Energy · Deep dive

Calpine Corporation

The largest independent power producer in the United States — a ~27 GW fleet of natural-gas combined-cycle plants, the world's biggest geothermal complex at The Geysers, and a retail electricity arm — that went bankrupt in 2005, was taken private by Energy Capital Partners in 2018, and was bought by Constellation Energy in a ~$26.6B deal that closed January 2026, handing ECP one of the most profitable private-equity exits in history.

well positioned

Calpine owns the largest fleet of flexible, dispatchable gas-and-geothermal generation in the US just as AI data-center load reverses a decade of bearishness on gas — the very reason its EBITDA doubled under private ownership, ECP cleared a ~4x return, and Constellation paid ~$26.6B to own it.

My take ★★ Very interesting ★ Interesting − Maybe ✗ Not interesting

Snapshot

Calpine is the largest independent power producer in the United States: a fleet of about 79 plants and more than 27 GW of capacity, overwhelmingly natural-gas combined-cycle turbines, plus The Geysers in California — the world’s largest geothermal complex — and a retail electricity arm serving roughly 60 terawatt-hours of load a year. It is a company that has died and been reborn: founded in 1984, it overbuilt on cheap debt, filed one of the largest industrial bankruptcies in US history in December 2005, and emerged in 2008. In March 2018 a consortium led by Energy Capital Partners took it private for about $17 billion including debt on a contrarian bet that flexible gas would stay valuable even as renewables scaled. That bet paid off spectacularly: as AI data centers reset US power demand, Calpine’s EBITDA roughly doubled, and in January 2026 Constellation Energy closed its ~$26.6 billion acquisition of the company, handing ECP a ~4x return and the title, in some accounts, of the most profitable private-equity deal in history by dollar value.

Founding story

Peter Cartwright started Calpine in 1984 in San Jose with about $1 million and an unfashionable idea: independent companies, not regulated utilities, should build and own power plants and sell into wholesale markets. Backed by Guy F. Atkinson Construction and Switzerland’s Electrowatt, and joined by Gibbs & Hill colleagues Ann Curtis and John Rocchio, he specialized in efficient combined-cycle gas turbines — plants that capture exhaust heat to spin a second turbine.

Cartwright took Calpine public in 1996 in the largest IPO ever for an independent energy company, then ordered 46 Siemens-Westinghouse turbines before he had contracts to run them — a bet that deregulation would create insatiable merchant demand. It made Calpine the biggest US gas generator, until the 2000-01 California energy crisis and overbuilding collapsed wholesale prices and the debt came due. In December 2005 Calpine filed for Chapter 11 with roughly $22 billion of debt — one of the largest industrial bankruptcies on record. It emerged in January 2008 with a $7.3 billion exit financing and a leaner ~24,000 MW fleet, relisted, and spent the next decade deleveraging and building retail under Thad Hill, CEO from 2014.

The ownership story that matters now begins in 2017-2018. Energy Capital Partners, with CPP Investments and Access Industries, took Calpine private for $15.25 a share — about $5.6 billion of equity and ~$17 billion including debt — closing in March 2018. The consensus view at the time was that renewables would strand gas plants; ECP bet the opposite, that dispatchable generation would stay essential to a grid adding intermittent wind and solar. Over the hold Calpine’s EBITDA roughly doubled, ECP took dividends along the way, and the AI data-center boom turned a contrarian bet into a landmark exit.

How it works

Calpine’s business is physical and unglamorous: it owns power plants and sells the electricity, and increasingly the reliability, they produce. The core asset is a fleet of natural-gas combined-cycle plants — the most efficient form of gas generation — across the three big competitive US markets: ERCOT (Texas), PJM (mid-Atlantic/Midwest), and CAISO (California). These plants are dispatchable, ramping up in minutes when demand spikes or wind and solar fade — exactly what a grid full of intermittent renewables needs.

Money is made three ways. Merchant energy sales: Calpine sells power at wholesale prices, earning the spark spread — the gap between the electricity price and the cost of the gas burned to make it — which widens when demand is high and reserves tight. Capacity payments: markets like PJM pay generators simply to be available in future peaks, a stream that has repriced sharply higher as capacity tightened. And the retail hedge: owning generation lets Calpine match its fixed-price retail obligations against physical supply, smoothing the volatility that sank the old Calpine.

The Geysers is the exception and the jewel: 13 plants across a 45-square-mile field in Sonoma County, California, producing ~725 MW of always-on, near-zero-carbon baseload — the largest geothermal operation on earth, with a 38 MW battery added in 2024 and a 25 MW expansion completed in 2026.

Product and business overview

Calpine sells electricity and reliability through a few named components. Merchant generation is the bulk: ~27 GW of gas combined-cycle plus geothermal, battery storage, and some solar. The Geysers is the flagship renewable-baseload asset. On the customer side, Champion Energy Services is the residential/commercial retailer in deregulated markets and Calpine Energy Solutions serves commercial, industrial, and governmental load; together they serve ~60 TWh annually, and Calpine increasingly markets firm power directly to data-center buyers. The logic is a barbell: dispatchable gas earning on scarcity, geothermal on carbon-free baseload, and retail as a hedge and channel.

Business model and pricing

There is no price list; Calpine earns on wholesale clearing prices, capacity auctions, and negotiated retail and bilateral contracts, driven by three levers: spark spreads, capacity prices, and retail load-following margins. As a private company it published little, but the deal math is clear: Constellation paid ~7.9x 2026 EV/EBITDA on a ~$26.6B EV, implying ~$3.3-3.4B of forward EBITDA, and guided $2B of added annual free cash flow. ECP’s framing — that EBITDA doubled over the 2018-2025 hold — tells the story: a fleet priced for decline instead saw cash flows expand as demand inflected. The leverage is heavy by design ($12.7B net debt), magnifying both the upside from rising power prices and the downside from any reversal.

Traction over time

The arc is a business that grew, over-levered, collapsed, rebuilt, and rode a demand super-cycle into a record exit. The key number is the doubling of EBITDA over 2018-2025 against a thesis that said it should have shrunk.

Market analysis

The structural force is the sharpest demand inflection US power has seen in a generation. After ~two decades of flat demand, AI data centers are driving load up fast: Goldman Sachs projected US data-center demand more than doubling from ~31 GW in 2025 to ~66 GW in 2027 (Goldman Sachs, 2025), and Bank of America estimated the US needs 230 GW-plus of new capacity over five years while regulated utilities add only ~93 GW — a 100 GW-plus gap merchant generators fill at market prices (BofA / Utility Dive, 2025). Gas supplied ~40% of US electricity in 2025, and the EIA warned data-center growth could push gas generation up 7%+ over 2025-2027 in a high-demand case (EIA, 2025). For a 27 GW dispatchable-gas-plus-geothermal fleet, this is nearly ideal: tightening reserve margins widen spark spreads, lift capacity prices, and put a scarcity premium on firm power. The offsetting force is long-run decarbonization, but near-term, gas is the swing supply the AI build depends on.

Competitive intel

The competitive set is a short list of scaled merchant generators, most AI-power beneficiaries too. Constellation was the largest competitive generator and is now the parent — the deal created a 60 GW producer pairing the biggest US nuclear fleet with the biggest gas-and-geothermal one. Vistra ($55.7B market cap, 2025-2026) is the closest independent peer: a Texas-weighted gas fleet plus nuclear and the TXU retail book, competing directly in ERCOT and PJM and chasing the same dispatchable capacity (~2.6 GW of gas for $1.9B in 2025). NRG Energy ($3.7-4.0B projected 2025 EBITDA) overlaps most in retail and Texas gas. Talen Energy is a smaller Pennsylvania nuclear-plus-gas operator whose Amazon co-location deal pioneered behind-the-meter supply. Regulated utilities — NextEra, PSEG, Southern — are the structural alternative but build far too slowly to close the gap alone. Calpine’s edge is scale and flexibility in exactly the fuel and markets where demand is landing; its exposure is that every peer is racing for the same firm capacity, compressing future build returns.

History and evolution

• 1984 — Peter Cartwright founds Calpine in San Jose with ~$1M and a merchant-power thesis.
• September 1996 — IPO, the largest ever for an independent energy company; orders 46 gas turbines.
• 2000-2005 — Rapid debt-funded build-out; the California energy crisis and overbuilding collapse wholesale prices.
• December 21, 2005 — Files Chapter 11 with ~$22B of debt, one of the largest industrial bankruptcies in US history.
• January 31, 2008 — Emerges from bankruptcy with ~24,000 MW and a $7.3B exit financing; relists on NYSE.
• 2014 — Thad Hill becomes CEO; builds out retail (Champion Energy, Calpine Energy Solutions).
• March 2018 — Energy Capital Partners, CPP Investments and Access Industries take Calpine private for $15.25/share (~$17B incl. debt).
• 2018-2024 — Deleveraging, dividend recaps, and EBITDA roughly doubling as power demand inflects.
• January 10, 2025 — Constellation Energy agrees to acquire Calpine for ~$16.4B equity / ~$26.6B incl. debt.
• July 2025 — FERC and DOJ clear the deal conditioned on PJM generation divestitures.
• January 7, 2026 — Acquisition closes; Calpine becomes a Constellation business unit; ECP realizes ~4x.
• March 2026 — Constellation agrees to sell PJM plants to LS Power to satisfy the FERC/DOJ remedy.

What people say

The case for. ECP and the deal’s backers frame Calpine as a contrarian thesis vindicated: flexible gas, written off as stranded in 2018, became the backbone of an AI demand surge, doubling EBITDA and delivering what several outlets called the largest-dollar PE gain ever. Constellation paid ~$26.6B and guided ~20% accretion to 2026 adjusted EPS and ~$2B of added free cash flow — a confident endorsement of the fleet’s cash generation. Employees are notably positive for a heavy-industry operator: Glassdoor ~4.0/5 across ~445 reviews, ~91% CEO approval, praising a flat culture and work-life balance. The Geysers is genuinely differentiated — the world’s largest geothermal complex, carbon-free and always-on.

The complaints. The business rests on burning gas, the central bear case: Calpine is one of the larger CO2-emitting generator fleets in the country, and its value depends on gas staying economically and politically viable through a decarbonizing decade. The capital structure is aggressive — ~$12.7B net debt and a PE history of dividend recaps — and its debt has long carried sub-investment-grade ratings (Moody’s rated recent secured term debt around B1, senior notes in the Ba2/B2 range). Merchant markets are volatile: the spark-spread leverage that doubled EBITDA on the way up can compress it fast on the way down, and the 2005 bankruptcy is the cautionary tale. The Geysers carries a decades-long record of hydrogen-sulfide (“rotten egg”) emissions and abatement-chemical exposure, and gas plants draw routine community and environmental-group opposition (Union of Concerned Scientists; Global Energy Monitor). The merger itself required FERC/DOJ concessions — divesting PJM plants to LS Power — and the demand surge that made the fleet valuable is pulling every peer toward new gas build, eroding scarcity returns over time.

Outlook: well positioned or at risk?

Well-positioned. Calpine owns the largest fleet of flexible, dispatchable generation in the US at the exact moment the grid is short of it. The bearish 2018 thesis — that renewables would strand gas — inverted: AI data centers reset demand, reserve margins tightened, spark spreads and capacity prices rose, and EBITDA roughly doubled over the private hold. That is not a forecast but a realized result, ratified by ECP’s ~4x exit and Constellation paying ~$26.6B to fold the fleet into a ~60 GW producer. The Geysers adds a scarce carbon-free baseload asset data-center buyers covet, and retail hedges the volatility that killed the old Calpine.

The honest risks are real but do not break the thesis on today’s evidence. The carbon-transition question is genuine — a gas-heavy fleet must stay viable through decarbonization — but gas is the swing supply the AI build depends on, and dispatchability is worth more, not less, as intermittent renewables grow. The leverage is heavy and the debt sub-investment-grade, so a sharp price reversal would bite; but the demand backdrop and Constellation’s investment-grade balance sheet now behind it cut that risk materially. Merchant volatility is structural, and competition for new firm capacity will compress build returns. Against the base case — the largest dispatchable fleet in the country, doubled cash flows, and a strategic acquirer that just paid a premium — the position compounds. The debate is how much of the AI-load premium is already priced, not whether Calpine owns the assets the grid needs.

How a challenger would attack it

Attack the scarcity premium, not the fleet. Calpine’s economics rest on tight reserve margins: spark spreads and capacity payments that repriced upward because dispatchable supply is scarce. A challenger doesn’t out-operate a 27 GW combined-cycle fleet — it collapses the scarcity that fleet monetizes. The fastest vector is speed-to-power for the marginal buyer: on-site generation (fuel cells, modular gas with pre-permitted sites, batteries paired with solar) sold directly to data centers lets hyperscalers bypass the wholesale markets where Calpine earns its widest margins, the way Talen’s Amazon co-location deal already showed load can be pulled behind the meter. The second vector is the balance sheet: ~$12.7B of assumed net debt and sub-investment-grade paper mean Calpine’s parent must defend cash flow, while a new entrant with cheap capital can build the next tranche of firm capacity at returns that compress everyone’s scarcity rents — Vistra’s ~$1.9B for 2.6 GW shows the land grab is on. Third, attack the carbon flank: Calpine is among the larger CO2-emitting fleets in the country, and every corporate buyer with a clean-energy mandate is a customer a geothermal, nuclear-SMR, or long-duration-storage challenger can peel off — including at The Geysers’ own game, where next-generation enhanced geothermal developers can offer the same always-on carbon-free product without a 45-year-old field’s hydrogen-sulfide baggage.

Same playbook, new buyer

The Calpine playbook — own dispatchable generation in deregulated markets, pair it with retail load as a hedge, and monetize scarcity — has been run almost entirely inside ERCOT, PJM, and CAISO for utility-scale and commercial buyers. Two shifts are open. First, the buyer shift: package firm power as a direct product for mid-size data-center and industrial customers — 10-100 MW loads too small for the bilateral mega-deals Constellation now chases with hyperscalers, but collectively enormous as AI inference pushes compute into secondary markets. A developer aggregating those buyers with dedicated gas-plus-battery microgrids replicates Calpine’s generation-plus-retail hedge at a scale the ~60 GW parent won’t bother structuring. Second, the geography shift: the same merchant thesis in power-short international markets — data-center build-outs in Southeast Asia, the Gulf, and Latin America face worse interconnection queues than PJM’s, and no incumbent runs Calpine’s flexible-fleet-plus-retail model there. Constellation cannot follow either path: its capital and integration bandwidth are consumed digesting the largest deal in its history plus the LS Power divestitures, its investment-grade rating disciplines new-market risk, and its strategic center of gravity is US nuclear — the merchant frontier is someone else’s to take.

Sources and further reading

• Constellation to Acquire Calpine; Creates America’s Leading Producer of Clean and Reliable Energy — Constellation Energy, January 2025. Deal terms: ~$16.4B equity, ~$26.6B net, 7.9x 2026 EV/EBITDA, ~$2B added FCF, ~60 GW combined.
• Constellation’s $16.4B purchase of Calpine would create largest US power generator — Utility Dive, January 2025. Structure, capacity, and regulatory approvals required.
• Constellation Completes Calpine Transaction — Constellation Energy, January 2026. Close on January 7, 2026; largest US power producer.
• Constellation Announces Agreement to Sell PJM Generation Assets to LS Power as Part of FERC, U.S. DOJ Resolution — Constellation Energy, March 2026. Regulatory divestiture remedy.
• Energy Capital Partners Nets 4x Return on Calpine Sale to Constellation — Transacted, 2026. ~$5.6B equity, ~$4.5B cash + ~$11.9B CEG stock, EBITDA doubled over hold.
• Consortium Led by Energy Capital Partners Completes Acquisition of Calpine — Energy Capital Partners, March 2018. Take-private close, $15.25/share, board and management.
• Calpine Makes $7.3B Exit from Chapter 11 Bankruptcy — Natural Gas Intelligence, January 2008. Bankruptcy emergence, ~24,000 MW, exit financing.
• History of Calpine Corporation — FundingUniverse. 1984 founding, Peter Cartwright, 1996 IPO, turbine order.
• US Data Center Power Demand Projected to Double by 2027 — Goldman Sachs, 2025. ~31 GW (2025) to ~66 GW (2027) data-center demand.
• Fossil generation could rise with faster-than-expected growth in data center power demand — U.S. EIA, 2025. Gas ~40% of US generation; data-center-driven gas demand growth.
• Calpine Reviews — Glassdoor, 2024-2025. ~4.0/5 across ~445 reviews, ~91% CEO approval, culture and work-life themes.
• Environmental Impacts of Geothermal Energy — Union of Concerned Scientists. Hydrogen-sulfide emissions and geothermal environmental concerns at The Geysers.