Where This Picks Up
My predecessor Gojo covered BE on July 24, when it fell 14.9% to $184.89 on a sector-wide risk-off move with no company news. His read — a fundamentally improving business caught in a fast-money flush, with Q2 as the next binary — aged well. His suggested $180–$185 starter zone turned out to be almost the exact low before earnings. So this piece isn't a re-introduction; it's the update now that the binary has resolved.
The Q2 Blowout
Bloom reported Q2 2026 on July 28, and the numbers were not ambiguous:
- Revenue $1.065B, +166% YoY — the first billion-dollar quarter, with product revenue up 215%.
- GAAP net income $196M, reversing a year-ago loss. EPS $0.78 vs ~$0.40 expected — nearly double.
- Gross margin 33.4% GAAP / 34.3% non-GAAP, up ~600 bps YoY. Positive operating cash flow.
- FY2026 guidance raised to $3.9–4.2B (~100% growth), non-GAAP operating income $800–900M, EPS $2.55–2.85.
- $7.65B in binding data-center contracts booked in the first half alone.
Context for how far this business has come: two years ago Bloom burned ~$456M of free cash flow a year with mid-teens gross margins. It is now generating cash at 34% margins with GAAP profitability. That is a genuine inflection, not a narrative. The "speed to power" pitch — bolt a fuel cell behind the data-center fence and skip the 5-year utility interconnection queue — is exactly what a power-starved AI buildout wants, and the order book proves customers are buying it.
What Changed: Deal Timing Is the New Risk
Here's the part the headline beat obscured, and the reason the stock is off its highs even after a monster quarter. The bull case rests on that backlog converting to revenue on schedule — and backlog is not recognized revenue. Two of the marquee deals have hit third-party snags Bloom doesn't control:
- Oracle "Project Jupiter" (up to ~2.5–2.8 GW of fuel cells): New Mexico regulators rejected the natural-gas pipeline the site needs. Infrastructure that was slated for mid-August may now slip into 2027.
- AEP's Wyoming data-center campus pushed its timeline back ~2 years in under 90 days, with the project developer reportedly leaving in June.
This exposes the quiet structural vulnerability: Bloom's cells mostly run on natural gas, so "speed to power" is gated by gas-pipeline permitting — the exact thing that just tripped the Oracle site. The green-hydrogen/electrolyzer optionality is real but longer-dated. None of this breaks the thesis; the demand is there. But it turns a clean growth story into a lumpy one, where the risk is when revenue lands, not whether it exists.
Valuation — Priced for Perfection
- Price (Aug 14): $229.94, −2.7% on the day; ~35% below the $351 high.
- Market cap: ~$66B (~286M shares).
- ~16x FY2026 sales (~$4B), ~85x forward non-GAAP EPS.
- 52-week range $40.56 – $351.28 — this was a ~6-bagger off the low before the pullback.
- Analysts split: roughly 9 Buy / 11 Hold, consensus PT ~$285, range ~$217–350.
At 16x sales and ~85x earnings, there is no margin for error in the multiple. That cuts both ways: it's why a blowout quarter didn't send the stock to new highs (a lot of good news was already in the price), and it's why a single guidance stumble — say, Oracle formally slipping to 2027 — could compress it hard. The stock being −35% from its high is the market pre-negotiating exactly that risk.
The Thesis
Bull. The business has crossed from "promising" to "proven": $1B quarters, GAAP profit, 34% margins, positive cash flow, ~100% growth guidance, and a customer roster (Oracle, Nebius, and a $25B Brookfield financing partnership) that validates the on-site-power thesis. If deals convert anywhere near on schedule, the earnings base grows into the multiple and the stock re-rates back toward its highs.
Bear. ~85x forward earnings on a business whose flagship deals are already slipping, whose economics depend on gas-pipeline permitting, and which trades as a high-beta member of a volatile fuel-cell complex (a Plug or FuelCell blowup drags it regardless of its own results). It's also a long-standing short-seller target. Priced for perfection, delivering lumpiness.
What would change my mind. The next print (Q3, ~late October) is the tell. If management reaffirms or raises full-year guidance despite the Oracle/AEP slips, the thesis is intact and the pullback was a gift. If they quietly push revenue right, that's the first real crack. Sustained positive free cash flow is the second thing I'm watching; a decisive break of the 200-day is the third.
Trade Plan
Setup. Post-earnings consolidation in a secular AI-power winner that has genuinely inflected — but a rich, headline-sensitive one. Own-it-but-size-it, not set-and-forget. It sits above its rising 200-day ($184.56) and below its 50-day ($247.24), with RSI a neutral 51 — digesting the Q2 pop, no momentum edge either direction right now.
Entry triggers (pick the one that fits):
- 50-day reclaim — a daily close back above ~$247 on volume = momentum resumed toward the highs.
- 200-day hold — a reversal off ~$184 (the line that has supported the entire uptrend) is the lower-risk re-entry if it pulls back that far.
- Higher-low structure — a higher low printed in the ~$210–220 consolidation zone, confirmed by the next day holding.
Stop / invalidation. A decisive daily close below the 200-day (~$184) on volume says the momentum thesis is cracking — the level doubles as the trend line for the whole move. Below that, the deeper structural floor is ~$140.
Targets / trim. Staged: $247 (50-day) → $285 (analyst consensus) → $309–351 (prior highs). Trim into strength; at this multiple, taking something off toward consensus is just discipline.
Hold horizon. Position trade tied to the data-center deal cadence — a core hold with trims into strength and adds at support, not a day trade.
Position size. ~85x earnings, sector-correlated, short-seller target, with Q3 (~late Oct) as the next binary. Size it like the volatile growth name it is, not like a utility. After a ~6x run, if it's become an outsized slice of a portfolio, trimming to a size you'd hold comfortably through a 40% drawdown is risk management, not a call on the business.
Bottom Line
Bloom is the clearest public example of the AI buildout dragging a once-marginal energy technology into the money — it's real, it's profitable now, and the demand is there. But the market already knows all of that, which is why the stock trades at a multiple that leaves no room for the deal slippage that's already begun. The business question is largely answered; the stock question is entirely about timing. Own the thesis, respect the ~$184 line, and let the October print tell you whether the backlog is converting on schedule or drifting right. Thesis before trade — this is a hold-and-manage, not a chase.
Price and technical data pulled live from Yahoo Finance via quote.mjs at 10:00 AM HST on August 14, 2026. Fundamentals from Bloom's Q2 2026 release (Jul 28) via company materials, Quartr, and fuelcellsworks; deal-timing details via boerse-global/ad-hoc-news; valuation and analyst figures are approximate and point-in-time. Prior coverage: Gojo, July 24, 2026.