AI compute is colliding with a power-constrained grid, turning electricity generation, nuclear capacity, gas turbines, and transmission equipment into the binding constraint on the AI buildout.
Data centers are emerging as one of the fastest-growing sources of electricity demand of the decade, with the IEA projecting global data-center consumption to roughly double from ~485 TWh in 2025 to ~945 TWh by 2030 — about 3% of world electricity. Demand from AI-optimized data centers alone is projected to more than quadruple over that window. The result is a structural scramble for firm, dispatchable power: hyperscalers are signing multi-decade power-purchase agreements, the U.S. interconnection queue has swelled past 2,600 GW, and grid operators are committing tens of billions to transmission, repricing every layer of the power stack.
Between 2026 and 2030, the power stack — generation, nuclear, gas turbines, and transmission — is positioned to be the throughput limiter for AI rather than chips. Independent power producers with firm nuclear and gas fleets are converting decades-old assets into long-dated, investment-grade revenue via hyperscaler PPAs, while equipment makers face turbine and grid-component slots booked out toward the end of the decade. Small modular reactors and nuclear uprates move from announcement to early licensing and construction, setting up the next supply wave beyond 2030.
The forces routing money into this theme right now.
Talen Energy's expanded 17-year PPA with Amazon (June 2025) supplies up to 1,920 MW from the Susquehanna nuclear plant, establishing a grid-connected template that Vistra and Constellation are replicating across their fleets.
GE Vernova's gas turbine backlog reached 100 GW in Q1 2026 with slots tight through 2030, and data-center orders for its electrification equipment in Q1 2026 alone exceeded all of 2025 combined.
Constellation's Three Mile Island Unit 1 (Crane Clean Energy Center) restart under a 20-year, 835 MW Microsoft PPA is now targeted for 2027, roughly a year ahead of schedule, backed by a DOE loan guarantee of up to $1 billion.
MISO, SPP, and PJM unveiled 2025 transmission plans totaling roughly $42 billion, with U.S. grid investment forecast near $1 trillion over the coming decade as the interconnection queue exceeds 2,600 GW.
Structural large-cap anchors — lower-variance exposure to the theme.
GE Vernova (GEV) stands as a foundational play on AI's escalating electricity demands, providing critical equipment across power generation and transmission. With a substantial backlog and expanding capacity, GEV is positioned to capitalize on the structural shift towards increased electrification.
Why the excitement: GE Vernova's backlog has grown to $163 billion, with an 80% increase in equipment backlog, signaling strong demand for its solutions.
The honest risk: GE Vernova's current ratio of 0.9 suggests potential short-term liquidity challenges in fulfilling its growing backlog.
Vistra Corp. stands to directly benefit from the escalating electricity demand of data centers, particularly through long-term power purchase agreements. With a diverse generation portfolio including nuclear and natural gas, Vistra is positioned as a structural anchor within the 'Electricity of AI' theme.
Why the excitement: Management sees a structurally improved demand environment with load growth remaining elevated, especially in ERCOT and PJM markets, creating meaningful opportunities for Vistra.
The honest risk: Vistra's high debt-to-equity ratio of 3.56 could constrain its ability to invest in new generation capacity to meet escalating demand.
Constellation Energy is positioned as the structural anchor of the AI electricity theme, leveraging its nuclear fleet to meet the surging power demands of data centers. The company is increasingly contracting directly with hyperscalers for firm, carbon-free energy.
Why the excitement: Management affirmed a compelling long-term outlook with a base earnings growth rate exceeding 20% through 2029, driven by nuclear production tax credits and long-term contracts.
The honest risk: Policy shifts in PJM, where Constellation has significant operations, could impact the economics of new capacity and customer contracting.
Smaller names with higher upside and deeper potential drawdowns.
Oklo is a speculative, pre-revenue bet on next-generation nuclear power, aiming to capitalize on the surging electricity demand from AI data centers. The company is pursuing an integrated platform across power, fuel, and isotopes.
Why the excitement: Oklo's management highlighted NRC modernization moving in a direction aligned with their targeted fleet deployment model of advanced reactors with repeatable designs.
The honest risk: As a pre-revenue company, Oklo faces significant execution risk in deploying its technology and achieving commercial viability.
Asymmetry: Large upside if SMRs reach deployment; deep drawdown given pre-revenue, regulatory and timeline risk.
NuScale Power aims to capitalize on surging baseload demand with its small modular reactor (SMR) technology, offering siting flexibility and scalability; the firm is a higher-risk, higher-reward play on the nuclear renaissance.
Why the excitement: NuScale's modularity allows incremental capacity additions as load grows, with first units generating revenue while others are deployed, management noted in the Q1 FY2026 earnings call.
The honest risk: NuScale's revenue decreased to $0.6 million for the three months ending March 31, 2026, compared to $13.4 million in the prior year, primarily due to the completion of prior licensing agreements.
Asymmetry: High upside on first deployments; sharp drawdown on project delays or financing gaps.
Talen Energy is a leveraged play on firm power pricing, anchored by an ~$18B, 1,920 MW Susquehanna data-center deal with Amazon. The company's strategic focus on data center contracting positions it to capitalize on surging electricity demand from AI.
Why the excitement: Talen Energy's Q1 FY2026 transcript highlights increasing demand with no meaningful increase in supply, validating the company's data center contracting strategy.
The honest risk: Talen Energy's high debt-to-equity ratio of 6.34 presents a significant risk if power prices or contract terms become unfavorable.
Asymmetry: Upside on power-price strength and new hyperscaler contracts; drawdown on rate or regulatory shifts.
The sub-layers and the leaders that anchor each one.
AI's binding constraint is electricity, not just chips. US data-center power demand is inflecting after two decades of flat load, so the companies that can supply firm power and grid equipment capture the spillover from AI capex.
Hyperscalers want firm, carbon-free, 24/7 power, and nuclear is the clearest source. Several have signed direct nuclear power-purchase agreements, which is why operators and SMR developers sit at the center of the theme.
Large operators offer lower-volatility exposure to rising power prices; SMR developers like Oklo and NuScale carry far higher asymmetry — bigger upside if next-gen nuclear deploys, deeper drawdowns if it slips.
It weights revenue and contract growth, margin and cash strength, and forward visibility — favoring producers with signed PPAs over pre-revenue developers.
The risks are a slowdown in data-center demand, regulatory or permitting delays for new nuclear and gas, and power-price volatility that can swing IPP earnings sharply.
AI InfrastructureCopper — The AI BottleneckBitcoin Miners to AI HPC
투자 자문이 아닌 교육 콘텐츠. 과거 성과가 미래 결과를 보장하지 않습니다.