Microturbine-cogen sub : CGEH as binary-outcome data-center-power play vs BE/SOFC
Thesis
Microturbine-cogen sub : CGEH as binary-outcome data-center-power play vs BE/SOFC
- Type: thesis-shift (subtype: branches — new sub-thesis under the generation-side bull)
- event_id:
2026-05-10-cgeh-microturbine-cogen-sub-thesis-three-source-convergence - Source: @Longviewres sponsored long-form 2026-05-04 (third independent voice on the thesis)
- Shift: Adds a fourth generation-side path alongside (a) BE/SOFC onsite-power, (b) BTC-miner-pivot neocloud, (c) IPP (CEG/VST/TLN), (d) gas-turbine OEM (GEV). The new path: microturbine-cogen via CGEH (Capstone Energy+, formerly Capstone Green Energy / Capstone Turbine, post-bankruptcy emergence). The physics-grounded core: microturbines are ~33% efficient standalone (vs SOFC ~60% LHV), but the "loss" becomes cogen-heat usable by absorption chillers for liquid-cooled racks, taking total-cycle to 80-90%+. Pitched PUE 1.05-1.10 vs industry 1.30-1.40 = ~$18M/yr power-cost savings at a 100MW IT-load AI data center. Same physics-down archetype as our existing key tickers — find the layer where the technology genuinely solves a bottleneck — but here the bottleneck is thermal (where to put rejected heat), not voltage (NuttyCLD's stack) or metallurgy (HWM/ATI blades) or substation-side (GEV/EHV transformers).
- CGEH is the bull side of a clean binary. Linchpin claim: 100MW data-center deal by FY28. Sell-side organic-only consensus ($107.5M FY26E → $156.9M FY28E) assumes zero data-center revenue. Either Capstone signs a sizeable deal in next ~24 months and rerates, or the bull thesis collapses. @Longviewres scenarios (bear $10.11 ~ where it trades / base $34.82 / bull $137.95) telegraph standard sponsored-research sandbagging, but the shape is binary, not ramp-up.
- BE relationship: CGEH is positioned as a complement not a substitute to Bloom. SOFC for steady-state base load (Bloom's 60% LHV story); microturbine-cogen for IT-load thermal management + supplementary power. Both can coexist on the same site. This sharpens our existing BE position rather than displacing it.
- Effect on thesis:
- CGEH added to
ai-power.jsonwatchlist as a 43rd symbol with full provenance trail. Not yet a key_ticker — promotion gated on (i) actual data-center deal-signing event, OR (ii) a CGEH deep-dive confirming the physics-grounded TAM case independently of Longview/phynvesting framing. - phynvesting Substack added as a tracked RSS source. The 2026-04-28 phynvesting capture was the first canary; it sat orphan in
inputs/for 8 days before integration — exact orphan-pattern the audit CLI was built to surface. Tracking the source going forward closes the discovery gap. - No re-grade of existing key_tickers. BE, GEV, CEG, VST, TLN, HWM, ATI, CRWV cohort all unchanged. Microturbine path is additive, not displacing.
- Methodology note: three-voice convergence across 8 days (Discord + Substack + sponsored long-form) is itself a signal worth filing. Different voice classes (anonymous community / independent author / sponsored sell-side) converging on the same physics-grounded conclusion is the substantive-class convergence pattern, opposite of the Aschenbrenner-recycler skip-class convergence rule (see 2026-05-07-skip-class-convergence-inverts-substantive-convergence).
- CGEH added to
- Per-ticker: CGEH ↑ (new watchlist entry; binary-outcome bull-case; not yet key_ticker). BE → (sharpened framing; cogen is complement not substitute). No other changes.
- Open questions for deep-dive (deferred — separate TASKS.md entry): (a) Verify the Monarch recap ($112.5M total, $80M Series A Preferred at $5.00 / 5.0% PIK) terms via primary 8-K; (b) confirm the 42,300 sqft Van Nuys factory's 350/700/1050 MW/yr 1/2/3-shift capacity claim; (c) check the rental-fleet floor at $60/bbl oil for EaaS demand; (d) verify the FY26→FY28 sell-side organic-only forecast assumes zero data-center revenue (most explicit place to check the bull-case math).
Related
5 events DateTypeEvent
May 6 SourceCGEH long-form pitch — third independent source on Capstone vs Bloom data-center-power thesis May 23 ArticleWeird-ETF rotation read — MSTR-leverage breakdown, AI-stack narrowing, defensives flat May 6 AnalysisThread: AI counterparty taxonomy — graduating tickers vs permanent watched entities, with Cerebras May 15 as the forcing function May 3 ArticleAI Power — Week in Review (Apr 27 – May 3, 2026) May 2 PodcastChase Lochmiller (Crusoe CEO) on AI data-center economics No direct external sources are attached to this read.