Bernstein Raises China Semiconductor Equipment Market Forecast to $101B, Top Pick: NAURA Technology
Nashnova编辑部
Bernstein has sharply raised its 2028 China wafer-fab equipment forecast to $101 billion, up 31% from prior estimates, while nearly doubling price targets on NAURA, AMEC, and PIOTECH — sustained overperformance in logic and memory capacity expansion is the core driver.
Where does a $101 billion market come from?
Bernstein now projects China's WFE — wafer fabrication equipment — demand at $57 billion, $73 billion, and $101 billion for 2026–2028, a cumulative $58 billion above prior estimates.
This means → China's annual equipment purchases are approaching or exceeding the rest of the world combined.
Two engines firing at once: logic chips (for AI and mobile processors) and memory chips (DRAM and NAND for data storage), both expanding faster than the market expected.
Why is Huawei's "Tau Scaling Law" the key catalyst?
Huawei has proposed the Tau Scaling Law, built on a technique called LogicFolding — in plain terms = stacking chip circuits that would normally sit on a single layer, trading vertical space for performance.
The cost is straightforward: each folded chip consumes at least twice the wafer area. Advanced-logic WFE demand is therefore projected to hit $33 billion in 2028, up 30% year-on-year.
This reflects a deeper signal: Huawei is using this architecture to work around advanced-node restrictions, aiming to build clusters more powerful than NVIDIA's NVL72 — covering both AI inference and training.
Where is the memory money coming from?
Bernstein raised 2027–2028 memory WFE forecasts by $1.7 billion and $8 billion respectively. Chinese memory makers' equipment demand is projected at $14 billion, $26 billion, and $42 billion for 2026–2028, implying year-on-year growth of 88%, 81%, and 62%.
Three players are scaling simultaneously: YMTC, CXMT, and now Fujian Jinhua is also adding capacity.
This means → CXMT's recent IPO and YMTC's upcoming IPO will give both companies more capital-expenditure firepower, potentially accelerating the buildout further.
How can the domestic-equipment share double in three years?
Bernstein expects China's semiconductor equipment self-sufficiency rate to rise from 23% in 2025 to 52% by 2028 — more than doubling.
In plain terms = by 2028, more than half of the equipment Chinese fabs purchase will come from domestic suppliers.
The inflection point is DRAM: new import restrictions on CXMT in early 2025 forced it to switch away from U.S. equipment vendors. This means → DRAM localization rates will jump in 2026, becoming the steepest segment of the self-sufficiency curve.
How to choose among the three equipment stocks?
NAURA Technology (北方华创) is named top pick: price target raised to RMB 1,380 (from 680), implying 32× 2028E EPS. The logic — it benefits most from advanced-logic expansion, with order exposure near 40%.
AMEC (中微公司) target raised to RMB 660 (from 336) at 39×; PIOTECH (拓荆科技) to RMB 1,200 (from 580) at 36×. All three saw 2028 EPS estimates roughly doubled.
Bernstein flags one caveat: NAURA's Q2 2026 results may miss expectations, making a post-earnings pullback a better entry point.
What are the next catalysts to watch?
Huawei's 2026 Kirin mobile processor teardown — this will be the first physical validation of a LogicFolding chip, a direct test of whether the Tau Scaling Law works in practice.
Huawei Atlas 950 super-computing cluster deployment — housing up to 8,192 chips, a landmark for China's domestic AI compute scale.
YMTC's IPO prospectus disclosure — will reveal its true capex plan and capacity roadmap, a critical window for validating memory-side demand forecasts.
Content is for reference only, not financial advice.