BE research

Position in the Physical Layer of AI thesis

BE is a Power & Grid pivot. VST is the anchor (large nuclear fleet + gas peakers); BE is the distributed-generation pivot. Bloom Energy builds solid-oxide fuel cells that hyperscalers deploy as on-site power generation for datacenters that the grid cannot connect fast enough. The thesis: "behind-the-meter generation, but for the customers that can't get a Microsoft-Talen-style nuclear PPA."

Quinn's book opened BE on 2026-05-05. The 2026-05-05 snapshot had BE at +1.15% (same-day entry). Sized at 2.8% of book.

Recent catalysts (last 30 to 60 days)

Oracle Project Jupiter validation. Oracle Cloud disclosed Bloom as the on-site power provider for the Project Jupiter AI datacenter cluster. First named-hyperscaler customer at scale.
Hydrogen-ready certification. Bloom's fuel cells are now certified for hydrogen feedstock, opening the path to zero-emission operation when hydrogen infrastructure catches up.
South Korea AI Energy MOU. Multi-year framework with SK Group for industrial-scale fuel cell deployment.
Q1 print mixed. Revenue beat, gross margin disappointed. The margin miss is on product mix (more lower-margin AC-coupled units shipped vs DC-coupled hyperscaler units).

The thesis (what has to be true)

1. Hyperscaler grid-interconnect queues stay 5-7 years long. If the FERC + utility reform shortens the queue meaningfully, on-site power demand softens. 2. The Oracle Project Jupiter customer expands. Single-customer concentration is the immediate risk. 3. Hydrogen feedstock evolves on a believable timeline. Today's BE units run on natural gas. The hydrogen pivot is the upside. 4. SK Group and Korean utility partners scale the industrial-side revenue.

Kill vectors (what would break the thesis)

Natural gas pricing shock. BE's economics depend on gas at <$8/MMBtu. A sustained spike makes the units uneconomic vs grid power.
Microsoft or Meta in-house power deal that bypasses BE. Either pivoting to a competitor (Plug Power, Cummins) or to direct nuclear PPAs at scale.
Hydrogen timeline slip. If green hydrogen costs don't fall, the zero-emission narrative stays a story.
Capital event. BE has historically used the capital markets aggressively. A dilutive secondary breaks the discipline.

Layer context

Power & Grid layer. Sister names: VST (anchor, large utility-scale fleet), SMR (pivot, small modular reactors). BE is the distributed/on-site pivot that complements the central-generation theses.

Position discipline

Pivot sleeve rules.
Trim trigger: +100 percent from cost.
4-quarter thesis-fail rule applies. If Oracle Project Jupiter cancels or natural gas spikes structurally, the clock starts.

Moat 6. Asym 7. The moat is moderate because solid-oxide fuel cell technology is patented but reproducible; the operational footprint is what locks customers. The asym is moderate-high because the on-site power category is real and BE is the dominant US name.

Real money. Real position. Real receipts.