AI’s power race heats up: TSMC’s record profits, US battery plants, and fusion funding surge—who wins the next grid?

TSMC is widely expected to extend its run of record profitability, with analysts pointing to a fifth straight quarter of record profit as the AI boom sustains demand for advanced chips. The reporting frames TSMC’s momentum as a direct beneficiary of hyperscaler and AI accelerator buildouts, reinforcing the idea that leading-edge foundry capacity remains a strategic bottleneck. In parallel, a US-based “organic flow battery” startup, CMBlu, is planning a factory in the United States aimed at powering AI data centers, signaling that the AI buildout is now pulling on grid-scale storage as much as on semiconductors. Separately, global fusion energy annual investments have reportedly reached a record of nearly $4.5 billion, indicating that governments and investors are continuing to fund long-horizon energy options that could later de-risk power constraints for high-load computing. Geopolitically, the cluster links three layers of strategic competition: compute, energy, and industrial capacity. TSMC’s profit trajectory underscores how Taiwan’s semiconductor ecosystem remains central to AI supply chains, giving it leverage but also exposing it to geopolitical risk premia and potential chokepoint concerns. The US battery factory plan suggests Washington is trying to localize critical components for the energy transition and for AI infrastructure resilience, reducing dependence on imported storage technologies and strengthening domestic industrial policy. Meanwhile, the fusion investment surge reflects a broader race to secure future baseload or firm power, where early capital allocation can translate into later technology leadership and regulatory influence. Overall, the winners are likely firms and jurisdictions that can scale both compute and power—while losers face higher costs, slower deployment, and greater exposure to supply disruptions. Market and economic implications span semiconductors, grid equipment, and energy technology funding. TSMC’s expected record-profit quarter implies continued strength in leading-edge wafer demand and supports bullish sentiment for AI-exposed semiconductor supply chains, with potential spillovers into equipment and materials used for advanced nodes. The CMBlu factory narrative points to incremental demand for battery manufacturing capacity and related components, which could influence sentiment around energy storage supply chains and data-center power procurement. The near-$4.5 billion fusion investment record is smaller than today’s mainstream power capex, but it can still move venture and early-stage capital flows, affecting valuations and funding availability for fusion startups and enabling technologies. In FX and rates terms, the story is more about risk appetite and sector rotation than immediate macro shocks, but it can raise the “strategic tech” bid and keep volatility elevated around supply-chain and energy-cost assumptions for AI data centers. Next, investors and policymakers should watch whether AI-driven power demand translates into measurable grid constraints, and whether storage deployments keep pace with data-center commissioning schedules. For semiconductors, the key trigger is whether TSMC’s record-profit guidance persists into subsequent quarters as customers shift from ramping to sustaining AI infrastructure buildouts. For energy, the critical indicators are permitting, construction milestones, and offtake agreements for CMBlu’s planned US factory, plus evidence that data centers are contracting for storage to manage peak loads and reliability. For fusion, watch funding rounds, government procurement signals, and any movement toward demonstration milestones that could unlock follow-on capital. Escalation risk is mainly indirect—if power shortages or supply bottlenecks intensify, it could accelerate industrial localization and raise trade or export-control tensions around advanced manufacturing and energy hardware.

Geopolitical Implications

• 01

  Semiconductor chokepoints and energy reliability are converging into a single strategic competition over AI deployment speed.

• 02

  Industrial localization of energy storage in the US may reduce exposure to cross-border supply disruptions and strengthen bargaining power in future grid and data-center contracts.

• 03

  Record fusion investment suggests governments and capital markets are preparing for a future where firm power becomes a strategic asset for high-load computing and national competitiveness.

• 04

  If power constraints worsen, policy responses could intensify export-control, procurement restrictions, and industrial subsidies across both compute and energy hardware.

Key Signals

• —TSMC guidance and customer order cadence for leading-edge nodes sustaining record-profit expectations
• —CMBlu factory permitting timeline, construction milestones, and signed offtake agreements with data-center operators
• —Evidence of grid congestion or rising capacity charges that force data centers to contract for storage and reliability services
• —Fusion funding rounds, government procurement announcements, and progress toward demonstration milestones that unlock follow-on capital