanded its focus on the Medical segment (as it sees this as an emerging area for semiconductor content) along with certain areas within Communications (e.g. MEMS/actuators). However, as we can see from Figure 86, Auto/Industrial accounted for 52% of sales in 2016, with C&C (Communications & Consumer) as high as 42% of sales. We believe C&C will continue to increase in the mix again in 2017 driven by two factors (mentioned below), before Auto/Industrial again starts to increase in the mix from Consolidation of X-FAB France (Altis) from Q416 following its acquisition. X-FAB France generated 80-85% of its $27mn of Communication & Consumer revenues (of a total $31.6mn in sales in Q416) from RF (radio frequency) chips. Figure 86: X-FAB's revenue breakdown by end market 100% 80% 60% 40% 6% 3% 4% 4% 4% 3% 0% 24% 25% 18% 38% 38% 2% 3% 24% 14% 14% 3% 3% 4% 11% 11% 11% Others Outsourced - Communications Communications & Consumer Medical 20% 54% 55% 41% 41% 43% Industrial 0% E 2018E Automotive Source: Company data, Credit Suisse estimates X-FAB (XFAB.PA) 50

51 Figure 87: X-FAB process portfolio and features During 2016, X-FAB had a large contract with a Chinese customer (which we assume to be Goodix (or Shenzhen Huiding), a fingerprint sensor company for the smartphone market in China; our assumption is based on the fact that on its website, Goodix identifies X-FAB as one of its wafer suppliers). Total revenues from this contract were $107mn for X-FAB, of which $94mn (or 18% of group revenues) was outsourced by X- FAB. X-FAB expects this particular contract to decline significantly during In Q117, this contract still generated $15.5mn of sales but the company expects this to decline to only around $3-5mn sales for Q217. X-FAB's key products and their applications X-FAB has six foundries (fabs) that produce mainly 8-inch (or 200mm) wafers, with certain products also being manufactured on 6-inch (or 150mm) wafers. Each of its foundries specialises in different process technologies, including CMOS, SOI, MEMS, SiC and RF (Figure 90). Given that it operates in the analog/mixed-signal space, the company only produces wafers with nodes of over 0.13μm or 130nm (Figure 87). M/S = Mixed-Signal, NVM = Non Volatile Memory, RF = Radio Frequency, SOI = Silicon On Insulator Source: Company data X-FAB (XFAB.PA) 51

52 The different node sizes are applied in different technologies, which have various features and applications. For instance, CMOS will use nodes from 1.0μm to 0.13μm, whereas SOI CMOS will use nodes of sizes 1.0μm, 0.6μm, 0.18μm and 0.13μm. Their key features and the applications they serve are presented in Figure 88. Figure 88: Overview of X-FAB's technology processes and applications Source: Company data X-FAB (XFAB.PA) 52

53 Figure 89: X-FAB's IC creation process IC creation process The company s Integrated Circuit (IC) creation process consists of six steps, starting from IC design, where X-FAB co-operates with a customer to develop a suitable design for the desired application, to IC assembly and testing, where the customer cuts the wafer into individual ICs ('dies') and tests whether the ICs function properly (Figure 89). In terms of wafer production only, typical production lead-times range from one to three months. Source: Company data Details of X-FAB's fabs X-FAB has three fabs in Germany. Post the acquisition of Altis' fab, X-FAB now has six foundries in total, with three in Germany, and one each in Malaysia, the US and France, specialising in differing process technologies, wafer sizes and node sizes. For instance, MEMS-related chips are manufactured in the Erfurt and Itzehoe fabs in Germany, with the Dresden and Erfurt fabs specialising in CMOS products on 200mm and 150mm wafers, respectively. The Malaysian fab has been a success story. From 2006 to 2016, X-FAB increased its customer base for the Sarawak (Kuching) fab in Malaysia from 25 to 170, and converted $5mn of prototyping revenues into $190mn of production revenues. Furthermore, the group does not expect to pay any taxes in Malaysia for the next years given its significant amount of tax loss carry forward ($335mn) and capital allowances ($1.48bn) in its Malaysian entity. In the past five years, the company spent c.$50mn on the fab's capacity increase and plans to invest more in machinery to avoid potential bottlenecks. The company intends to increase this fab's capacity further (total increase of 7k 200mm equivalent wafers per month in the Kuching and Dresden fabs combined). Based on its successful integration experience with X-FAB Sarawak, X-FAB plans to implement a similar strategy in the newly acquired X-FAB France. X-FAB (XFAB.PA) 53

54 Lubbock fab in Texas. The Lubbock fab in Texas is one of X-FAB's oldest foundries and the company is planning to convert it to specialise it in SiC (Silicon Carbide) technology, which can be used in Industrial and Auto applications. In the past five years, the company has invested c. $12mn in the conversion from CMOS to SiC and intends to invest another $12mn over the next three years to develop the technology further and convert 6k WSPM of the site's existing capacity from CMOS to SiC. The SiC investment programme started in 2014 and is expected to be finalised by The company expects to continue SiC expansions after Figure 90: X-FAB's foundry details including specialisation (process technology and wafers) and capacity Foundry Location Year acquired Specialisation Wafer size Fab 1 Erfurt (Germany) 1992 CMOS, SOI, MEMS 150mm, partly 200mm (MEMS) Node size (μm) Capacity in WSPM (200mm equiv. wafers/month) 1.0; 0.8; ,000 Fab 2 Lubbock (Texas, USA) 1999 CMOS, SiC 150mm 1.0; 0.8; ,000 Fab 3 Sarawak (Kuching, Malaysia) 2006 CMOS, SOI 200mm 0.35; 0.25; 0.18; 0.13 Capacity upside potential upside for MEMS in special MEMS clean room upside for WBG, non- CMOS 24,000 upside above 30k WSPM Fab 4 Dresden (Germany) 2007 CMOS 200mm 0.6; ,000 up to 10k WSPM Fab 5 Itzehoe (Germany) 2011 (fully owned since 2015) MEMS 200mm NM Fab 6 Corbeil-Essonnes (France) 2016 CMOS, SOI, RF 200mm 0.18; ,000 Fab 7 Plymouth (UK) 2002 (sold to Plessey in 2009) The company does not provides capacity for Itzehoe fab (MEMS) Source: Company data, Credit Suisse research 94,000 upside in 9k sq mts of unused fab space Details of the recently acquired fab in France. X-FAB France, located in Corbeil-Essonnes, is the most recent fab addition (acquired from Altis in 2016 for 9.2mn). It has the largest capacity of X-FAB's six plants, at 35k wafers per month (200mm equivalent), and currently runs at 70% of its capacity utilisation. 80% of its products are in Consumer & Communication (mainly RF) and the remaining 20% in Automotive and Industrial. Management intends to invest up to $60mn by 2019 to shift the focus of the fab more towards Automotive and Industrial as well as converge the fab's capacity utilisation to the group average by Q317. For further details on each of X-FAB's foundries, please see Figure 90 and Figure 91. Figure 91: X-FAB's foundry plans and other details including historical investments Foundry Location Capacity upside potential Company plan Other details Fab 1 Fab 2 Erfurt (Germany) Lubbock (Texas, USA) upside for MEMS in special MEMS clean room upside for WBG, non- CMOS Fab 3 Sarawak (Kuching, Malaysia) upside above 30k WSPM Fab 4 Dresden (Germany) up to 10k WSPM Fab 5 Fab 6 Itzehoe (Germany) Corbeil-Essonnes (France) Source: Company data, Credit Suisse research upside in 9k sq mts of unused fab space 1. Invest up to ~$32mn in to build 200mm pilot line in Erfurt & invest into TSV (through-silicon via) and Micro Transfer Printing (related to MEMS) 1. Lubbock to be specialised in SiC 2. Invest up to ~$12mn into SiC technology in Lubbock 3. Convert up to 6k WSPM of CMOS capacity into SiC capacity; program started in 2014 and expected to be completed by Invest in machinery to remove potential "bottlenecks" 2. Increase capacity of Kuching and Dresden by 7k WSPM by 2019 (investment program to start in 2017) 3. Insourcing of external maintenance contracts 1. Invest in machinery to remove potential bottlenecks 2. Increase capacity of Kuching and Dresden by 7k WSPM by 2019 (investment program to start in 2017) Also consists of Thesys' acquisition in 1999 (foundry business). MEMS capex spent in last 5 years: ~US$ 50mn on MEMS in total (likely in Itzehoe & Erfurt) Capex spent in last 5 years: ~US$ 12mn (into SiC conversion) Increased customers from 25 in 2006 to 170 in 2016, prototyping revenues of US$ 5mn converted into US$ 190m production revenues in Capex spent in last 5 years: ~US$ 50mn (into capacity increase). It is the most profitable fab in the group. Capex spent in last 5 years: ~US$ 35mn (into 200mm capacity). MEMS capex spent in last 5 years: ~US$ 50mn on MEMS in total (likely in Itzehoe & Erfurt) 1. X-fab plans to copy success from Kuching to France Acquired for 9.2m in 2016, financials consolidated to group 2. Invest up to ~$60mn to make capacity of 35k WSPM available to from Q416, currently runs at 70% utilisation. Current products product mix (to finalise by 2019) are 80% is RF (wireless / handset business), with remaining 3. Shift focus from Consumer/Communications to Automotive 20% in automotive / industrial. 4. For 2017 outsourcing revenue from a Chinese OEM, X-fab will conduct in France ~$20mn (with remaining $30mn still outsourced) 5. Utilisation rate to converge to group average by Q317 X-FAB (XFAB.PA) 54

55 Figure 92: Capital expenditure plan until 2019 Capex investment plans of $ mn ahead Overall, X-FAB plans to invest $ mn over the next three years to increase its capacity in 200mm wafers, enhance its technology in MEMS and SiC, and also transform the newly acquired fab in France. Of this total, $150mn will be maintenance-related capex (in line with history), with the remaining $ mn of capex targeted at expansion plans (Figure 92). Investment areas Target investment Capacity impact Commentary 200mm Up to c.$120mn Up to 7,000 WSPM - Increase in capacity at Kuching and Dresden sites - Investment program to start in 2017 and completed by 2019 MEMS Up to c.$32mn n/m - To build a 200mm pilot line in Erfurt and invest into TSV and Micro Transfer Printing - Investment program to start in 2017 and completed by 2018 SiC Up to c.$12mn Up to 6,000 WSPM (convert CMOS capacity into SiC capacity) - Invest into SiC technology in Lubbock - Investment program started in 2014 and expected to be completed by 2019 France Up to c.$60mn Make capacity of 35,000 WSPM available to X-FAB product mix Maintenance Up to c.$150mn none - Utilization rate expected to converge to group average by Q Investment program to be finalized by Approximately in line with historical maintenance requirements Total capex $350 $370mn Investment spread over Source: Company data, Credit Suisse research Figure 93: X-FAB has multiple suppliers Diverse set of suppliers X-FAB has multiple suppliers for the different materials it uses in the chip-manufacturing process. For each commodity, it tries to use at least 2-3 sources. For example, it uses around 25 suppliers in total for wafers, with 2-3 suppliers for each material. A few specialty chemicals are an exception, where there is no second source. In 2016, the total cost of these supplies accounted for ~24% of the group's revenues (Figure 93). Commodity / Product Material group / Number of suppliers Key suppliers as % of 2016 sales Wafer ~25 sources in total; 2-3 sources per material SUMCO, Siltronic, Shin-Etsu, SunEdison (formerly MEMC) 11% Electricity 4 sources (note: each fab has 1 source) Drewag, EDF, SESCO, Lubbock Power Light 5% Process chemicals ~45 sources in total; 2-3 sources available per material; a few special chemicals purchased from monopolists (no second source) Dow, Honeywell, Du Pont, Sumitomo, Mitsui Chemicals 4% Photo masks 7 sources (multi sourcing for various mask types) Taiwan Mask Corp, Toppan, Photronics, Portal Photomask, Hoya Group 1% Process gases ~15 sources in total; 2-3 sources available per material Air Liquide, Showa Denko, Praxair, Linde, Leeden National Oxygen 2% Metal targets 6 sources in total; 2-3 sources available per material JX Nippon Mining & Metals, Materion, KFMI, Tosoh, Sumitomo Corp <1% Total 24% Source: Company data, Credit Suisse research X-FAB (XFAB.PA) 55

56 X-FAB financials focus on revenue growth and margin expansion Continued revenue growth and margin expansion through 2019 In terms of guidance and mid-term objectives, X-FAB is looking to grow revenues from $513mn in 2016 to over $750mn by 2019 with EBITDA margins improving from 20% in 2016 to more than 27% by Figure 94: Guidance from X-FAB management Short-term guidance (2017) Mid-term objectives (2019) Revenue EBITDA Revenue EBITDA 19-20% of revenue >$750mn >27% of revenue Source: Company data Potential for robust growth From a base of $258.5mn in 2012, X-FAB's revenues grew 12% in 2013 and 14% in 2014, reaching $330.5mn. In 2015, reported sales were flat, but this included a significant FX headwind (of close to 5ppt) given a 16% depreciation in the EUR vs. the USD. In 2016, revenues grew from $331mn to $513mn (+55% yoy), helped by the acquisition of Altis ($32mn only for Q416) and the large contract we assume to be with Goodix in China ($94mn of sales for this Goodix contract had to be outsourced to another foundry in Malaysia). So excluding these two items, organic growth would have been 17% in Figure 95: Revenue breakdown by end market (2016) Figure 96: Revenue breakdown by technology node (2016) Outsourced - Communications, 18% Others, 4% Automotive, 41% Other (MEMS and specific apps) 9% 250nm (0.25 μm) and smaller 12% >350nm (>0.35 μm) 31% Communications & Consumer, 24% Medical, 3% Industrial, 11% 350nm (0.35 μm) 48% Source: Company data, Credit Suisse research Source: Company data, Credit Suisse research X-FAB (XFAB.PA) 56

57 Our approach to triangulating revenue objectives for 2019 Approach #1 Modelling revenue by segment + Altis acquisition + changes in outsourcing contract Here we try to model revenues for different end markets for the core X-FAB business and then adjust them up for the Altis acquisition (to capture the impact of four quarters of contribution in 2017 vs. only one quarter in 2016) and also adjust down for the expected decline in revenues from the outsourcing contract. Robust revenue guidance. The company expects to deliver sales of over $750mn by However, given the sharp decline in the Goodix contract from $107mn in 2016 (of which $94mn was outsourced) to $15.5mn in Q117 and management expecting only $3-5mn for Q217, we believe that $750mn of sales by 2019 may prove difficult to achieve. We model $580mn of sales for 2017E, which would imply 13% yoy growth in reported terms. But we need to adjust it for the Altis acquisition and changes in Goodix revenues, which implies 10% organic growth adjusted for 2017E on our estimates. This assumes the following: Adjustment for Altis or X-FAB France acquisition. X-FAB France contributed $32mn of sales in 2016 (note this was the revenue number for Q416 as it was consolidated for only one quarter). Total X-FAB France revenues in 2016 were estimated to be $122mn. We expect this contribution to grow to $133mn for 2017 driven mainly by new product wins around RF in the communications market, also driving higher utilisation. Adjustment for change in revenues from the outsourcing contract. In 2016, X-FAB outsourced a large part of the contract with a Chinese customer (we assume to be Goodix, which is a China fingerprint sensor company for smartphones), from which it generated $107mn of sales, of which $94mn or 18% of group sales was outsourced. While X-FAB expected this contract to significantly decline during 2017, visibility has significantly reduced based on management comments during Q117 earnings call. Given the company delivered $15.5mn from the contract in Q117 and expects this to decrease meaningfully in Q2 (around $3-5mn), we estimate this contract will generate around $20mn of sales in 2017E. We believe a 10%+ revenue CAGR is plausible for Specialty Analog foundries over the next four years. While we acknowledge the challenges in trying to forecast growth in specialty foundry revenues at an industry level, we look at two trends over time to get a sense of potential growth over the mid-to-long term for such companies. We discuss this in more detail in an earlier section of this report. Figure 97: Scenario analysis: If specialty analog foundry revenues as a % of total industry revenues continue to rise at the same pace as we have seen since 2011, we expect a revenue CAGR of 11-12% over E Revenues (US$ mn) E 2018E 2019E 2020E CAGR CAGR E Global semiconductor industry revenues 306, , , , , , , , , , % 3.7% % change yoy 1.8% -2.3% 5.2% 8.6% -2.3% 1.5% 7.2% 3.8% 0.8% 3.2% Semi TAM served by specialty analog foundries 67,465 64,711 63,265 69,339 69,463 72,623 76,242 79,455 82,250 86, % 4.5% % change yoy 1.1% -4.1% -2.2% 9.6% 0.2% 4.5% 5.0% 4.2% 3.5% 5.4% Top5 speciality analog foundries 2,250 2,697 2,544 3,154 3,271 3,956 4,554 5,053 5,422 5, % 10.7% % change yoy 6.0% 19.8% -5.7% 24.0% 3.7% 20.9% 15.1% 11.0% 7.3% 9.5% as % of semi industry revenues 0.73% 0.90% 0.81% 0.92% 0.98% 1.16% 1.25% 1.34% 1.42% 1.51% Top5 speciality analog foundries 2,250 2,697 2,544 3,154 3,271 3,956 4,475 4,999 5,523 6, % 11.8% % change yoy 6.0% 19.8% -5.7% 24.0% 3.7% 20.9% 13.1% 11.7% 10.5% 12.0% as % of specialty analog foundries core TAM 3.34% 4.17% 4.02% 4.55% 4.71% 5.45% 5.87% 6.29% 6.71% 7.14% Source: Gartner, Company data (Revenues), Credit Suisse estimates *Core TAM served by specialty analog foundries includes analog, discrete, ASIC and Non-Optical Sensor products in the overall semiconductor market First, we look at the trend for specialty analog foundry revenues (for the top 5 players) as a percentage of the total semiconductor industry revenues over time. As X-FAB (XFAB.PA) 57

58 shown in Figure 97, this rose from 0.73% in 2011 to 1.16% in 2016, which is close to a 10bp increase per annum. If we were to assume that specialty analog foundry revenues as a percentage of global semi revenues continue to grow at 10bp pa, it would be around 1.5% by 2020E, which would then imply specialty analog foundry revenues of $5.9bn for the top 5 players in 2020E compared with $4.0bn in 2016 (implying a CAGR of 11%). Second, we look at the trend for specialty analog foundry revenues (for the top 5 players) as a percentage of semis industry revenues for core products over time. Core products (or TAM) in this case would include analog, ASIC, discrete and non-optical sensor chips. As we show in Figure 97, this rose from 3.34% in 2011 to 5.45% in 2016, which is an around 40bp increase per annum. If we were to assume that specialty analog foundry revenues as a percentage of semis revenues for core TAM continues to grow at 40bp pa over E, it would be around 7.1% by 2020E, which would then imply specialty analog foundry revenues of $6.2bn for the top 5 players in 2020E compared with $4.0bn in 2016 (implying a CAGR of 12%). Figure 98: We model 10% organic growth for 2017, followed by 9% in 2018/2019 to get to ~$690mn of sales Sales (US$ mn) E 2018E 2019E Automotive of which X-Fab excl. France % change yoy 1% 15% 10% 13% 10% of which X-FAB France (only Q416 included) 2.3 of which X-FAB France (for full year) % change yoy 10% 40% 40% Industrial of which X-Fab excl. France % change yoy 3% 12% 10% 10% 7% of which X-FAB France (only Q416 included) 2.3 of which X-FAB France (for full year) % change yoy 10% 30% 20% Communications of which X-Fab excl. France % change yoy 34% -20% 10% 8% 6% of which X-FAB France (only Q416 included) 27.0 of which X-FAB France (for full year) % change yoy 9% 9% 6% Incremental sales from insourcing of outsourced project into X-FAB France Communications - Outsourced % change yoy -79% -100% Decline in sales from insourcing of outsourced project into X-FAB France Consumer % change yoy -17% 52% 10% 10% 5% Medical % change yoy 39% 22% 25% 55% 30% Others % change yoy -42% 79% 5% 10% 0% Total X-FAB sales (US$ mn) % change yoy 14.0% 0.2% 54.9% 13.1% 9.0% 9.1% % change yoy (excl. M&A) 45.4% -3.8% % change yoy (excl. M&A and Outsourcing) 17.0% 9.7% Source: Company data, Credit Suisse estimates We forecast 10%+ organic sales growth in X-FAB's Automotive segment. We expect the Automotive semiconductor market to record a 7% CAGR, and potential upside of another 7% (driven by higher electric car adoption, stricter CO2 emission norms and silicon X-FAB (XFAB.PA) 58

59 carbide opportunity). We also think analog foundries have grown faster than their end markets over the past few years (a trend that we expect to continue in the medium term). In light of these factors, we believe a CAGR of 10% or slightly higher is a plausible estimate for X-FAB's Automotive segment. In addition, we believe there will be some product shift at the French fab with Automotive rising in the mix. We forecast high single-digit organic growth in X-FAB's Industrial segment. For Industrial, we model X-FAB's revenues to grow at 10%/10%/7% over the next three years versus the Industrial semiconductor market recording a CAGR of 10% over E. In addition, similar to Automotive, we believe there will be some product shift at the France fab with Industrial also rising in the mix at the expense of Communications. We forecast around 10% organic growth in X-FAB's Communications segment for 2017, but slowing to 6-8% beyond that. For Communications, we model X-FAB's revenues growing at 10% organic driven by the company's expectations around increasing contribution from MEMS and utilisation in the French fab rising towards the group average before end We expect 30%+ growth in Medical. Medical is only 3% of group revenues, but this is an area where we are seeing increasing use of semiconductor-based applications, potential for content increases and even one-time-use sensors (like those used for DNA testing or cancer diagnostics). What is also interesting is that Medical is 14% of X-FAB's prototyping revenues already, which again highlights the potential for growth in this segment. Approach #2 Capacity utilisation analysis In addition to end-market trends, we analyse the company's capacity utilisation to gauge how achievable X-FAB's mid-term targeted sales of over $750mn in 2019 seems. Capex plans over should allow X-FAB to reach production capacity to deliver $750mn of sales by X-FAB has indicated plans to spend $ mn over the next three years to increase its capacity in 200mm wafers, enhance its technology in MEMS and SiC, and transform the newly acquired fab in France. Of this total capex, the company also seeks to spend up to $150mn on maintenance-related capex (in line with history), with the remaining $ mn of capex targeted at expansion plans. This is likely to take its production capacity from 59k wafer starts per month (WSPM) currently to 66k by In addition, the recently acquired France fab adds 35k WSPM (for 2016 it contributed to X-FAB's capacity for only one quarter). So total capacity at X-FAB would reach 101k WSPM after all the expansion plans are executed over the next three years. Figure 99: With current capex plans, X-FAB should be able to have enough fab capacity (at 100% utilisation) to deliver close to $750mn of sales in 2019E Capacity analysis F2016 F2019E Total WSPM (200 mm equivalent) capacity (excl. France fab) Utilization rate (%) 82% 100% Total WSPM (200 mm equivalent) capacity in France fab 9 35 Utilization rate (%) 70% 100% Total WSPM (200 mm equivalent) capacity (incl. France fab) Target Utilization rate (%) 79% 100% Total X-FAB revenues split X-FAB (Germany, US, Malaysia fabs) X-FAB (France fab) Outsourcing revenues MEMS products Total X-FAB (US$ mn) Source: Company data, Credit Suisse estimates At 100% capacity utilisation, sales could be around $750mn. X-FAB notes that capacity utilisation ending 2016 was 82% for all fabs combined (excluding the France fab where utilisation was only 70%). Assuming capex plans allow the company to increase capacity X-FAB (XFAB.PA) 59

60 along with the 100% utilisation potential, it would imply capacity to deliver over $750mn of revenues (including MEMS and Outsourcing) by 2019, up from around $513mn (including MEMS and Outsourcing) in We reduce Outsourcing revenues from $94mn in 2016 to $0 in 2019E. Furthermore, we see potential for MEMS-related revenues to increase from around $40mn in 2016 to at least $100mn by 2019E. This would suggest that X-FAB can deliver slightly over $750mn of sales with its planned increase in fab capacity (see Figure 99). At ~90% targeted capacity utilisation, X-FAB could deliver $690mn of sales in Another way to look at it is that annualised wafer starts were 55k per month for 2016 (assuming only one quarter of capacity for the fab in France) given blended capacity utilisation of ~80% for With capacity planned to rise to 101k WSPM by 2019, and also X-FAB targeting ~90% utilisation in 2019, it would imply 91k WSPM in 2019, which is an around 67% increase from the 2016 base. X-FAB guides to sales rising from $419mn (excluding Outsourcing of $94mn) in 2016 to at least $720mn (excluding our estimate of $30mn for Outsourcing revenues) in However, we estimate revenues in 2019 at $690mn, which would imply a 65% increase over the 2016 base versus a 67% increase in wafer production at the targeted utilization (Figure 100). Figure 100: Our capacity utilisation analysis suggests that at a 90% utilisation rate, X-FAB has the potential to deliver sales of around $690mn in 2019 Capacity (WSPM in '000 of 200mm equivalent) E 2018E 2019E Erfurt (Germany) since Lubbock (USA) since Kuching-Sarawak (Malaysia) since Dresden (Germany) since Corbeil-Essonnes (France) since Itzehoe (Germany) since 2011 (fully owned since 2015) - MEMS Total WSPM (200 mm equivalent) capacity at the end of the year ( 000) Utilization rate (%) 47% 55% 66% 68% 79% 84% 87% 90% Total WSPM (200mm equivalent) produced at target utilization rate % change yoy 17% 20% 13% 36% 48% 7% 6% Total X-FAB sales (US$ mn) Total X-FAB sales excl. Outsourcing (US$ mn) % change yoy 12% 14% 0% 27% 34% 13% 9% Source: Company data, Credit Suisse estimates Approach #3 Customer analysis Another way to look at X-FAB's sales guidance is to break down its revenues by key contributor (Melexis, Goodix, the X-FAB France acquisition and then all other customers in aggregate). Looking at the period , we see that X-FAB's sales from Melexis recorded a 17% CAGR, while revenues from all other customers (excluding X-FAB France and Goodix) grew by 5% pa (Figure 101). X-FAB (XFAB.PA) 60

61 Figure 101: X-FAB sales by customer: If we assume that X-FAB's revenues from Melexis grow in line with Melexis' sales (based on consensus) and it does not win any next-generation fingerprint business with Goodix, then revenues from Other customers would need to grow at 13% pa to hit the $690mn in 2019 Revenue build up by customer (US$ mn) E 2018E 2019E CAGR CAGR E Melexis % 11% % change yoy 21% 23% 0% 24% 13% 10% 10% % of X-Fab sales 36% 39% 42% 42% 34% 34% 34% 34% Goodix (Fingerprint - China) NM -100% % change yoy -81% -100% NM % of X-Fab sales 21% 3% 0% 0% of which Outsourcing (to 3rd party fab in Malaysia) X-Fab France (without Goodix) NM 10% % change yoy NM 12% 9% % of X-Fab sales 6% 23% 24% 24% All Other customers % 13% % change yoy 7% 8% 0% 5% 15% 16% 8% % of X-Fab sales 64% 61% 58% 58% 39% 40% 43% 42% Total X-FAB sales (US$ mn) % 10% % change yoy 12% 14% 0% 55% 13% 9% 9% Source: Company data, Credit Suisse estimates If we were to assume that X-FAB's sales from Melexis grows in line with Melexis sales, revenues from other customers would have to grow at 13% pa to reach $690mn in 2019 (well below its over $750mn guidance) (Figure 101). Note that consensus does not expect any of these customers, such as IDT, Sensata, Knowles and Microsemi, to grow at that pace as shown below. Figure 102: Sales growth forecasts at key customers of X-FAB (excluding Melexis and Goodix) Goodix E 2018E 2019E Revenues (Rmb '000) 685, ,694 1,119,497 3,079,303 3,996,000 4,825,000 6,311,000 % change yoy 25% 31% 175% 30% 21% 31% Revenues ($ '000) 113, , , , , , ,965 % change yoy 22% 25% 157% 31% 21% 31% IDT (Mar year end) Revenues ($ '000) 487, , , , , , , ,174 % change yoy -7% -1% 18% 22% 4% 11% 9% 5% Sensata Revenues ($ '000) 1,913,910 1,980,732 2,409,803 2,974,961 3,202,000 3,226,000 3,328,000 3,448,000 % change yoy 5% 3% 22% 23% 8% 1% 3% 4% Knowles Revenues ($ '000) 1,117,992 1,214,803 1,141,300 1,084, , , , ,996 % change yoy 14% 9% -6% -5% -21% 4% 6% 5% Microsemi Revenues ($ '000) 1,012, ,944 1,138,263 1,245,600 1,655,000 1,809,000 1,905,000 1,998,000 % change yoy 21% -4% 17% 9% 33% 9% 5% 5% X-FAB's top customers (excl. Melexis) ($ '000) 4,531,633 4,769,511 5,399,875 6,174,937 6,887,637 7,323,651 7,768,690 8,289,135 % change yoy 9% 5% 13% 14% 12% 6% 6% 7% X-FAB's top customers (excl. Melexis & Goodix) ($ '000) 4,531,633 4,656,258 5,262,271 6,002,537 6,444,540 6,743,680 7,068,400 7,373,170 % change yoy 9% 3% 13% 14% 7% 5% 5% 4% Source: Company data, Credit Suisse estimates (using consensus estimates for 2017/2018/2019 for all companies using Reuters) As such, in order to deliver on its 2019 sales objective, we believe X-FAB will have to rely on the following drivers: Goodix: X-FAB may need to win back some of the fingerprint business from Goodix as the Chinese fingerprint company has started using another foundry for next-generation products (manufactured at lower node). X-FAB France: Given the low utilisation at X-FAB France (70% in Q416), we believe X- FAB needs to target new customer wins or new product ramps (first in the communications market before moving gradually more to Auto/Industrial). Melexis: Another possibility is that Melexis can continue to grow faster than consensus expectations given exposure to Automotive sensors. X-FAB (XFAB.PA) 61

62 Areas of growth SiC, MEMS and healthy prototyping pipeline X-FAB focusing on SiC investments. X-FAB is committed to developing new technologies to address the changing needs of its clients. Since 2014, the foundry group has invested $12mn in 150mm wafers for its SiC (Silicon Carbide) capabilities and established the world's first 6-inch (150mm) SiC foundry offering before that only 4-inch (100mm) wafers were available. The company is also part of the 'Power America' consortium (to access potential partners and customers) and completed the installation of a high-temperature implanter to process SiC wafers in October Management intends to invest another $12m over the next three years to develop the technology further in its Lubbock fab. X-FAB currently does SiC prototyping for more than 10 customers and plans to begin volume production in 2017, with these chips having use case in both Industrial and Automotive. Increasing investments for MEMS technology. In MEMS, the company has over 20 years of experience. Its foundry in Itzehoe (added to the group in 2011 and fully acquired in 2015) is dedicated purely to this technology. X-FAB also expanded its MEMS production to the site in Erfurt, where it created a new cleanroom in The company has spent $50mn on MEMS capacity in the past five years and intends to invest a further $32mn in the next two years to build a 200mm pilot line in Erfurt and in TSV and Micro Transfer Printing (details of its capex plans are shown in Figure 92). X-FAB estimates that MEMSbased sensor and actuator sales for the whole industry will record a 9% CAGR. Track record of converting prototyping revenues into production revenues is a positive indicator. The lag between prototyping revenue ramp-up to production revenue is usually 4-6 years depending on the complexity of application. An example is X-FAB's 0.35μm node technology, for which the company saw prototyping ramping up in It transformed into production revenue later and then experienced strong growth from 2011 to 2016 (Figure 103), accounting for nearly 50% of group revenues in 2016 (largest percentage among all technology nodes). The company observes that a similar trend can be seen for its 0.18μm node technology, for which prototyping revenues started growing rapidly from 2010 and production revenue started to become meaningful only in As such, X-FAB believes that its robust prototyping portfolio is a good indicator of future production. Figure 103: Healthy prototyping can be a good indicator for future production Prototyping revenue is also called Non-recurring engineering (NRE) revenue Source: Company data X-FAB (XFAB.PA) 62

63 Cost structure suggests that GMs can expand by 700bp to 26.5% by 2019 X-FAB has seen gross margin expansion every year since 2012, having delivered a 20.5% GM in Over the same period, capacity utilisation at its fabs on a blended basis went up from 47% to 79% (80% for Q416 including the France fab). With capacity utilisation targeted to rise further to around 90% by 2019, we believe GMs will continue to expand. To understand the potential for GM expansion driven by targeted revenue growth, we look at the detailed split of COS provided by the company over time. Figure 104: Detailed split of COS (for Q416 excluding any impact from Outsourcing) Figure 105: Around two-thirds of X-FAB's COS is fixed and one-third is variable 100% 80% 60% 40% 28% 27% 25% 47% 46% 49% Total fixed costs Fixed COS (Facility costs + D&A) Fixed & Partly Variable COS (Employee-related + Fixed asset costs) 20% 0% 25% 26% 26% Total variable costs Variable COS (excl. Outsourcing) (Material + Other costs) Source: Company data Source: Company data, Credit Suisse research Material expenses (excluding Outsourcing) as a percentage of sales has remained in the 19-20% range since Material expenses (all variable cost items) account for roughly 25% of the total COS at X-FAB. Of the total material costs, we believe that ~55% is associated with wafer purchases, followed by process chemical-related costs (at ~20%). The remaining ~25% is attributed to photo masks, chemical gases, metals and other costs. In 2016, material expenses increased to 31% of sales but that was driven by the major outsourcing contract. Excluding that contract, material costs remained at 20% of sales. This is a ratio we expect to remain stable as we model material costs at ~20% of sales over E. D&A (reported within COGS) as a percentage of sales to rise to 10% by 2019E. X-FAB expects D&A to rise given capex investments in recent years along with plans for further capex over the next three years. We model D&A (reported within COGS, which is the majority) as a percentage of sales to rise from 8% in 2016 to 10% by Employee-related costs are mainly fixed but have some variable component. Total employee costs at X-FAB declined from 31% of sales in 2012 to 30% in 2014 and to 27% in 2016 (excluding the impact from outsourcing) as revenues have grown. If revenue growth continues, we believe this ratio would continue to reduce to 25% by 2019E. Facility costs are all fixed. Facility costs look set to rise in 2017 due to the Altis acquisition and then grow only gradually thereafter. But more importantly as a percentage of sales, this line item should continue to trend down from 14% in 2013 to 10.3% in 2016 (excluding the impact from outsourcing) to 9.6% in 2019E. X-FAB (XFAB.PA) 63