Gas turbines have become the tightest bottleneck in the AI power chain. In the US, a new gas turbine now takes six years on average to deliver. Worldwide orders top 100 GW, while global production capacity is estimated at 60 GW to 70 GW, according to Reuters. That gap has sent data center developers looking for used turbines, mothballed machines and converted jet engines. It has also led power producers to buy entire operating gas plants.

The turbine manufacturer at the choke point and four merchant generators are the focus here. All five bring in revenue and generate positive cash flow, but their cash flows work differently. GE Vernova earns from an equipment and services backlog. The four independent power producers (IPPs) earn from wholesale power prices, capacity auctions and hedges, and none has a regulated utility’s rate base to guarantee its returns.

GE Vernova: Turbine Delivery Slots Now Sold Into 2032

GE Vernova (NYSE:GEV) is the supplier everyone in this story is waiting on. In the third quarter it signed 11 gigawatts of gas orders and slot reservations, which raised the total under contract from 116 to 122 gigawatts. It expects to reach at least 130 gigawatts by the end of 2026, up from 83 gigawatts at the start of the year. The company has already sold delivery slots for 2032. Chief executive Scott Strazik said aeroderivative capacity is “sold out so far”. Annual gas turbine output is set to rise from 20 GW in Q3 2026 to 24 GW in 2028 and 30 GW by 2030, which still leaves the order book ahead of the factories.

Third-quarter revenue was $12.64 billion, up 26.8% year over year. GAAP diluted EPS of $4.81 beat the $4.01 consensus. Total backlog reached $197 billion, and management raised 2026 free cash flow guidance to $16 billion to $17 billion from $11.5 billion to $12.5 billion.

Bull case: Scarcity gives GE Vernova pricing power. Management expects 2026 gas orders to be priced about 50% higher per kilowatt than 2025 orders. Power EBITDA margin reached 21.4%, up 830 basis points. Each gigawatt of its large HA-class turbines is expected to produce $0.5 billion of high-margin services revenue over the first 20 years. Data center customers make up about 20% of its roughly 100 gas power customers.

Risk: The Wind segment is guided to $400 million to $500 million of EBITDA losses for 2026. The stock is also sensitive to politics: Barron’s linked recent pressure on GE Vernova and other industrial stocks to a Democratic sweep in the elections. Shares traded at $829.12 on Monday morning, up 7.11% on the session. They are down 16.8% over one month and up 27.16% year to date.

Vistra: Buying Running Gas Plants Instead of Waiting for Turbines

The clearest example of the secondary market working at the scale of whole plants is Vistra (NYSE:VST). It closed the Cogentrix acquisition on October 30, 2026, adding about 5,500 megawatts of mostly combined-cycle gas generation. The net price was about $4 billion, or about $729 per kilowatt. For comparison, Reuters reports that new turbines cost about $2 million to $2.5 million per MW for the equipment, and buyers still face the delivery queue. Vistra has now closed over 12 gigawatts of acquisitions at less than 7x forward EBITDA on average. Its fleet totals about 50,000 megawatts.

Third-quarter revenue rose 27.7% to $6.35 billion, and adjusted EBITDA came in at $2.378 billion. Management guided 2027 adjusted EBITDA to $7.7 billion to $8.5 billion. Vistra has also contracted some of its existing gas plants to data centers. It signed a 20-year, 207 MW power purchase agreement with New Era Energy and Digital at its Odessa combined-cycle plant in Texas.

Bull case: Chief executive James Burke called flexible gas “the cheapest and quickest source of incremental generation.” Vistra can contract plants it already owns at little added capital, and it has nuclear capacity at Beaver Valley and part of Comanche Peak still uncontracted.

Risk: ERCOT (the Texas grid) forward prices have been trending lower. Vistra is 78% hedged for 2028 overall but much less hedged in ERCOT, so Texas pricing is its largest open exposure. Shares traded at $133.49, up 6.96% Monday and down 16.88% year to date.

NRG Energy: Selling Older Gas Plants at a Premium While Keeping Its Turbine Slots

NRG Energy (NYSE:NRG) is on the selling side of the secondary market. It signed a definitive agreement to sell just under 4.9 gigawatts of gas-fired plants in New York, Texas and New England to LS Power for $4 billion. That price equals eight times expected 2027 adjusted EBITDA. The deal implies a value of about $1,400 per kilowatt for the Jack County facility. NRG says it is developing 1.5 gigawatts of new Texas gas generation for just above $1,000 per kilowatt. In other words, existing megawatts are selling for more per kilowatt than NRG expects to pay to build new ones.

Third-quarter adjusted EPS of $4.68 beat the $3.03 consensus, and revenue rose 24.0% to $9.47 billion. NRG plans $3 billion of total repurchases across 2026 and 2027 and $2 billion of debt reduction when the sale closes. The 2027 dividend is planned to rise to $2.05 from $1.90 annualized.

Bull case: NRG holds turbine allocations under a development agreement with GE Vernova and Kiewit, and management says its project pipeline “exceeds the turbine allocations and EPC capacity” under that agreement. Management also said it will hold the reserved turbines rather than resell them: “We’re looking to put them to work for long-term contracts.” Its 1.2-gigawatt Texas data center project with a hyperscaler has approved agreements, but the company has not formally announced it yet; management expects an announcement by year end. The project is excluded from NRG’s long-term financial framework, so it would add to that plan.

Risk: The LS Power sale still needs HSR, FERC and New York PSC approvals, and liquidity fell to about $6.0 billion from about $9.6 billion. Shares traded at $97.01, up 9.88% Monday after the earnings report but down 38.18% year to date.

Talen Energy: Scarce PJM Capacity It Already Owns

Buying existing plants also figured in the expansion at Talen Energy (NASDAQ:TLN). It closed the Cornerstone acquisition on June 15, 2026, adding about 2.6 GW of gas generation at Waterford, Darby and Lawrenceburg. It then cleared more than 10 GW in PJM’s 2028/2029 capacity auction at $325.00/MWd, a price that reflects scarcity on the largest US grid. Chief executive Mac McFarland summed it up on the November 3 call: “The demand is here, the supply isn’t, and we already own the assets the market needs.”

Revenue rose 72.3% to $1.33 billion, and total generation reached 23.9 TWh versus 11.1 TWh a year earlier. GAAP EPS of $2.88 missed the $7.46 consensus. The miss came from $190 million of unrealized derivative valuation losses and $148 million of acquisition-related interest. Adjusted EBITDA was $669 million, and 2027 guidance calls for $2.50 billion to $2.85 billion.

Bull case: McFarland put the all-in cost of new combined-cycle power north of $120 per MWh, which makes Talen’s existing fleet a cheaper option for buyers. It also holds 4 GW of powered-land sites that could connect to the grid in 2028 to 2029. Talen has not disclosed any new signed data center deal beyond its existing campus contract. McFarland’s line on future deals, “They’re coming. Trust me,” is a statement of intent that has yet to be backed by a signed contract.

Risk: Leverage. Long-term debt rose to $9.54 billion from $6.78 billion at year-end 2025. Talen has also already sold about $1.5 billion of its future PJM capacity payments, and 2028 output is only about 35% hedged. Shares traded at $359.45, up 6.15% Monday and down 4.11% year to date.

Constellation Energy: Nuclear Contracts Priced Below the Cost of New Gas

The largest US nuclear fleet is paired with Calpine’s gas plants at Constellation Energy (NASDAQ:CEG), and those gas plants generated 38,387 GWh in the third quarter. It is also buying gas capacity secondhand: it agreed to acquire the 609 MW Rhode Island State Energy Center for $715 million. Management estimates a new combined-cycle plant costs $110 to $130 per megawatt hour, assuming $3,000 per KW in capital costs. It uses that range as the benchmark when pricing nuclear contracts.

Revenue rose 32.9% to $9.55 billion, and adjusted EPS of $5.06 beat the $3.88 consensus. The nuclear fleet ran at a 97.2% capacity factor. Constellation has signed about 8,500 MW of long-term agreements so far this year. They include a 20-year, 690 MW hyperscaler PPA backing more than $3 billion of investment at Calvert Cliffs and a 15-year, 2,700 MW load-following agreement. Management raised and narrowed 2026 adjusted operating earnings guidance to $12.50 to $12.80 per share.

On the bull case side, nuclear uprates are expected to add nearly 1,200 megawatts, mostly between 2029 and 2032, with no new turbine order required. Chief executive Joe Dominguez also pointed to unused capacity on existing plants: “You have a lot of plants just sitting idle on the grid.”

Risk: Rules for large-load interconnection have not been finalized, and Constellation does not expect regulators to act on co-location until the first half of 2027. Shares traded at $270.06, up 8.05% Monday after the earnings report and down 23.21% year to date.

Where the Turbine Shortage Leaves These Five Stocks

The turbine shortage gives pricing power to whoever owns working megawatts or the factory slots to build new ones. GE Vernova benefits from both its order book and decades of services revenue. Vistra, NRG, Talen and Constellation are making money on operating plants because new turbines will not arrive before the 2030 to 2031 delivery window. All of that expansion still has to be powered, cooled and connected by someone, which is why we pulled together seven suppliers behind the AI data-center boom in a free report you can grab here. Three things to track next: GE Vernova’s fourth-quarter slot reservations, NRG’s year-end data center announcement and the December PJM capacity auction, which NRG and Talen both expect to clear at or near the top of its allowed range.

Disclaimer:This article represents the opinion of the author only. It does not represent the opinion of Webull, nor should it be viewed as an indication that Webull either agrees with or confirms the truthfulness or accuracy of the information. It should not be considered as investment advice from Webull or anyone else, nor should it be used as the basis of any investment decision.