📊 Full opportunity report: The bridge. Why the AI buildout runs on a nuclear story and a gas reality. on ThorstenMeyerAI.com — validation score, market gap, and execution plan.
TL;DR
AI data centers are primarily powered by natural gas today, despite major nuclear procurement deals promising future clean energy. This gap between supply and demand reveals a complex energy reality driven by timing and infrastructure constraints.
Major hyperscalers such as Meta, Microsoft, Google, and Amazon are investing heavily in nuclear power deals, but the actual energy powering their data centers today is predominantly natural gas. This discrepancy between the nuclear procurement rush and the immediate energy needs highlights a critical timeline gap in the AI industry’s energy strategy.
While companies have announced nuclear deals totaling up to 6.6 gigawatts, most of this capacity will not be available until the late 2020s or early 2030s. For example, Microsoft’s restart of Three Mile Island is expected to deliver 835 megawatts by 2027, and Google’s small modular reactors (SMRs) are not expected online until between 2030 and 2035. Meanwhile, the demand for power at AI data centers is urgent, with construction timelines of 18 to 24 months and grid interconnection delays ranging from three to seven years in the US, and up to thirteen in parts of Europe.
To bridge this gap, industry sources report that more than 40 gigawatts of behind-the-meter and co-located generation are being deployed, primarily natural gas turbines, reciprocating engines, and fuel cells. These installations are built on-site or off-grid, bypassing grid constraints and regulatory delays, and are the actual infrastructure powering AI data centers today. This behind-the-meter gas buildout is driven by the need for immediate, reliable, and firm power, contrasting sharply with the long-term, clean energy promise of nuclear.
The bridge.
Why the AI buildout runs
on a nuclear story and
a gas reality.
to early 2026 · the real rush
2027-2035, grid 3-7 years
generation · near-term mostly gas
(~10M cars) · Cornell analysis
- A data center is built in under two years
- Data center electricity use +17% in 2025, doubling by 2030
- Gartner: 40% of AI data centers electricity-constrained by 2027
- Three Mile Island ~2027 · Oklo ~2030 · Kairos 2030-2035
- No commercial SMR yet operates in the US
- Grid interconnection 3-7 years (up to 13 in Europe)
early 2030s
· mostly gas
The industry leads with the nuclear it has bought for the end of the decade and builds the gas it needs for now — and sites that gas behind the meter where it moves fastest and shows least. The behind-the-meter siting is the tell that the bridge will be here longer than the word implies.Thorsten Meyer · The Bridge · AI Energy 03
Implications of the Energy Buildout Timeline Mismatch
This divergence between the nuclear procurement narrative and the gas-based infrastructure being built today has significant implications for the AI industry’s carbon footprint. While the long-term vision includes a shift to nuclear power, the immediate reliance on fossil fuels means that current AI operations are contributing to ongoing emissions. The gap also raises questions about the industry’s ability to meet its sustainability commitments and the true timeline for decarbonization.
Furthermore, the situation underscores the structural challenge of aligning infrastructure development with demand, highlighting how market, regulatory, and technical delays influence energy choices. The reliance on gas as a bridge is both a pragmatic response to current needs and a potential obstacle to achieving the promised green transition.
Westinghouse 14500 Peak Watt Tri-Fuel Home Backup Portable Generator, Remote Electric Start, Transfer Switch Ready, Gas, Propane, and Natural Gas Powered
Perfect as a backup power source for larger homes or a dependable source of portable power
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Nuclear Deals and Construction Delays in AI Power Planning
Major technology companies have signed nuclear procurement agreements, including Meta’s deals for up to 6.6 gigawatts and Google’s first corporate SMR agreement, with capacity expected to come online between 2030 and 2035. However, actual nuclear capacity, such as Microsoft’s restart of Three Mile Island, is only expected to deliver a fraction of that power in the near term. Meanwhile, nuclear construction projects like Vogtle in the US have experienced significant delays and cost overruns, illustrating the challenges of deploying new nuclear capacity on the required timeline.
In contrast, the immediate power needs of AI data centers are being met by behind-the-meter gas generation, which can be deployed rapidly and flexibly. Industry insiders report that this gas buildout is outpacing the arrival of nuclear capacity, creating a temporal mismatch that shapes the current energy landscape for AI infrastructure.
“The nuclear deals are real and coming; the gas is real and here; and the years between them are the bridge.”
— Thorsten Meyer
Handbook of Small Modular Nuclear Reactors: Second Edition (Woodhead Publishing Series in Energy)
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Unresolved Questions About Long-Term Energy Strategy
It remains unclear whether the nuclear capacity will arrive on schedule or continue to slip, which could extend reliance on fossil fuels. The future of SMRs and their commercial viability remains uncertain, as no operational SMRs currently exist in the US, and past nuclear projects have faced significant delays and cost overruns. The question of whether the gas buildout is a temporary bridge or a long-term solution is also unresolved.
off-grid natural gas turbines
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Upcoming Milestones and Industry Developments
In the coming years, the industry will closely monitor the progress of SMR commercialization and nuclear project completions. Simultaneously, the deployment of behind-the-meter gas generation will continue to expand, potentially shaping the immediate energy landscape. Regulatory and grid interconnection processes will also influence how quickly new capacity, nuclear or gas, can be integrated into the power supply for AI data centers.
Further analysis will be needed to assess whether the nuclear promises will materialize in time or if the reliance on fossil fuels will persist, impacting the industry’s emissions trajectory and sustainability commitments.
backup power fuel cells for data centers
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
Why are AI data centers currently powered by gas if nuclear deals are so significant?
Because nuclear capacity, even if promised, will not be available in the short term due to delays in construction and licensing. Gas turbines and other behind-the-meter generation are deployed rapidly to meet immediate power needs.
Are the nuclear deals genuine or just marketing?
The nuclear deals are real and reflect significant investment and commitment by hyperscalers. However, the capacity from these deals will not be available until late in the decade, making them a long-term solution rather than an immediate one.
What are the emissions implications of relying on gas now?
Using natural gas for power generation produces carbon emissions, which counteracts some of the industry’s sustainability goals. The current reliance on gas creates a gap between the industry’s clean energy aspirations and its actual emissions footprint.
Could SMRs accelerate the transition to clean energy for AI data centers?
If SMRs become commercially viable and are deployed on schedule, they could significantly reduce reliance on fossil fuels and help meet sustainability targets. However, their current unproven status and potential delays mean this remains uncertain.
Is the reliance on gas a temporary or permanent solution?
It is unclear. If nuclear capacity is delayed or fails to meet expectations, the reliance on gas could become a more permanent feature of the energy landscape for AI infrastructure.
Source: ThorstenMeyerAI.com
