Energy IPOs Hit a Record $11.6 Billion as Wall Street Chases AI Power

Wall Street poured a record $11.6 billion into power and clean-technology IPOs so far in 2026, wagering that electricity has become the AI boom’s tightest bottleneck. At least ten power and clean-tech companies have gone public this year, the busiest stretch on record for the sector. Geothermal driller Fervo Energy set the pace, surging 35% on its Nasdaq debut in May.

The wager looks uniform from a distance. Underneath it, Fervo has slipped about 10% since that debut, X-energy has dropped 38% since its own listing, and one nuclear startup raised barely a quarter of what it hoped to. The AI power trade is real, and it is already sorting winners from losers.

A Record $11.6 Billion Chases the AI Power Problem

At least ten power and clean-technology companies have gone public in 2026, raising more than $11.6 billion combined, the most on record for the sector, Bloomberg reported. The rush spans geothermal drillers, small nuclear reactor makers, gas-engine manufacturers and the companies that keep data center servers cool.

Chips can be shipped from Taiwan. Memory can be shipped from Korea, though chip prices climbing 2.5 times in a year already rattled that supply chain earlier this year. Electricity has to be made where it gets used, and that single fact is now pulling investor money toward an entirely different set of companies.

Fervo Energy, a Houston-based geothermal developer, set the tone in May. Its shares jumped 35% on their first trading day, pushing the company’s valuation past $10 billion just nine years after it was founded.

Why Electricity Became AI’s Tightest Bottleneck

The scale of coming demand explains the rush into public markets. US data centers will need more than 77 gigawatts of capacity by 2030, up from 41 gigawatts in 2025, according to BloombergNEF estimates. Other forecasters see the same curve bending just as sharply.

• Gartner: worldwide data center electricity demand is set to climb 27% this year to 132 gigawatts, then reach 290 gigawatts by 2030.
• International Energy Agency: global data center electricity use, about 415 terawatt-hours in 2024, is on pace to nearly double by 2030.
• US Energy Information Administration: the country is entering its strongest four-year stretch of electricity demand growth since 2000, driven largely by data centers.

Each estimate leans on different modeling assumptions. All of them point toward the same shortfall: the grid was not built for this scale of load, and someone with capital has to close the gap.

Fervo Energy Turns Geothermal Into a $10 Billion Wager

Fervo Energy is the clearest example of how far investors will go. The company drills for geothermal heat using the same horizontal-drilling and hydraulic-fracturing techniques the oil and gas industry perfected, aiming to produce carbon-free electricity around the clock rather than only when the wind blows or the sun shines.

Google, B Capital, Breakthrough Energy Ventures and Devon Energy backed the company privately before its IPO, including a $462 million round in December 2025 that brought Fervo’s total financing to roughly $1.5 billion since its 2017 founding. The May listing dwarfed all of that: $1.89 billion raised and a first-day pop of 35%.

In its own securities filing, Fervo said the top four hyperscalers, Amazon, Google, Microsoft and Meta, had locked in over 133 gigawatts of energy supply agreements by the first quarter of 2026, and that AI data centers are projected to need 552 terawatt-hours of energy over the next decade, citing research firm Rystad.

“New technologies don’t have access to the same capitalization tools,” David Ulrey, Fervo’s chief financial officer, said of the decision to go public instead of relying only on private funding rounds.

The Early Scorecard Is Already Mixed

Not every name in the boom is thriving equally. A few months of trading tell a more complicated story than the headline total suggests.

Deep Fission’s shortfall stood out most. John Vetterli, a partner and co-lead of the global capital markets practice at law firm White & Case, said some companies are struggling to get through the IPO process given the scrutiny and due diligence it now demands.

The pattern is consistent so far. Infrastructure suppliers with paying customers today, like Forgent and Madison Air, are outperforming pure science bets that still need years to reach commercial scale.

Will SpaceX and OpenAI’s Mega-IPOs Crowd Out Smaller Energy Bets?

Probably not, according to bankers arranging these deals, though the risk is real. Wall Street executives argue a strong debut from SpaceX or OpenAI would widen investor appetite across the entire AI infrastructure trade. They agree a stumble in AI’s biggest names would hit smaller energy IPOs hardest of all.

“Capital is concentrating around companies perceived as important to AI deployment and infrastructure build-out,” said Natalie Hwang, founding managing partner at investment firm Apeira Capital Advisors. If the next series of thematic IPOs performs well, she added, it “can boost confidence for a broader set of companies.”

Beau Bohm, global co-head of equity capital markets at Cantor Fitzgerald, put the risk more specifically. He told Semafor the real danger to energy tech IPOs is a softening of the broader AI demand story, not direct competition from SpaceX or OpenAI for investor cash.

Jefferies strategist Christopher Wood struck a more cautious note before SpaceX’s debut, writing to clients that “the most vulnerable stocks should logically be those which have drawn the highest incremental inflows of late, which are the AI picks and shovels plays, as well as of course the hyperscalers.”

SpaceX’s own debut showed how fast that sentiment can flip. Wild swings in its stock after Elon Musk’s aerospace and AI conglomerate went public left pre-IPO perpetual futures traders underwater, a reminder of how quickly these speculative corners can turn.

• Beau Bohm, Cantor Fitzgerald: a strong AI mega-IPO lifts the whole infrastructure trade; the bigger risk is cooling AI demand, not competition for capital.
• Kim Zou, Sightline Climate: data centers give energy startups a “crystal-clear customer base,” but whether pre-commercial technologies like advanced nuclear can scale in time remains an open question.
• Jim Chanos, veteran short seller: markets are confusing a temporary infrastructure bottleneck, permitting delays, transmission limits, turbine shortages, with a permanent electricity shortage.

Chanos argues that markets are confusing a temporary bottleneck with a permanent shortage, pointing to permitting delays, transmission constraints and turbine shortages as the real source of today’s grid strain.

The Next Wave Is Already Filing

Innio Group, a gas-engine maker backed by private equity firm Advent International and Abu Dhabi’s sovereign wealth fund, the Abu Dhabi Investment Authority (ADIA), priced its own offering at $2.4 billion this spring, above the $2 billion it originally targeted. A rival gas-generator maker, ERock, raised $600 million days later.

Behind the scramble is a financing gap Morgan Stanley expects to keep widening. The bank forecasts a 49 gigawatt power shortfall by 2028 in the US alone, even as hyperscalers commit more than $1 trillion in combined capital spending through 2026.

That math is why some operators are skipping the traditional IPO line entirely. Shahal Khan, chief executive of investment firm Burkhan World, is merging his privately held data center power project, Tachyon9, with publicly listed AI infrastructure company Nixxy, skipping a conventional IPO altogether.

“I usually wouldn’t do something like that,” Khan said of the merger. The capital available and how little time he had left him little choice, he said.

If we don’t have power, the entire AI trade can collapse in on itself.

General Fusion is chasing the same urgency at the riskier end of the spectrum. The Canadian company has not yet proven nuclear fusion can generate power outside a laboratory, and it plans to merge with a blank-check company at a $1 billion valuation within weeks.

Frequently Asked Questions

What is the biggest energy IPO of 2026 so far?

Fervo Energy holds that title. The geothermal driller’s May offering raised $1.89 billion and was oversubscribed roughly 15 times, pushing its valuation past $10 billion. Madison Air Solutions’ $2.23 billion April raise separately set a record as the biggest US industrial listing in almost three decades.

Why do AI data centers need so much more electricity than regular servers?

A single hyperscale AI campus can draw more than 2 gigawatts of power on its own, roughly equal to a medium-sized city’s entire electricity use, according to Fervo Energy’s own securities filing. Training and running large AI models requires constant, high-density power that traditional cloud or office servers never needed, which is why hyperscalers are locking in supply agreements years before a data center even opens.

Are energy IPOs part of a broader AI stock bubble?

Some prominent investors think so. JPMorgan chief executive Jamie Dimon warned there is “a lot of exuberance out there,” and Goldman Sachs raised its 2026 IPO forecast to $225 billion from $160 billion the same week SpaceX and Anthropic advanced their own offerings, Fortune reported. Energy IPOs are riding that same wave of capital, which means a broader AI selloff would likely hit them too.

What is a SPAC, and why are risky energy startups using one to go public?

A special purpose acquisition company, or SPAC, is a shell company that already trades publicly and merges with a private business to take it public without a traditional IPO’s vetting and disclosure process. General Fusion, a nuclear fusion startup, and Tachyon9, a data center power project, are both using SPAC mergers in 2026 because it moves faster than a conventional listing for technology still years from commercial proof.

Which AI-power stock had already surged before this IPO wave began?

Bloom Energy is the clearest example. The fuel-cell maker went public back in 2018 and traded near its $15 IPO price for years, then rallied more than 1,300% over the past year as CoreWeave, Oracle, Equinix and AEP signed on to power AI data centers with its on-site fuel cells. Short sellers, including Jim Chanos, are now testing whether that kind of pricing power can last.

Disclaimer: This article is for informational purposes only and does not constitute investment advice. Energy and AI-related stocks and IPOs carry significant volatility and risk, including the possibility of substantial loss. Readers should consult a qualified financial professional before making investment decisions. Figures are accurate as of publication on July 16, 2026.