Why the oil majors' integrated gas supply creates a structural pricing moat over independent power producers. As grid constraints and private credit financing challenges mount, integrated oil majors offer the only lifeline for hyperscalers to continue the growth trajectory.
GE Vernova GEA35768 (05/2025) · GridLab/APPA (2025) · CPV/OPC (2020) · Model audit: 138 checks, 0 errors
Now that data centers have scaled beyond what the grid can support, gas price becomes the most important value driver in their economics (which are measured on a $/token/watt basis). Those located near cheap, abundant natural gas, i.e. Permian Basin and Marcellus Shale, will be structurally advantaged relative to others.
Competitive Power Ventures (CPV) capitalized on low-cost shale gas to deliver the most advanced fleet of combined cycle gas turbine projects in the U.S., most drawing from Marcellus Shale gas. This framework will be the formula that delivers the next phase of hyperscale data centers.
| Project | MW | Heat Rate | Turbine | Market | COD | Contract |
|---|---|---|---|---|---|---|
| Shore, NJ | 725 | 6,698 | GE 7FA.05 | PJM-EMAAC | 2016 | Revenue put (BP) |
| St. Charles, MD | 745 | 6,856 | GE 7F.05 | PJM-SWMAAC | 2017 | Revenue put (Shell) |
| Valley, NY | 720 | 6,844 | Siemens 5000F | NYISO-LHV | 2018 | Revenue put (MS) |
| Towantic, CT | 805 | 6,425 | GE 7HA.01 | ISO-NE | 2018 | Capacity lock $9.55/kW-mo |
| Fairview, PA | 1,050 | 6,419 | GE 7HA.02 | PJM-MAAC | 2019 | Revenue put (BP) |
| Three Rivers, IL | 1,258 | 6,356 | GE 7HA.02 | PJM-COMED | 2023 | Gas netback (47%) |
| CPV Wtd Avg (Operating) | 4,045 | 6,637–6,844 | ||||
| Non-CPV Regional Fleet Avg | 8,595–11,391 | Older F/E-class | Heat rate disadvantage = dispatch risk | |||
Same playbook, next chapter: 7HA-class turbines + cheap shale gas + behind-the-fence delivery. The only difference is the offtaker is now a hyperscaler instead of PJM.
Source: OPC Energy / CPV Acquisition Investor Presentation (Sep 2020). Heat rates on HHV basis. Note: behind-the-meter (BTM) hyperscale deployments carry limited dispatch and heat rate risk relative to grid-connected CPV fleet, as BTM plants run baseload to a dedicated offtaker and are not subject to ISO merit-order dispatch or wholesale market clearing dynamics.
Contract structure note: The derivative option market (revenue puts, gas netbacks, capacity locks) is not deeply liquid and is controlled by a few strategic IOCs and investment banks (BP, Shell, Morgan Stanley). If this market is closed to IPPs for strategic considerations, it will enhance negotiating leverage for equity positions in the power generation projects for the IOCs.
Illustrative hyperscale PPA: 10yr initial term (fixed price), 10yr second term reset at "market rates" w/ fixed spark spread
| Henry Hub ($/MMBtu) | $2.00 | $2.50 | $3.00 | $3.50 | $3.75 | $4.00 | $4.50 | $5.00 |
|---|---|---|---|---|---|---|---|---|
| Nameplate (5,944) | 32.5% | 29.1% | 25.7% | 22.3% | 20.5% | 18.8% | 15.4% | 12.2% |
| Operating (6,300) | 31.5% | 27.9% | 24.3% | 20.6% | 18.8% | 17.0% | 13.5% | 10.1% |
| Conservative (6,400) | 31.2% | 27.5% | 23.9% | 20.1% | 18.3% | 16.5% | 12.9% | 9.5% |
| XOM Internal Transfer | XOM at ~$2.00 → 31–33% levered returns | |||||||
Gas breakeven for 10% levered IRR: Nameplate ~$4.10 · Operating ~$3.80 · Conservative ~$3.65
Methodology: Levered equity IRR on 40-year cash flow. CapEx: $1,677mm all-in ($1,300/kW EPC + 10% contingency + $35mm dev). 60/40 D/E at 5.30%. 5-yr MACRS (80% eligible, 10% bonus). IRR computed via scipy.optimize (±2 bps vs Excel XIRR).
Capacity factor: 85% CF × 95% availability = 80.75% effective utilization (~70 days/yr downtime). Accounts for: hot gas path inspection (every ~25,000 EOH, 2-4 weeks), annual combustion inspection (1-2 weeks), major overhaul (every ~50,000 EOH, 4-6 weeks), and forced outage allowance (~1-2 weeks/yr). Conservative for a dedicated BTM baseload plant with no ISO merit-order dispatch risk — grid-connected CCGTs average 57-64% CF (EIA, 2022) due to renewable curtailment, not equipment limitation.
At every return target, the integrated major undercuts the IPP by ~$11/MWh. Even at 30% ROE, the major's price is at or below the IPP's minimum viable offer.
| Provider / Scenario | 18% ROE | 20% ROE | 22% ROE | 24% ROE | 26% ROE | 28% ROE | 30% ROE |
|---|---|---|---|---|---|---|---|
| XOM / CVX — Internal Gas @ $2.00 | |||||||
| Operating (6,300) | $53 | $55 | $57 | $58 | $60 | $62 | $64 |
| Conservative (6,400) | $53 | $55 | $57 | $59 | $60 | $62 | $64 |
| IPP — Market Gas @ $3.75 HH | |||||||
| Operating (6,300) | $64 | $66 | $68 | $70 | $71 | $73 | $75 |
| Conservative (6,400) | $65 | $67 | $68 | $70 | $72 | $74 | $75 |
| IPP — Stressed Gas @ $4.50 HH | |||||||
| Operating (6,300) | $69 | $71 | $73 | $74 | $76 | $78 | $80 |
| Conservative (6,400) | $70 | $71 | $73 | $75 | $77 | $78 | $80 |
The $11/MWh advantage holds regardless of construction cost. As EPC inflation pushes IPPs toward $80/MWh, the integrated major remains competitive at $68.
| Base | +20% | +50% | |
|---|---|---|---|
| EPC ($/kW) | $1,300 | $1,560 | $1,950 |
| All-in Cost ($mm) | $1,677 | $2,005 | $2,497 |
| XOM — Internal Gas @ $2.00 | |||
| Operating (6,300) | $53 | $59 | $68 |
| Conservative (6,400) | $53 | $60 | $69 |
| IPP — Market Gas @ $3.75 | |||
| Operating (6,300) | $64 | $70 | $80 |
| Conservative (6,400) | $65 | $71 | $80 |
| IPP — Stressed Gas @ $4.50 | |||
| Operating (6,300) | $69 | $75 | $84 |
| Conservative (6,400) | $70 | $76 | $85 |
| Hyperscaler Saves | $11.2 | $11.2 | $11.2 |
$80/MWh approaches retail power prices. Hyperscalers will refuse. The IPP is priced out of the market. XOM/CVX at $68 remains competitive.
Construction cost inflation hits both XOM/CVX and IPP equally. The $11/MWh savings doesn't change because it's driven entirely by the $1.75/MMBtu gas cost differential, not construction cost. Rising CapEx doesn't erode the advantage, it eliminates the IPP as a viable competitor, leaving XOM/CVX as the only counterparties that can deliver power at a price hyperscalers will accept.
Market is trending toward $2,000+/kW (GridLab, Sep 2025). At that level, only integrated majors can build new CCGTs at competitive PPA pricing.
CapEx build-up: All-in cost = (EPC $/kW × 1,148 MW) + 10% contingency + $35mm fixed development cost. The +20% and +50% scenarios apply to the EPC rate only. Development costs are fixed, so all-in totals do not scale proportionally with EPC inflation (e.g. +50% EPC = +49% all-in).
"I'm hopeful that many of these hyperscalers are sincere when they talk about the desire to have low emission facilities, because certainly in the near to medium term we're probably the only realistic game in town to accomplish that."
Table 2 proves it. XOM can offer the hyperscaler a lower PPA price at 30% ROE ($64/MWh) than the IPP can offer at its minimum viable return of 18% ($64/MWh). The integrated major earns a premium return while simultaneously undercutting every competing bid. The $11/MWh gap isn't a negotiating outcome, it's embedded in the cost structure.
CCS widens the moat. When you layer carbon capture costs (~$15-20/MWh) onto both providers, the integrated major's advantage grows because XOM's CCS infrastructure (Denbury network, $4.9B acquisition) is internalized while the IPP must contract at market rates. The more the hyperscaler cares about decarbonization, the fewer counterparties can deliver it. And it's not economics alone: CCS at scale requires subsurface engineering, CO₂ pipeline permitting, and reservoir management expertise that only the integrated majors possess. An IPP can't buy this capability off the shelf.
"We're not interested in going into a utility business like power generation. Utility returns for power generation would not compete in our portfolio."
Table 1 shows why. At $2.00 gas, XOM earns 31-33% levered. The 18% floor in this model is conservative — XOM strategic projects target 25%+. At 25% ROE, XOM offers $59/MWh — still $5 below the IPP's 18% floor.
Table 3 proves it. At $1,950/kW, IPP needs $80/MWh. XOM needs $68. The $11 savings holds regardless of construction cost. Rising CapEx raises barriers to entry, eliminates IPP competition, and concentrates pricing power with the integrated major. In a normal market, cost inflation would deter additional investment. But AI is existential and a major national security priority. The hyperscalers will continue to scale regardless of cost, and they will contract with whoever can deliver.
Sources: CNBC (Oct 31, 2025) · Fortune (Feb 1, 2025) · NGI (Feb 7, 2025) · Semafor (Dec 13, 2024). Model: 138 checks, 0 errors. Python (scipy.optimize) verified ±2 bps vs Excel XIRR.
Every major announced project uses GE Vernova 7HA-class turbines, integrated gas supply, and bypasses the grid
| Project | Capacity | Turbines | EPC / Developer | CapEx | Gas Source | COD | Model |
|---|---|---|---|---|---|---|---|
| Homer City Energy Campus | 4.5 GW | 7× GE 7HA.02 (H₂-enabled) | Kiewit EPC; Knighthead Capital | $10B+ total (incl. site remediation, coal ash closure); est. ~$1,400/kW gen-only | EQT / Marcellus (TETCO + Eastern Gas) | 2027 | Co-located DC campus on 3,200 acres; legacy 345kV PJM/NYISO interconnection; avoids 5-7yr queue* |
| Crusoe / Engine No. 1 / Chevron | 4.5 GW | 7× GE Vernova (purchased by Engine No.1 / CVX) | Crusoe (behind-the-meter) | Undisclosed | Chevron JV partner (gas source TBD) | TBD | Behind-the-meter islanded; bypasses grid entirely |
| ExxonMobil Data Center Power | 1.5 GW | GE Vernova (expected) | TBD (Kiewit/Bechtel relationships) | ~$1.9B est. (~$1,200/kW) | XOM integrated supply (basin TBD) | 2027-28 | Dedicated DC power plant; XOM integrated gas + power + CCS |
| Olara Base Case (This Model) | 1.1 GW† | 2× GE 7HA.02 (2x1 MS config) | Greenfield assumptions | $1.7B ($1,460/kW) | XOM internal (est. ~$2.00/MMBtu) | 2028 | Integrated major behind-the-fence; $65+2% esc / market reset |
Three of the four largest announced gas-for-data center projects use GE Vernova 7HA.02 turbines. Total announced islanded capacity: 10.5+ GW across Homer City (4.5), Crusoe/Chevron (4.5), and XOM (1.5).
All three bypass the grid interconnection queue. Homer City uses legacy coal plant 345kV transmission infrastructure, Crusoe operates behind-the-meter, and XOM plans dedicated islanded power.
The common thread: at multi-GW scale, the grid cannot deliver. Islanded gas with integrated fuel supply is the only model that solves for timeline, cost, and reliability simultaneously. This is exactly the thesis the Olara model quantifies.
Sources: ENR/Business Wire (Homer City, Apr 2025); HCR/EQT gas supply agreement (Jul 2025); DCD (Crusoe/Chevron, Mar 2025); Reuters (XOM, Dec 2024). Homer City $10B+ is all-in incl. brownfield remediation, coal ash closure, site acquisition, and DC readiness; generation-only CapEx est. ~$1,300-1,550/kW after stripping site costs.
*Homer City risk factor: Site straddles PJM/NYISO territories. Gas interconnection permitting may face headwinds given NYISO exposure. The Hochul administration has been broadly hostile to new fossil generation in New York, which could complicate pipeline permitting and air quality approvals for any capacity serving the NYISO side of the interconnection.
†Model assumes 2x1 MS configuration (1,148 MW). Economics would improve at 1.5 GW+ as fixed costs (pipeline interconnection, site development, BOP infrastructure, $35mm development) spread over ~30% more capacity, reducing all-in $/kW by an estimated $50-80/kW.
This document is a confidential research publication prepared by the Sheephill Group for informational and discussion purposes only. It does not constitute investment advice, a solicitation, or an offer to buy or sell any securities or financial instruments.
Not a Registered Investment Adviser. The Sheephill Group is not registered with the SEC, FINRA, or any state securities regulatory authority. No content in this presentation should be construed as personalized investment advice.
Methodology. Sensitivity outputs computed via Python (scipy.optimize.brentq) replicating the Excel Cash Flow model. IRR uses annual discounting via root-finding; verified within ±2 bps of Excel XIRR. Full model audit: 138 checks, 0 errors. Heat rates on HHV basis (LHV × 1.108).
Sources. Turbine: GE Vernova GEA35768 (05/2025). CapEx: GridLab/APPA (2025). CPV fleet: OPC Energy (Sep 2020). CEO quotes: ExxonMobil Q3 2025 and Q4 2024 earnings calls.
Conflicts of Interest. The Sheephill Group and its principals may hold positions in securities discussed, including ExxonMobil (XOM).
By receiving this presentation, all recipients acknowledge that this document is provided for informational purposes only.