⚠ The AI Capital Cycle Has Never Been TestedHigh threat

Beneath the specific risks to Bloom's demand lies a systemic one: the entire surge rests on the artificial-intelligence capital-spending cycle, which is historic in scale and has never been tested by a downturn. The hyperscalers, neoclouds, and AI labs buying Bloom's on-site power are spending unprecedented sums to build AI infrastructure on the expectation that AI will generate returns to justify it — and that expectation, not any durable feature of Bloom's business, is the ultimate source of the demand. If the AI build-out pauses to digest what it has built, if the returns on AI investment disappoint, or if the frenzy simply cools from its current fever, the demand for fast on-site power would soften quickly, and Bloom sits at the leveraged end of that chain.

The exposure is doubly acute because Bloom is a capital-intensive, lumpy-order business tied to the most sentiment-driven spending in the economy. A digestion phase in AI capital expenditure — a normal and likely feature of any technology build-out of this magnitude — would hit Bloom's product orders directly, and because its recurring base is not yet large enough to cushion the blow, the effect would flow straight through to revenue and, via operating leverage, to profits. The AI demand may well prove durable for years, and the structural case for AI infrastructure is serious. But an investor must recognize that Bloom's entire current story is a leveraged bet on the continuation of an AI-spending cycle that no one has seen through a full boom-and-bust — the same untested cycle that underpins the valuations across the AI-infrastructure complex, and one whose eventual normalization is a question of when, not whether — a question a ~$63B market cap answers optimistically¹.