UK Needs More Affordable Power Plants To Drive AI: Miller

Matt Miller said the UK needs more affordable power generation, including nuclear energy, to support growth in the country's AI sector. The comments are strategic and policy-oriented rather than event-driven, with no immediate financial figures, regulatory action, or company-specific catalyst. The article also notes Miller founded Evantic, a $400 million UK-based venture capital fund last year.

Analysis

The key market implication is not “more nuclear” in the abstract, but that compute growth is increasingly gated by firm power, not just GPUs or capital. In the UK, that shifts advantage toward developers and owners of dispatchable baseload, grid interconnection, and behind-the-meter solutions that can guarantee uptime; pure-play AI infrastructure without contracted power access becomes a lower-quality asset. Second-order, this favors industrials and utilities with permitting expertise more than semis, because the bottleneck moves from chip supply to electrons and transmission rights.

The second-order competitive effect is geographic. If the UK fails to lower all-in power costs over the next 12-24 months, AI investment will continue to leak toward the US, Nordics, and Middle East, where energy economics are structurally better and permitting is often more centralized. That creates a winner-take-most dynamic for regions that can offer power certainty, while UK venture-backed AI companies face higher burn and slower model deployment, pressuring private-market valuations.

The contrarian point is that “build more nuclear” is a 5-10 year solution to a 12-24 month competitiveness problem. The near-term fix is more likely gas peakers, grid upgrades, demand-response, and long-dated power contracts; nuclear helps the strategic narrative but won’t materially change AI economics before the current funding cycle resets. So the market may be overpricing the long-dated policy promise and underpricing the interim winners in grid modernization and flexible generation.

Catalyst risk runs both ways: if the UK announces credible fast-track power procurement or AI-zone incentives, domestic AI multiples could re-rate within weeks; if not, the gap versus US AI infrastructure names likely widens over months. The biggest tail risk is that power scarcity becomes a binding constraint on AI cluster buildouts, causing delayed revenue ramps and higher failure rates in venture portfolios that assumed compute was abundant.

Key Decisions for Investors

• Long grid modernization and power infrastructure exposure over a 6-18 month horizon: favor UK/EU utilities and transmission-linked names with regulated returns (e.g., NG.L, SSE.L) versus unprofitable AI venture proxies; thesis is that the bottleneck shifts to grid capex and connection queues, offering lower drawdown than speculative AI software.
• Pair trade: long nuclear supply-chain/engineering beneficiaries, short UK AI venture-sensitive small caps or software names dependent on cheap compute. Use a 3-6 month horizon; risk/reward is attractive if policy headlines keep the nuclear narrative alive while power economics remain tight.
• Long US and Nordic data-center infrastructure beneficiaries over UK-exposed peers for 12 months: prefer firms with contracted power and land over pure AI application names; the edge is power availability, not model quality.
• Sell volatility on UK AI policy headlines via event-driven options if listed names rally on nuclear announcements; the near-term fundamental transmission is weak, so upside can fade before capex or permitting changes arrive.
• Watch for a tactical long in flexible generation and demand-response providers on any UK AI cluster announcements; these can monetize within quarters, whereas nuclear returns are too delayed to support current-cycle AI growth.