The artificial intelligence boom requires far more electricity than computing power, forcing tech giants and startups to rethink how they secure energy for massive data center campuses. McKinsey estimates AI-related infrastructure spending could exceed $5 trillion by 2030, while developers may need roughly 100 gigawatts of new data-center capacity over the same period. But here's the catch: the electrical grid simply cannot keep up.

For years, investors focused on chips, models, and software as the keys to AI dominance. Today, energy has become the true bottleneck. Utilities across North America and Europe are receiving requests for hundreds of megawatts from individual AI campuses, creating multi-year waits for interconnection studies, transmission upgrades, and transformer deliveries before a single server can be switched on.

Why Is Power the New Limiting Factor in AI?

The scale of AI infrastructure buildout is unprecedented. Technology companies like Microsoft, Amazon, Google, Meta, Oracle, and OpenAI are collectively committing hundreds of billions of dollars annually to AI infrastructure. The largest tech companies can raise capital almost without limitation, but they cannot create transmission lines, substations, grid connections, power plants, transformers, permits, and suitable sites overnight.

According to Kevin O'Leary, investor and Shark Tank personality, the situation is dire: "When you go and look at opportunities in the U.S., I would say 50 percent or more of the data centers that have been announced won't be built because there is no power on the grid". This reality has forced a strategic reckoning across the entire industry.

The challenge extends beyond simply finding power. Hyperscalers need firm, dispatchable baseload power that can run continuously, not intermittent renewable energy. This requirement has made nuclear power increasingly attractive, but traditional nuclear projects face decades-long timelines and massive cost overruns that make financing nearly impossible.

How Are Companies Pairing Gas and Nuclear to Solve the Energy Crisis?

A new strategy has emerged to address the financing gridlock that has stalled nuclear development for years. Several developers are now pairing natural gas turbines with small modular reactors (SMRs), a hybrid approach that delivers power in the near term while building toward clean energy in the long term.

Blue Energy and GE Vernova announced in May 2026 a 2.5-gigawatt collaboration to advance what they describe as the world's first gas-plus-nuclear power plant. The project, planned for the Port of Victoria in Texas, combines two GE Vernova 7HA.02 gas turbines expected to provide approximately 1 gigawatt of power as early as 2030, before transitioning to approximately 1.5 gigawatts of nuclear power as GE Vernova Hitachi Nuclear Energy's BWRX-300 small modular reactors come online as early as 2032.

This phased approach solves a critical problem: it separates early power delivery from later nuclear island completion, reducing financing risk for all parties involved. Blue Energy's core innovation is a reactor-agnostic, modular plant architecture that shifts much of nuclear construction from the field to centralized manufacturing at fabrication yards and existing shipyards, borrowing delivery practices from liquefied natural gas, offshore oil and gas, and offshore wind.

The company claims its model could reduce nuclear plant costs from more than $10,000 per kilowatt to about $2,000 per kilowatt and shrink build times from roughly a decade to two years, with later materials targeting a 48-month-or-less time-to-power timeline for the gas-to-nuclear sequence.

Steps to Understanding the Gas-to-Nuclear Strategy

• Phase One (Near-term): Natural gas turbines provide immediate power within two to three years, allowing data centers to begin operations while nuclear construction continues in parallel.
• Phase Two (Long-term): Small modular reactors come online within four to six years, replacing or supplementing gas generation with clean, firm baseload power for decades.
• Risk Mitigation: By separating power delivery timelines, developers can secure fixed-price contracts and reduce cost overrun risk that has historically derailed nuclear projects.

Blue Energy and Crusoe Energy announced a strategic partnership in October 2025 to develop a nuclear-powered AI data center campus at the Port of Victoria. Under the agreement, Blue Energy secured a site to design, develop, and operate an advanced nuclear power plant of up to 1.5 gigawatts that would deliver power to Crusoe-developed AI factories. Crusoe, an energy-first AI infrastructure company, has increasingly tied its data center growth to dedicated energy supply, with contracted AI infrastructure capacity reaching 4.9 gigawatts across data center projects and cloud capacity as of June 2026.

Other developers have adopted similar strategies. In January 2025, Oklo and RPower unveiled a three-stage model that deploys RPower's natural gas generators over approximately 24 months before transitioning to Oklo's Aurora powerhouses. Oklo later extended that strategy through a July 2025 alliance with Liberty Energy, which brings Liberty's Forte natural gas generation platform to large data center and industrial customers ahead of nuclear baseload.

What Role Are Existing Companies Playing in the Power Race?

Beyond nuclear and gas strategies, other companies are securing power through different approaches. Bitzero, a company that began as a Bitcoin mining operation, has assembled a portfolio of sites, transmission access, land, permits, fiber connectivity, and power agreements spanning Norway, Finland, and North Dakota, representing more than one gigawatt of potential capacity.

The company's flagship Namsskogan campus in Norway operates with direct access to the 132-kilovolt transmission network and low-cost hydroelectric power. In Finland, Bitzero controls the Kokemäki development, where engineering studies outline a path toward approximately 520 megawatts of buildout on its controlled lands with longer-term expansion approaching one gigawatt.

In a significant milestone, Bitzero announced a binding letter with OneQode Networks for a lease on the full 110 megawatts of capacity at its Namsskogan campus. The 15-year lease carries an estimated value of approximately $2.6 billion and is expected to support GPU clusters dedicated to enterprise AI, sovereign AI initiatives, and large-scale model training workloads. This represents the first time a third party entered into a binding letter to commit the full capacity of one of Bitzero's flagship developments to AI infrastructure under a long-term contract measured in billions of dollars.

Microsoft has taken a more direct approach, helping restart a nuclear reactor at Three Mile Island. Google has signed agreements tied to next-generation nuclear power. Amazon, Meta, Oracle, OpenAI, and others are pursuing long-term power arrangements, dedicated generation assets, and large-scale infrastructure investments to secure electricity for future expansion.

The race for power is reshaping investor attention across the broader technology sector. Monolithic Power Systems, which designs and sells semiconductor-based power management chips that convert and control electricity inside AI data center servers and storage, generates about $3 billion in revenue almost entirely from semiconductors. The company sits at the heart of the AI buildout, supplying power management chips to AI data centers with design wins among large customers and exposure to 48-volt and high-voltage architectures.

The fundamental reality is clear: without electricity, there is no AI infrastructure. As hyperscalers commit hundreds of billions of dollars to data center expansion, the companies that can secure reliable, affordable power will determine the future of artificial intelligence development. The gas-to-nuclear hybrid model represents a pragmatic solution to a problem that has stalled nuclear development for decades, finally unlocking the financing and timelines needed to power the next generation of AI factories.