Three energy companies have gone public in 2026, and they have one thing in common: they are all selling electricity to a grid that the AI industry has put under serious strain.
Solv Energy, which builds solar and battery storage projects, debuted in February at a valuation of roughly $6 billion. It mentioned data centers more than a dozen times in its filings with US regulators before listing. X-energy, building small modular nuclear reactors, followed in April and its valuation hit $11.5 billion on its first day of trading. Amazon is both a client and a reported shareholder holding close to 20% of the company. Fervo Energy, a geothermal firm backed by Google, raised $1.89 billion in its May IPO and now trades at a market cap of around $12.4 billion.
This is not a coincidence. The electricity demand from AI data centers is forecast to roughly double globally by 2030. In the US, forecasters including Gartner put data center electricity usage rising from 4% of national consumption today to nearly 8% by 2030. One research group estimates that in a high-growth scenario, data centers could account for 14% of total US power generation by 2030. That is a staggering shift for a grid that had been essentially flat in consumption for the better part of a decade.
What makes the clean energy angle particularly interesting is the political moment. The current US administration has cut support for wind energy and slowed approvals for new projects. Yet clean energy stocks have been on a strong run for over a year, because the demand pressure from data centers overrides the political headwinds. Solar and geothermal are among the fastest technologies to build and deploy. Nuclear takes longer but promises round-the-clock output, which AI data centers specifically need: they cannot simply shut down when the wind stops.
The tech giants driving AI infrastructure have already locked in long-term power agreements with these companies. Google has been investing in Fervo for years and has a dedicated clean energy tariff agreement with them. Amazon holds a stake in X-energy and is a paying customer. These deals happen years before the power flows, which means the best available supply is increasingly committed to a handful of hyperscale buyers before anyone else can access it.
For a business operator outside the tech sector, this creates a practical problem. If you run a manufacturing facility, a large retail distribution center, or a hospital that is expanding, you are competing for grid connections and power supply in the same regional markets where data centers are growing fastest. Supply chain bottlenecks, backlogged interconnection queues, and slow permitting processes are already cited by analysts as the main constraints on adding new generation. The IPO valuations for these three companies reflect investor confidence that the supply problem is large and persistent enough to be very profitable to solve. That is not good news for anyone who needs power and is not a tech giant.
The Illinois AI safety law passed this week adds a separate layer worth watching. The state passed a bill requiring independent third-party safety audits of the largest AI companies, and the governor has said he will sign it. It passed the House 110 to 0 and the Senate 52 to 5. The law only applies to companies with over $500 million in annual revenue building the most advanced AI models, so it does not touch most businesses directly. But it is the first US law to mandate external audits of AI safety practices, and California and New York already have related laws. The pattern of state-level regulation filling the gap left by federal inaction is worth watching: just as California's privacy law effectively became a national standard because companies found it easier to comply uniformly, Illinois may be setting a floor that the entire industry ends up meeting.
Separately, the public disagreement between OpenAI and Anthropic over AI's impact on jobs is a signal in itself. OpenAI's CEO Sam Altman said this week he was wrong about AI eliminating entry-level white-collar jobs at the pace he predicted, while Anthropic's co-founder said large-scale displacement remains a real possibility. Both companies are reportedly preparing IPOs this year at valuations in the hundreds of billions. The data available so far, including research from Yale's Budget Lab, shows no meaningful change in unemployment in jobs with high AI exposure through early 2026. The jobs story is genuinely unresolved, and anyone telling you otherwise with confidence is working from incomplete information.