UBS is bullish on 2026 allocation opportunities in Chinese semiconductors; equipment, AI interconnect, power semiconductors, and edge AI are the key themes
AI summary card
UBS is bullish on 2026 allocation opportunities in Chinese semiconductors; equipment, AI interconnect, power semiconductors, and edge AI are the key themes
The report focuses on 2026 positioning for China tech stocks and emphasizes that domestic substitution, AI compute infrastructure, rising memory capex, and automotive intelligence will drive differentiated gains across China's semiconductor value chain.
- On the equipment chain, NAURA and AMEC are highlighted as preferred allocation names, benefiting from the multi-year expansion cycle in advanced logic and memory, as well as accelerating domestic substitution in etch, deposition, cleaning, and process control.
- AI-related demand is extending from cloud compute to interconnect, memory, packaging, testing, and thermal management; the report believes demand for server GPUs/accelerators, HBM, DRAM, and PCIe and memory interconnect chips remains an important growth theme.
- In power semiconductors, names such as CR Micro benefit from ASP recovery in power discrete devices, product-mix improvement, higher SiC penetration, and a tighter supply-demand balance driven by more disciplined capex.
- Edge AI and autonomous driving are driving localization in AD/ADAS SoCs, CMOS image sensors, and automotive-grade semiconductors, with Omnivision and Horizon Robotics viewed as related beneficiaries.
Report interpretation
Overview
This is an industry and company coverage report from UBS on Chinese semiconductor stock allocation, centered on the question of how to position China tech stocks in 2026. The report covers the Chinese semiconductor value chain, from end-market demand, the global semiconductor cycle, AI cloud computing demand, China AI accelerators, wafer fab equipment, interconnect chips, power semiconductors, to autonomous-driving edge AI, and lists the ratings, target prices, valuations, and earnings growth assumptions for major covered companies.
Core views
The report's core view is that Chinese semiconductors are not a single-cycle trade, but are driven by multiple structural themes: first, memory and advanced logic expansion lifts demand for China's wafer fab equipment, while localization is accelerating; second, AI infrastructure drives demand for server GPUs/accelerators, HBM, DRAM, packaging and testing, interconnect chips, and power management; third, power semiconductors are seeing earnings improvement under supply discipline, AI capacity reallocation, price recovery, and higher SiC penetration; fourth, end-to-end autonomous-driving models and edge AI increase demand for vehicle-side compute, ADAS SoCs, and CMOS sensors; fifth, stock selection favors names with visible growth, rising domestic-substitution share, and valuations that remain attractive.
Analysis framework
The report uses a top-down and bottom-up approach: it first uses indicators such as smartphones, PCs, servers, semiconductor revenue, inventories, wafer fab equipment spending, memory supply and demand, AI token usage, and cloud-capex trends to determine the cycle position, then maps that to different segments of the Chinese semiconductor value chain; it then compares stock ratings, target prices, PE, P/BV, ROE, and EPS growth through company pages and the coverage table.
Methodology notes
Use global semiconductor revenue, inventory, and stock-price leading indicators to identify cycle turning points
The report cites global semiconductor revenue forecasts, revenue and earnings growth charts, inventory analysis, and the relationship between the 3-month moving YoY growth rate of semiconductor revenue and the SOX index to judge cycle bottoms, peaks, and stock-price performance.
Use bit demand, capacity, capex, and pricing conditions to judge the memory cycle
The report separately presents UBS's DRAM supply-demand summary, NAND Flash supply-demand summary, HBM demand forecasts, and supplier shares, showing the impact of AI servers and memory capex on upstream equipment and the memory value chain.
Assess domestic substitution opportunities through expansion projects, import data, and equipment vendors' product progress
The report focuses on China's WFE spending, lithography and SPE imports, lithography equipment imports in major provinces and cities, the global equipment competitive landscape, and the product launches and qualification progress of AMEC, NAURA, Piotech, and others.
Break large-model demand down into compute, memory, interconnect, packaging, and cluster-scaling needs
The report attributes the needs of powerful LLMs to more parameters, larger datasets, and longer inference, and further maps them to HBM, DRAM, PCIe retimers, memory interconnects, CoWoS/SOIC packaging, rack-scale up, and cluster scale-out.
Asset mapping & comparison
Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).
- NAURAPreferred China semiconductor equipment name
- Strengths
- Leading revenue scale; covers etch, PVD, cleaning and other equipment; benefits from advanced logic and memory expansion and localization.
- Weaknesses
- Valuation requires high visibility into growth, and equipment ramp pace and customer qualification progress may affect realization.
- Comparison
- The report views it as one of the top-priority WFE stocks and highlights rising share at domestic wafer fabs and expanding applications.
- Risks
- Wafer-fab capex below expectations, slower-than-expected localization progress, and valuation compression.
- AMECPreferred China semiconductor equipment name
- Strengths
- Has proprietary solutions in etch and deposition tools; benefits from localization and accelerating new-equipment revenue contribution.
- Weaknesses
- Commercialization of some new tools and customer qualification still require time.
- Comparison
- Along with NAURA, it is a core beneficiary of the equipment chain, and the report forecasts WFE revenue CAGR above 40% for 2026-2028E.
- Risks
- Equipment qualification below expectations, volatility in advanced-node demand, and intensifying competition.
- USIAI data center and interconnect-related beneficiary
- Strengths
- AIDC-related businesses, including NIC PCBA and optical modules, are becoming new revenue drivers; it also benefits from Apple business content and AI glasses opportunities.
- Weaknesses
- May still be affected by consumer electronics and customer project timing.
- Comparison
- Compared with traditional packaging/testing or EMS businesses, AI server and interconnect-related revenue provides additional growth optionality.
- Risks
- AIDC order conversion below expectations, customer concentration, and price competition.
- CR MicroPreferred power semiconductor name
- Strengths
- Benefits from ASP recovery in power discrete devices, product-mix improvement, and better profitability.
- Weaknesses
- Power semiconductor demand is still tied to industrial, new energy, and consumer electronics cycles.
- Comparison
- The report highlights its margin-expansion opportunity within Chinese power semiconductors.
- Risks
- Slower-than-expected price recovery, weakening downstream demand, and intensifying SiC competition.
- OmnivisionBeneficiary of CMOS image sensors and autonomous driving
- Strengths
- Emerging applications are expected to drive sequential recovery, with L2+ autonomous driving and localization of China's automotive semiconductors providing long-term drivers.
- Weaknesses
- Image sensors are still affected by fluctuations in smartphone, automotive, and consumer electronics demand.
- Comparison
- Compared with a pure consumer-electronics chain, autonomous driving and in-vehicle sensors provide it with structural demand.
- Risks
- Slower-than-expected penetration of automotive intelligence, price competition, and fluctuating customer demand.
- Horizon RoboticsBeneficiary of edge AI and autonomous-driving SoCs
- Strengths
- The report focuses on localization of China's AD/ADAS SoCs, higher vehicle-side compute, and the adoption of end-to-end models by major automakers.
- Weaknesses
- Profitability realization and market-share gains still depend on vehicle ramp-up and customer adoption.
- Comparison
- It benefits together with CMOS sensors and automotive-grade power semiconductors from automotive intelligence.
- Risks
- Slower autonomous-driving rollout, chip competition, and automaker in-house substitution.
Key data
- Forecast growth in China's WFE spending2026E/2027E/2028E expected YoY growth of 7%/18%/14%, respectivelyUsed to support the medium- to long-term expansion and localization opportunities in the semiconductor equipment chain.
- Memory capacity expansionTotal memory capacity expansion in 2026 of about 120k wpm or higherThe report says most of the newly added capacity will be used for DRAM and expects the 2026-2030 memory capex upcycle to be longer and stronger.
- NAURA WFE revenue growth2026-2028E WFE revenue CAGR above 40%The report believes NAURA benefits from higher technology maturity, expanding product coverage, and greater share at domestic wafer fabs.
- AMEC WFE revenue growth2026-2028E WFE revenue CAGR above 40%The report believes AMEC benefits from localization, R&D investment, new equipment revenue contribution, and qualification of high-aspect-ratio etch tools.
- NAURA coverage table dataRating Buy, current price Rmb721.04, target price Rmb800.00The coverage table shows its 2026E/2027E/2028E P/E at 77.1x/45.1x/31.3x, respectively.
- AMEC coverage table dataRating Buy, current price Rmb360.00, target price Rmb403.00The coverage table shows its 2026E/2027E/2028E P/E at 91.2x/53.8x/36.9x, respectively.
- CR Micro coverage table dataRating Buy, current price Rmb73.59, target price Rmb83.40The report believes it benefits from ASP recovery in power discrete devices, product-mix improvement, and stronger profitability.
- USI coverage table dataRating Buy, current price Rmb37.51, target price Rmb51.50The report notes that its AIDC-related businesses, including NIC PCBA and optical modules, are becoming new revenue drivers.
Impact & implications
In investment terms, the report suggests building China semiconductor exposure around structural incremental opportunities with higher conviction: upstream equipment leaders benefit from expansion and domestic substitution; AI interconnect, packaging/testing, and power management benefit from greater compute-cluster complexity; power semiconductors benefit from price recovery and supply discipline; autonomous driving and edge AI drive localization in vehicle SoCs and sensors. By contrast, names exposed to consumer electronics demand, those whose valuations already fully reflect expectations, and segments more constrained by overseas restrictions warrant greater caution.
Risks
- The global semiconductor cycle or end-market demand may be weaker than expected, especially if demand recovery in smartphones, PCs, and consumer electronics falls short.
- China's wafer-fab expansion pace, memory capex, or WFE localization speed may be slower than expected.
- AI infrastructure capex may slow, leading to weaker-than-expected demand for server GPUs/accelerators, HBM, interconnect chips, packaging/testing, and power management.
- Some names have high valuations, and if earnings growth or order visibility declines, valuation compression risk is significant.
- Overseas export controls, equipment supply-chain constraints, or geopolitical changes may affect advanced process, equipment, and AI chip value chains.
- If power semiconductor price recovery, SiC penetration, and new-energy demand fall short of expectations, earnings improvement for related names may be delayed.
What to watch
- China's WFE spending and major wafer fabs' capex guidance for 2026-2028.
- Changes in DRAM, NAND, and HBM supply-demand balance and memory pricing trends.
- New product qualification, orders, and revenue realization at equipment companies such as NAURA and AMEC.
- Cloud providers' capex, AI token usage, server GPU/accelerator shipments, and AI cluster architecture upgrades.
- Order trends for PCIe retimers, memory interconnect chips, 2.5D packaging, testing equipment, and thermal management demand.
- ASP trends for power discrete devices, lower SiC costs, and penetration rates in EV and energy-storage applications.
- Localization rates for China's AD/ADAS SoCs, end-to-end model adoption by automakers, and trends in vehicle-side compute upgrades.