AI’s power, chips, and capital scramble: China’s cheap electricity meets Japan’s chip supply shake-up

AI is accelerating from boardrooms to balance sheets as multiple reports point to a fast-moving scramble for talent, compute, and funding. On May 28, 2026, a Dutch cloud provider Nebius surged in premarket trading after an ex-OpenAI employee’s fund disclosed a sizeable stake, signaling how quickly insider-adjacent capital is rotating into AI infrastructure plays. In parallel, foreign investors bought Japanese stocks for an eighth consecutive week on an AI rally, while Japan’s state-backed fund was reported to be considering a sale of chipmaking materials maker JSR, a potential signal of how strategic supply chains may be reshuffled. Separately, ByteDance is reportedly developing custom CPU chips to support AI rollout, underscoring that the next competitive edge is increasingly about owning the compute stack rather than just training models. Geopolitically, the cluster highlights a widening gap in the ability to scale AI at cost and speed. China’s “cheap energy” advantage is framed as a key enabler for data-center expansion, effectively turning electricity pricing and grid capacity into a strategic weapon in the US–China AI race. Meanwhile, the reported push by ByteDance to design custom chips suggests Beijing is trying to reduce dependency on external semiconductor ecosystems and compress supply-chain bottlenecks. The Japan angle matters because state-linked capital decisions around chip materials can influence downstream manufacturing resilience and bargaining power with global foundries and equipment suppliers. Even the European cooperation story—Mistral launching partnerships with BMW and Airbus—signals that AI competitiveness is also being operationalized through industrial deployment, not only model performance. Market and economic implications are immediate and cross-sector. AI rally flows are pulling capital into Japanese equities and into AI-adjacent infrastructure and semiconductor supply chains, while Nebius’s premarket jump shows how quickly sentiment can reprice cloud and compute exposure when credible stakes emerge. The ByteDance custom-chip report raises the probability of intensified competition in CPU design and accelerators, which can pressure suppliers that rely on standardized components and benefit firms that can provide tooling, memory, and packaging. China’s energy advantage implies structurally lower marginal costs for training and inference, which can translate into more aggressive pricing for AI services and greater developer adoption, potentially weighing on higher-cost competitors in Europe and the US. The optical-tech squeeze referenced by Nikkei and Huawei’s “chip comeback” narrative further suggests that AI demand is reshaping component demand curves—favoring firms that can deliver at scale while compressing margins for legacy optical and less-integrated hardware. What to watch next is whether these signals translate into concrete capex, product launches, and supply-chain transactions. For China, monitor MiniMax’s rapid sales growth ahead of a new model launch and track whether custom CPU milestones are paired with measurable deployment volumes in data centers. For Japan, the key trigger is whether the state-backed fund proceeds with any sale process for JSR and how that affects pricing, counterparty selection, and downstream materials availability for chipmakers. For Europe, watch whether Mistral’s BMW and Airbus cooperation expands into broader enterprise rollouts that can lock in data, tooling, and long-term contracts. Across markets, the near-term stress test will be whether AI-driven equity inflows persist while hardware bottlenecks ease or worsen—an outcome that could swing optical, semiconductor materials, and cloud infrastructure valuations within weeks.

Geopolitical Implications

• 01

  Electricity pricing and data-center scaling capacity are becoming strategic levers in the US–China AI contest.

• 02

  Custom-chip development reduces exposure to external semiconductor ecosystems and can accelerate technological decoupling.

• 03

  State-backed portfolio decisions in Japan can influence downstream manufacturing resilience and leverage in global chip-materials markets.

• 04

  Industrial AI partnerships in Europe (BMW/Airbus) indicate a shift toward securing enterprise deployment pipelines, not only research wins.

Key Signals

• —Any confirmation of ByteDance custom CPU milestones and early deployment volumes in AI workloads.
• —Progress or denial of the reported JSR sale process, including potential bidders and timing.
• —Evidence that China’s data-center buildout is translating into measurable cost advantages for inference and developer adoption.
• —Sustained foreign inflows into Japanese equities beyond the eighth week, and whether they broaden into semiconductor supply-chain names.