That is the largest commercial deal in the company's history. It followed a $5 billion partnership with Brookfield Asset Management in October 2025 to deploy fuel cells for AI data centers globally.
Two deals. $7.65 billion in committed capital. Eighteen months.
The signal here is not about Bloom Energy specifically. It is about what these deal structures reveal regarding where the grid bypass market for data center power actually sits.
Hyperscalers and utilities are not treating on-site fuel cell deployment as an experiment anymore. AEP is a regulated utility committing to 1GW of SOFC capacity to service data center load it cannot serve through the grid queue on a commercially viable timeline.
That is a procurement decision, not a pilot program.
The underlying driver is straightforward. Grid interconnection queues in the US run three to five years. A hyperscaler breaking ground on a new data center campus today cannot wait for grid capacity. Bloom's solid oxide fuel cells deploy in 90 days and operate at roughly 60% electrical efficiency.
The technology readiness is not the question. The question is whether the supply chain can execute at gigawatt scale.
Bloom announced 2GW of annual manufacturing capacity expansion in August 2025. MTAR Technologies secured a $43.9 million supply agreement for fuel cell components in September 2025. The supply chain is being built in parallel with the order book.
The signal I am watching: AEP deployment commencement confirmation.
The $2.65 billion AEP agreement is an offtake commitment. The commercial validation that matters next is the first confirmed deployment milestone under that agreement.
A physical installation update from AEP in H2 2026 converts the order book into operational evidence. That is what moves the broader utility and hyperscaler procurement conversation from interest to replication.
Watch the commissioning announcement, not the deal announcement.