The artificial intelligence revolution is transforming not only technology but also how we consume energy. The rise of generative AI and high-performance computing (HPC) is pushing data centers to their electrical limits. In this context, Equinix, one of the world’s largest interconnection and data center operators, has decided to take a step that could redefine the industry’s energy future: investing in fuel cells now and advanced nuclear energy in the future.
Unprecedented Power Consumption
According to projections from the U.S. Department of Energy, data centers could increase their share of the country’s electricity consumption from 4.4% in 2023 to between 6.7% and 12% by 2028. The main driver is the explosion of AI, which requires enormous amounts of real-time computing power.
Today, a single GPU rack for AI can exceed 100 kW of power, and next-generation projects like NVIDIA Rubin clusters are approaching half a megawatt per rack. This poses a challenge not only for tech companies but also for local power grids, many of which are already under stress.
Equinix: From Fuel Cells to Nuclear Power
In response, Equinix has outlined a dual strategy:
Immediate solutions with Bloom Energy fuel cells
- Currently operating 75 MW of fuel cell capacity across 19 IBX centers in six U.S. states.
- Building an additional 30 MW to power critical facilities more cleanly and efficiently.
- Bloom Energy’s solid oxide fuel cells generate on-site electricity, reducing grid dependence and providing more stable supply during demand peaks.
Long-term focus on next-generation nuclear energy
- In 2024, signed a pioneering agreement with Oklo for 500 MW of nuclear capacity using fast reactors that convert waste into fuel.
- Recently added deals with three other partners:
- Radiant: Pre-purchasing 20 Kaleidos microreactors, container-sized, capable of replacing backup diesel generators and deployable in less than 48 hours.
- ULC-Energy: Contract for 250 MW in the Netherlands with Rolls-Royce Small Modular Reactors (SMRs) that can be built in just 500 days.
- Stellaria: Order for 500 MW of its Stellarium molten salt reactor, capable of recycling used fuels and burning long-lived nuclear waste.
The Hybrid Model: Securing Today, Preparing for Tomorrow
Raouf Abdel, Equinix’s EVP of Global Operations, summarized it as:
“Access to continuous electricity is critical for everything from AI-driven drug discovery to cloud video streaming. We have the opportunity and responsibility to invest in reliable, sustainable, and scalable energy infrastructure.”
The key lies in combining solutions: fuel cells to address immediate needs, and advanced nuclear to ensure future capacity. This strategy allows Equinix to meet the explosive demand growth without relying solely on overstressed power grids.
Beyond the Cloud Impact
Equinix emphasizes that these investments will benefit more than just their clients or AI infrastructure. They will positively impact local communities through:
- Strengthening grids with new substations and transmission improvements.
- Creating jobs in constructing, operating, and maintaining new plants.
- Supporting sustainability with emission- and waste-reducing technologies.
The company has dedicated over $1 million in local aid projects alongside its Kansas City data center expansion, aiming to replicate this model with its new energy investments.
A Global Race for the Digital Atom
Equinix is not alone. Other tech giants are also turning to nuclear energy to meet AI’s energy challenges:
- Google partnered with Kairos Power to develop SMR reactors in Tennessee by 2030.
- Amazon announced commitments for 5 GW of nuclear energy, collaborating with Dominion Energy and X-energy.
- Microsoft is exploring fission reactors and has shown interest in fusion energy to power its AI campuses.
The common theme is clear: AI cannot grow indefinitely without stable, safe, and sustainable energy sources.
Challenges and Open Questions
The sector’s push toward nuclear energy raises important questions:
- Regulation and timetables: Many projects depend on government licensing and approvals, which could delay deployments beyond 2030.
- Public acceptance: Despite safety improvements in SMRs and microreactors, nuclear remains controversial in public opinion.
- Technical integration: Deploying modular reactors alongside data centers introduces new engineering and safety challenges.
Nonetheless, the direction is clear: the tech industry can no longer rely solely on renewables and gas to meet increasing AI demand.
Conclusion
Equinix is paving a pioneering path by combining fuel cells with next-generation nuclear power. Their strategy not only addresses immediate data center energy needs but also establishes a sustainable energy foundation for the future of AI and HPC. As data centers become the “factories of knowledge,” energy becomes the new digital gold. Alongside industry leaders like Google, Amazon, and Microsoft, Equinix appears set to lead this atomic race for tomorrow’s cloud.