China Semicap: The surging memory capacity expansion We update our China WFE model in this call, (updated model here). From our channelcheck in the past quarter, Chinese memory IDMs have significantly sped up their capacityexpansion plan for the next few years due to robust memory demand and ample availablecapital, we revised up memory capex in 2027/28 to reflect that change. Local equipmentstocks might correct a little bit after a strong rally, but we believe the order growth is stillmuch stronger than expected, creating more upsides for the next few months.ReiterateOutperform for AMEC, Piotech and NAURA. Qingyuan Lin, Ph.D.+852 2123 2654qingyuan.lin@bernsteinsg.com Stacy A. Rasgon, Ph.D.+1 213 559 5917stacy.rasgon@bernsteinsg.com David Dai, CFA+852 2918 5704david.dai@bernsteinsg.com China WFE is entering another phase of secular growth driven by AI & memory supercycle along with increasing localization.We now project China WFE spending to be USD58/67/77bn in 2026-28, significantly revised up from previous 55/59/61bn, mainly dueto stronger memory capex. This upward revision on demand mainly benefits local and non-US vendors, as both CXMT and YMTC are forced to switch away from US supply chain. Wenow expect both CXMT and YMTC to ramp up one new fab each in 2026 and another twonew fabs each in 2027, with more to come in 2028 (although much less visibility on exactscale at this point). Given most Chinese vendors recognize revenue about 1 year after theyreceive the order, most of the memory capex upside will reflect in 2027/28 WFE demand.We believe the accelerated memory capacity expansion is driven by both stronger demandand increased available capital. The memory supercycle is creating a significant shortagein DRAM/NAND supply, making local customers a lot more willing to adopt local memorychips, especially for DRAM where the technology has been behind and adoption is lagging.On the other hand, the improved profitability allow CXMT and YMTC to generate tens of bnUSD cash in 2026, together with the cash that will be raised through IPO they now bothhave ample capital to deploy for the capacity expansion. Note that Chinese memory playerscan ramp up new clear room just within one year compared to global players that need 2-3years, allowing them to accelerate the capex spending much faster than global peers forthe next two years. Zheng Cui+852 2123 2694zheng.cui@bernsteinsg.com Francis Ma+852 2123 2626francis.ma@bernsteinsg.com Alrick Shaw+1 917 344 8454alrick.shaw@bernsteinsg.com Arpad von Nemes+1 917 344 8461arpad.vonnemes@bernsteinsg.com Juho Hwang+852 2123 2632juho.hwang@bernsteinsg.com Carmine Milano, CFA+44 20 7762 1857carmine.milano@bernsteinsg.com AI could bring upside to our advanced logic projection.We didn’t revise up ourprojection on logic capex for the next two years, but as the AI demand continues to growand Nvidia continues to be banned by both Chinese and US government from shipping toChina, we believe the advanced logic capacity expansion could further accelerate as well. Atthe same time, the demand for AI related peripheral chip also will boost the local foundriesto continue to expand mature logic capacity. With limited visibility at this point, we projectlogic capex to be flattish throughout 2026-28, which we believe there is upside risk to ourprojection. The AI chip localization also indicates that China will need its own HBM andeSSD to power the local AI datacenters, which could further boosted China’s demand onDRAM and NAND capacity and also brings upside to our memory projection. Local vendors continue to take share from global peers, especially in DRAM in 2026.Before 2025, the increased in localization ratio was mainly driven by advanced logic andNAND due to higher urgency, but in 2026 Matured logic localization is picking up as theperformance of local tools improves, and in 2027 we expect DRAM to have a step changeas they switch away from US vendors, continue to boost the localization ratio. BERNSTEIN TICKER TABLE INVESTMENT IMPLICATIONS NAURA (Outperform, CNY680.00): As the domestic WFE leader, NAURA has the broadest product portfolio coveringDeposition (PVD, CVD), Dry Etch (ICP), Thermo Processes, and Cleaning, as well as a more diverse client base covering leadinglogic, DRAM, NAND players, benefiting from the WFE domestic substitution in China with acceleration share gain. AMEC (Outperform, CNY500.00): Primarily focused on Dry Etch (CCP, ICP) with rapid expansion in Deposition (ALD, LPCVD,EPI), commonly perceived as the domestic WFE company with the best technology and widest global recognition, continue tobenefit from the WFE domestic substitution in China with acceleration share gain. Piotech (Outperform, CNY580.00): Rising domestic WFE vendor primarily focused on Deposition (PECVD, HDPCVD, SACVD,ALD) with expansion in W2W and C2W hybrid bonding equipment for advanced packaging. Piotech has a strong track record ofproduct innovation, which should allow it to benefit