Why $1.7B Is Betting on Fuel Cells Over Nuclear for AI

Industrial Development Funding and Oaktree just committed $1.7 billion to power Nebius' AI data centers with Bloom Energy fuel cells. It's the second nine-figure fuel cell bet in under a year, following Brookfield's $5 billion partnership with Bloom. Fuel cells are winning AI power contracts on speed. Emissions profile alone is not the deciding factor.

>> In Other News: Airbus Unveils Study on the Potential of Sustainable Aviation Fuel to Support Economic Growth and Accelerate Canada's Climate Transition

What Did IDF and Oaktree Just Announce?

Industrial Development Funding and Oaktree announced $1.7 billion in project financing on July 16, 2026. The money funds Bloom Energy fuel cell systems for Nebius, an AI cloud computing provider.

The power runs behind the meter. That means it sits directly at the data center instead of routing through the public grid. IDF leads development on the project. Oaktree holds a minority equity stake.

Morgan Stanley served as the sole tax equity investor and placement agent. MUFG Bank provided the senior debt financing. This deal pushes the IDF-Bloom relationship past $2.6 billion in combined project value.

How Does This Fit the Brookfield-Bloom Pattern?

This is not Bloom Energy's first billion-dollar AI power deal. Brookfield signed a $5 billion strategic partnership with Bloom in October 2025 to power AI data centers globally.

By June 30, 2026, Brookfield had grown that commitment fivefold, to $25 billion. Two of the largest infrastructure investors in the world have now placed multibillion-dollar bets on the same fuel cell platform. Both moves came within nine months of each other.

The repeat pattern points to one reason buyers keep choosing Bloom. Nebius picked the technology for its installation speed, its cleaner-burning profile, and its ability to track AI workload demand. Speed is doing most of the work on that list.

RELATED: US Carbon Capture Race: $77B Industry Shifts Global Balance

Why Are Fuel Cells Beating Nuclear SMRs to the AI Market?

Fuel cells win because they deploy in months, not years. Bloom's systems install directly on site. They can start producing around 100 megawatts in about 90 days.

Small modular reactors cannot match that pace. Kairos Power, X-Energy, and TerraPower do not expect a fully working SMR before 2030. New reactor builds also need fresh federal licensing. That step alone can take 3 to 5 years.

Grid-scale renewables and new transmission lines face similar delays. New transmission typically takes 5 to 10 years to build in U.S. markets. The data center industry expects electricity shortages as early as 2028.

Power Source	Typical Deployment Timeline
Bloom Energy fuel cells	Around 90 days on site
Natural gas peaker plants	3 to 5 years
New grid transmission lines	5 to 10 years
Small modular reactors	First units expected around 2030

That gap explains why $1.7 billion is landing on fuel cells this month, not a reactor order. Right now, speed decides who wins AI power contracts. Emissions profile is a secondary factor for buyers racing the clock.

Are Bloom's Fuel Cells Running on Clean Hydrogen?

Not yet, mostly. Bloom's solid oxide fuel cells at sites like this one run mostly on natural gas today. Green hydrogen still costs more. The hydrogen supply chain is also still young.

Even so, solid oxide fuel cells burn cleaner than standby diesel or gas turbines. They produce next to no nitrogen oxides, sulfur oxides, or particulate matter. They are not zero-carbon when running on natural gas, and that distinction matters for an honest read of the technology.

The gap between promise and current fuel mix has mattered before. Amazon backed out of a 2023 deal to run Bloom fuel cells at three of its Oregon data centers. It withdrew from that plan in 2024, after state regulators flagged that natural gas-powered cells would add real emissions in a region already served largely by clean hydropower.

The hardware itself is not locked to one fuel. Bloom's solid oxide platform also powers its electrolyzer line, built to produce hydrogen on the same underlying technology.

As regional hydrogen hubs scale up supply, and as hubs like California's ARCHES and the HyVelocity Gulf Coast hub work through funding shifts, the cost gap between natural gas and clean hydrogen should narrow. Fuel cells today work best as a bridge technology. They start on natural gas now and move toward cleaner feedstock as it becomes affordable.

What's Next for Fuel Cells in AI Infrastructure?

Expect more deals shaped like this one. Behind-the-meter power solves a real bottleneck for developers stuck waiting years for a grid interconnection.

The broader decarbonization sector is watching closely. AI power demand is pulling capital into adjacent fields too. Direct air capture developers are pairing carbon removal with data center power sales. DAC deployment more broadly is racing to scale alongside the same AI infrastructure boom.

• Fuel cells offer 90-day installs versus multi-year timelines for gas peakers or transmission upgrades
• IDF and Bloom's combined project portfolio now tops $2.6 billion
• Brookfield's Bloom financing framework grew from $5 billion to $25 billion in nine months

Other decarbonization sectors are riding the same wave. Carbon storage capacity is expanding alongside data center growth. Sustainable aviation fuel production has climbed sharply on the strength of similar federal incentives. SAF production capacity roughly doubled in early 2025, showing the same investment appetite now flowing toward fuel cells.

Fuel cells did not win this round of AI power contracts because they are the cleanest option on paper. They won because they are the only technology that can show up on time. Whether that lead holds once hydrogen supply chains catch up is the question worth watching next.

For ongoing coverage of carbon removal, BECCS, and corporate CDR procurement, subscribe to Decarbonfuse.com.