Whalesbook
AI Race Hinges on Compute Power
Bernstein’s latest analysis identifies compute power as the key factor in the US-China AI race. The firm believes whoever leads in computing power will likely lead the global AI revolution, with a focus on processing power and its infrastructure. Projections indicate China's AI compute power could surge significantly by 2035, driven by strategic investments in energy and semiconductor manufacturing. This expansion could reshape the global AI landscape.
Key Players in the Compute Race
Bernstein has identified several companies set to benefit from this rising demand for AI infrastructure. Chinese battery maker CATL and solar firm Sungrow are expected to gain from the push for more energy infrastructure to power compute operations. AI chip designers Cambricon and Hygon are direct beneficiaries as China pushes for domestic chip capacity, crucial for AI. These companies cover key areas, from energy supply to specialized hardware.
Valuations and Competition Challenges
However, analysis reveals significant disparities and risks. CATL, with an estimated market capitalization around $70 billion and a P/E of approximately 25x, faces intense global competition from players like LG Energy Solution and SK On, while navigating price pressures. Sungrow, valued near $15 billion with a P/E of roughly 18x, operates in the solar inverter market against formidable rivals such as Huawei Digital Power. The semiconductor segment faces tougher challenges. Cambricon ($5 billion market cap) and Hygon ($3 billion) are strong in China's domestic market but face major hurdles from US export controls on advanced semiconductor technology. These controls restrict their access to cutting-edge manufacturing, putting them at a disadvantage against global leaders like Nvidia and AMD, who still have broader tech access despite relying on foundries like TSMC.
Key Risks for China's AI Ambitions
The intense US-China tech rivalry creates significant geopolitical and supply chain risks that could derail these companies' projected growth. Export controls and potential sanctions are a clear threat to Chinese chip firms like Cambricon and Hygon. This could fragment global markets and create artificial demand for lower-spec domestic components. Unlike global rivals with diverse market access, these Chinese firms struggle to expand internationally or use the most advanced manufacturing. Focusing solely on compute capacity for national goals might divert resources from innovation and mask the long-term viability of projects built on artificial demand. Past tech rivalries show that such geopolitical focus can lead to price swings and sharp valuation drops when market conditions or policies change. The semiconductor sector is cyclical. While AI chips are in demand, broader market conditions and raw material costs for energy firms like CATL and Sungrow are sensitive to global economics and policy changes.
Outlook: Navigating Geopolitical and Tech Uncertainty
Looking ahead, these Bernstein-identified companies operate in a sector marked by rapid tech advances and strong geopolitical influence. Demand for AI compute power is clear, but its future depends on tech strength, international relations, regulations, and supply chain resilience. Analysts remain cautious on several Chinese tech firms due to these geopolitical factors, balancing growth potential against high risks in their ratings. Their long-term success will hinge on innovation, adapting to trade policies, and competing beyond national mandates.