Data Center Power Race: Bloom Competes on Speed, Fervo on Cost Curve, Ormat on Low Risk
AI summary card
Data Center Power Race: Bloom Competes on Speed, Fervo on Cost Curve, Ormat on Low Risk
Bernstein uses the BTM/FTM/BYOG framework to compare BE, FRVO, and ORA hyperscaler power deals, arguing that Bloom benefits from fast deployment in the near term, Fervo becomes more attractive in the medium term if it lowers EGS costs to $3,000/kW, and Ormat is lower risk but structurally less scalable.
- Although BE, FRVO, and ORA all target demand anchored by hyperscalers or utilities, their commercial structures differ: Bloom is a true BTM model, while Fervo and Ormat are mostly FTM; Bloom/AEP is a BYOG hybrid model with utility financing.
- Bloom's core advantage is speed and optionality, with deployment in about 55-90 days, which is materially faster than the standard 3-8 year interconnection queue, but its LCOE is about $82-115/MWh and it bears natural gas price and stack-replacement costs.
- Fervo's current Cape Station cost is about $7,000/kW, with LCOE around $92-112/MWh; if it achieves a $3,000/kW target, LCOE could fall to about $48-60/MWh, and it has no fuel-price exposure.
- Ormat primarily uses existing geothermal assets for brownfield expansions, with the lowest risk, but resource availability limits its ability to scale deals from the 13-150 MW range to GW-level expansion.
- If interconnection reforms by PJM, ERCOT, and FERC take significant effect before 2030, the queue-bypass premium currently enjoyed by BTM could narrow.
Report interpretation
Overview
The report discusses how Bloom Energy, Fervo Energy, and Ormat Technologies compete for hyperscaler capital under rapid growth in data center power demand using different supply models. It argues that the three companies cannot be simply grouped into one "data center power" theme: Bloom's BTM fuel cell solution bypasses interconnection queues and wins on near-term speed; Fervo's FTM enhanced geothermal play bets on scalable drilling and cost declines; Ormat's FTM geothermal deals are based on existing assets, with lower risk but limited scalability.
Core views
In the 2026-2028 short-term delivery window, Bloom's roughly 90-day deployment speed has a decisive edge versus the 3-4 year FTM interconnection wait. In the medium term from 2028-2033, if Fervo brings EGS costs down to $3,000/kW, FTM geothermal could be cheaper than BTM fuel cells and avoid fuel-price risk. Ormat is attractive on a risk-adjusted basis but is limited in scalability by the number of expandable geothermal assets and is unlikely to be the largest-scale winner.
Analysis framework
The report classifies deals by BTM, FTM and BYOG structures instead of only grouping by a "data center power" theme, then compares deployment speed, interconnection-queue risk, capital expenditure, LCOE, fuel-price exposure, tax credits, project execution risk, and scalability.
Methodology notes
Classify data center power supply models by the location of the power assets, grid-interconnection path, and which party bears the risks.
BTM is on-site generation behind the customer meter that bypasses interconnection queues for that load; FTM is grid-connected generation delivered through utilities or an ISO; BYOG is utility-financed onsite or near-site generation at contracted capacity, where the utility shoulders more credit and financing risk.
Compare fuel cells, EGS geothermal, natural gas, and nuclear solutions using levelized cost of electricity.
Bloom's LCOE is affected by natural gas fuel and stack replacement costs; Fervo remains relatively high-cost today, but if drilling and modular design drive a learning curve, LCOE could compress substantially.
Differentiate near-term winners and medium-term winners using delivery speed versus long-term scalability.
Bloom wins in the near-term capacity gap; Fervo's investment thesis depends on its cost curve coming through after 2028; Ormat wins on lower risk but has a lower scalability ceiling.
Asset mapping & comparison
Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).
- Bloom Energy Corporation (BE)Beneficiary of BTM fuel cell and BYOG hybrid transactions
- Strengths
- Fast deployment of about 55-90 days; can bypass load-related interconnection queues; Oracle and AEP agreements provide clear demand anchors; strong reliability positioning.
- Weaknesses
- Higher LCOE than most wholesale power prices; exposed to natural gas prices, stack replacement costs, and sustainability of the 48E ITC.
- Comparison
- In the short term it has higher delivery certainty relative to FTM geothermal, but if interconnection reforms accelerate or Fervo's costs decline, the speed premium may narrow.
- Risks
- Rising natural gas prices, changes to 48E tax credit, data center demand weaker than expected, and grid upgrades reducing onsite power demand.
- Fervo Energy Company (FRVO)FTM enhanced geothermal scale-up exposure
- Strengths
- If EGS costs fall to $3,000/kW, LCOE could drop to $48-60/MWh; no fuel-price exposure; Google GFA provides large-scale demand support.
- Weaknesses
- Cape Station Phase I is the first commercial-scale project; current cost is about $7,000/kW; still affected by interconnection queue and project execution risk.
- Comparison
- In the medium term, if the cost curve materializes, it could be cheaper than Bloom's BTM fuel cells; compared with Ormat it has greater scaling potential but higher risk.
- Risks
- The drilling learning curve may underperform, permitting challenges, project overruns, startup ramp issues, and thermal conductivity or thermal drawdown performing worse than expected.
- Ormat Technologies, Inc. (ORA)Low-risk FTM incremental supply from existing geothermal assets
- Strengths
- Uses existing interconnection infrastructure and brownfield expansion, giving relatively lower project risk; Google/NV Energy and Switch contracts provide data center demand exposure.
- Weaknesses
- Resource availability limits expansion, with transaction sizes mostly 13-150 MW, making it hard to replicate GW-scale potential of Bloom or Fervo.
- Comparison
- Robust on a risk-adjusted basis but not the largest-scale winner; without internal EGS breakthroughs, scalability is weaker than Fervo.
- Risks
- Expandable geothermal assets are scarce, PUCN approval and commercialization timing are uncertain, competing technology expansion, and data center power demand or power prices below expectations.
Key data
- Bloom Oracle MSAApproximately 2.8 GW; 1.2 GW signed, expected for delivery in 2026-2027A representative BTM case, with fuel cells at Oracle Project Jupiter and directly serving IT load.
- Bloom AEP agreement$2.65B; up to 1 GW; 20-year offtakeDefined in the report as a BYOG hybrid structure, where the utility bears the credit and financing risk for the large load.
- Bloom deployment speedAbout 55-90 daysMaterially faster than standard interconnection waiting cycles, and the source of near-term upside.
- Bloom installation cost and LCOEInstallation cost $3,100-4,000/kW; $2,200-2,800/kW after ITC; LCOE about $82-115/MWhCosts are impacted by natural gas, stack replacement reserve of about $1,000/kW every 5-7 years, and 48E ITC effects.
- Fervo Google GFAUp to 3 GW through 2033; prioritize 1 GW in the first two yearsAn FTM geothermal deal dependent on utility intermediation and interconnection review.
- Fervo Cape Station cost targetCurrently about $7,000/kW; target $3,000/kWIf achieved, LCOE could decline from about $92-112/MWh to about $48-60/MWh.
- Ormat data center dealsSwitch 13 MW; Google/NV Energy up to 150 MWBased on existing geothermal assets and the NV Energy Clean Transition Tariff, and expected to require PUCN approval.
- Ratings and targetsBE Market-Perform / $276; FRVO Outperform / $47; ORA Underperform / $115Current prices for the three are $244.61, $27.13, and $109.77, respectively.
Impact & implications
The investment takeaway is that data center power supply should not be viewed only through theme exposure; it also depends on whether a business can avoid interconnection bottlenecks, whether unit costs can come down, and where project risk sits. Bloom's near-term edge comes from speed and reliability, but valuation and LCOE premium depend on persistent queue bottlenecks; Fervo's upside is tied to validation of its cost curve; Ormat is more of a low-risk, lower-scale incremental supply story.
Risks
- If interconnection reform accelerates, the speed premium from BTM queue bypassing may be compressed.
- Higher natural gas prices would weaken the cost competitiveness of Bloom's fuel cell solution.
- Changes in 48E ITC availability, direct-pay, and transfer mechanisms would affect Bloom's after-tax economics.
- Fervo may fail to lower EGS costs to $3,000/kW as targeted.
- As Fervo Cape Station Phase I is its first commercial-scale project, it faces permitting, construction, cost-overrun, startup ramp, and resource-performance risks.
- Ormat's lower-risk brownfield model is constrained by available geothermal resources and is difficult to scale to GW-level.
- If data center power demand or PPA pricing is weaker than expected, it would reduce deal value across all three names.
What to watch
- Progress on Fervo Phase I startup expected in Q4 2026.
- Fervo disclosures on progress toward a near $3,000/kW cost target.
- Actual interconnection-queue improvements from PJM fast-track, FERC Order 2023, and ERCOT Batch Zero.
- Delivery pacing of Bloom under the Oracle Project Jupiter and AEP BYOG transactions.
- Policy changes to 48E ITC, 45V hydrogen tax credits, and credit transferability mechanisms.
- Approval and go-live timing for Ormat's Google/NV Energy and Switch transactions.