Wall Street money is moving down the AI supply chain, and power producers are becoming a bigger destination.
Global energy IPOs raised $12.6 billion in the first half of 2026, the highest first-half total since the peak of the dot-com boom in 1999, according to Dealogic data cited by BlockTempo. The figure was also the highest ever for the same period and far above the $4.3 billion raised across all of 2025.
Power is turning into a hard limit for AI expansion
The report ties the fundraising surge to one issue: AI data centers are consuming much more electricity than many people had expected, and securing enough supply is becoming one of the main constraints on the broader AI investment cycle.
A data center may take one to two years to build. A new power plant, or even a major grid upgrade, can take five or 10 years. That gap in construction timelines is now a central problem for the sector.
Capital is lining up around generation and grids
Markets have already started to reflect the shift. GMO launched a power infrastructure ETF this week focused on companies tied to generation, grids and electrification. Nuclear startup Standard Nuclear is also expected to list in the U.S. later in July. Set against the Dealogic numbers, those moves point in the same direction: investors are increasingly trying to buy into electricity supply.
The momentum is not limited to listing volume. Geothermal company Fervo Energy raised about $1.9 billion in its IPO this year, and its shares jumped more than 33% on the first trading day. BlockTempo described it as the largest clean-energy IPO on record.
Market data cited in the report also shows that at least 10 power infrastructure and clean-tech companies are planning listings in 2026. In parallel, U.S. utility M&A reached $203.6 billion in the first five months of the year, already ahead of the full-year 2025 total.
From buying chips to buying electricity
Chris Dendrinos, a clean-energy analyst at RBC, summed up the rotation this way: “Investors first bought AI names like Nvidia, then they realized every chip needs electricity to run.”
That line captures the logic behind the trade. Chips shape what AI can do, but power shapes how much AI can actually be deployed. In that framework, buying electricity-related assets is a way to back the whole AI race rather than trying to pick a single winner among chipmakers or model developers.
The report frames this as a version of the old “selling shovels” thesis. It may be difficult to know which AI company ultimately wins, but demand for electricity rises either way.
A single data center can consume city-scale power
BlockTempo said a typical AI data center uses about 876,000 megawatt-hours of electricity a year. It compared that figure to the annual household electricity consumption of an entire city such as Glasgow in the U.K. or Salt Lake City in the U.S.
The article also notes that AI training and inference require baseload power, meaning supply that remains stable around the clock. On that view, nuclear and natural gas are better suited to support always-on data center operations, while intermittent sources such as wind and solar are less able to carry the load on their own.
Wall Street is reshuffling allocations
Consulting firm ICF expects U.S. power demand to grow 39% from 2026 to 2035, with data-center demand as the main driver.
Manish Kabra, head of U.S. equity strategy at Societe Generale, said, “Power capacity expansion, the return of manufacturing to the U.S., and AI-related infrastructure investment remain core strategic allocations for us.”
The article presents that view as part of a broader shift in how Wall Street is positioning capital, with money moving beyond headline AI equities and into the infrastructure needed to keep the buildout running.
Echoes of the 1999 infrastructure trade
BlockTempo closes by comparing the current move to the late-1990s internet boom, when not only internet companies but also router and fiber suppliers benefited from the rush to build infrastructure. This time, the equivalent assets are power plants, grids and nuclear facilities.
In that reading, the boom in energy IPOs is more than a sector financing story. It is a sign that AI's electricity constraint is starting to shape where capital goes next.