Two companies are about to test whether nuclear power can solve AI's massive electricity problem. Aalo Atomics and Crusoe Energy announced a partnership on July 30 to build a nuclear-powered AI data center at Idaho National Laboratory (INL) in 2027, combining Aalo's advanced reactor technology with Crusoe's modular data center platform. This proof-of-concept could reshape how the industry tackles one of its biggest challenges: powering the next generation of artificial intelligence infrastructure without overwhelming existing electrical grids.

Why Is AI's Power Demand Becoming a Crisis?

AI data centers consume roughly 10 times more electricity per server rack than traditional internet data centers, creating unprecedented strain on power systems. The International Energy Agency projects that global data center electricity consumption could more than double to around 945 terawatt-hours by 2030, accounting for just under 3 percent of global electricity use. In the United States alone, data centers used approximately 4.4 percent of electricity in 2023 and could account for 6.7 to 12 percent by 2028, according to the Department of Energy.

This explosive growth has sparked genuine concern about grid stability and local power availability. When multiple data centers cluster in the same region, they can overwhelm local infrastructure and drive up electricity costs for everyone else. Some communities have responded by imposing moratoriums on new data center construction, fearing the economic and environmental consequences.

How Does the Aalo-Crusoe Partnership Work?

The partnership follows a staged approach. In 2027, Aalo will power a Crusoe Spark modular data center at INL as an initial demonstration. Crusoe Spark is a prefabricated, turnkey data center platform that integrates power, cooling, remote monitoring, and fire suppression into repeatable units that can scale from hundreds of kilowatts to tens or hundreds of megawatts.

Looking further ahead, the companies plan to deploy Aalo Pods and Aalo's 50-megawatt extra-modular reactors (XMR) at Crusoe data centers by the end of 2029. Aalo's Pods are designed specifically for data centers, using five 10-megawatt reactors connected to a shared turbine, arranged as 50-megawatt modular blocks that can scale to larger campuses.

Aalo has already achieved a major milestone. Its Critical Test Reactor reached criticality on July 4, making it the fourth Department of Energy-authorized advanced reactor to reach this stage in about a month. The company is now working on Project Ascension, a 10-megawatt commercial-scale system designed to produce electricity for an on-site data center in the coming months.

What Makes This Different From Other Nuclear-AI Partnerships?

Unlike some other nuclear-AI initiatives that use natural gas as a bridge fuel before transitioning to nuclear power later, the Aalo-Crusoe project tests a nuclear-first model. This means the reactor would power the data center from the start, rather than serving as a future upgrade.

"As one of only four companies to reach criticality by President Trump's deadline for companies in the DOE Reactor Pilot Program, we are now well positioned for not only the full power operation of our 10 MWe Aalo-X reactor, but also the deployment of the Aalo Pod to power commercial data centers," said Matt Loszak, CEO of Aalo Atomics.

Matt Loszak, CEO, Aalo Atomics

Aalo claims it expects future Pod power solutions to be delivered within 12 months from order placement, with additional Pods on shorter repeat intervals. However, these timelines remain contingent on successful full-power reactor operation, factory scale-up, regulatory approvals, fuel supply, and site-specific requirements.

How Are Communities and Policymakers Responding to Data Center Growth?

The debate over AI data centers has become increasingly polarized. Some local governments, like Covington, Georgia, plan to fund their entire government services through property taxes on AI data centers, potentially eliminating property taxes for homeowners. Major companies like Amazon Web Services and Google have announced multibillion-dollar data center investments in North Carolina, generating substantial construction activity and tax revenue.

However, other communities have pushed back. U.S. Senator Bernie Sanders and U.S. Representative Alexandria Ocasio-Cortez introduced a bill earlier in 2026 calling for a federal moratorium on AI data center construction, arguing that Congress needs time to develop regulatory frameworks before deployment accelerates further. Many local governments in North Carolina and elsewhere have passed temporary moratoriums despite having no active data center proposals in their areas.

Steps to Address Data Center Power and Environmental Concerns

• Ratepayer Protection: President Trump's Ratepayer Protection Pledge requires utilities and AI companies to build, bring, or buy new power supply rather than relying on existing infrastructure. Duke Energy and other major utilities have signed the pledge, which now covers roughly 80 percent of power delivered to American homes and businesses, protecting 263 million Americans.
• Transparent Environmental Reporting: The European Union has established a common reporting framework under the Energy Efficiency Directive requiring data centers to disclose electricity consumption, renewable energy contracts, water withdrawal and consumption, cooling methods, and performance during extreme weather.
• Site-Specific Planning: Local authorities should require information about expected water withdrawals, seasonal peaks, emergency operating conditions, and whether public infrastructure costs will be recovered from developers before approving projects.
• Cooling Technology Innovation: Direct liquid cooling can remove heat more efficiently from densely packed processors and may enable higher operating temperatures, reducing mechanical refrigeration needs. Hybrid systems can switch between cooling modes based on temperature and water availability.

What Does This Mean for the Future of AI Infrastructure?

The Aalo-Crusoe partnership represents a critical test case for whether advanced nuclear technology can become a practical solution for powering AI at scale. If successful, it could demonstrate that small modular reactors can be deployed faster and more flexibly than traditional large nuclear plants, potentially addressing both the power and environmental concerns that have sparked community backlash.

However, significant challenges remain. Efficiency improvements in data center design and AI hardware can reduce energy consumption per calculation, but these gains may be offset by rebound effects as companies deploy larger systems and use them for more products. Environmental assessment must therefore measure absolute electricity, water, and emissions alongside efficiency metrics.

The coming years will reveal whether nuclear power becomes a cornerstone of AI infrastructure or remains a niche solution. What's clear is that the industry cannot continue relying on existing grids to absorb the explosive growth in AI computing demand. The 2027 INL demonstration will provide crucial data on whether advanced reactors can fill that gap.