Baker Hughes Wins 1.1GW AI Power Order

Baker Hughes will supply 43 gas turbines and a set of dry coolers to an unnamed “major U.S. technology company”, giving about 1.1 gigawatts of power for an AI computing development. The energy technology group announced the order on 27 October but did not name the customer, the site or the value of the contract.

According to the company’s release, the order covers 34 Frame 5 gas turbines, nine NovaLT16 turbines, BRUSH generators and associated gearing. The power will support “AI model training, inference, and physical AI workloads”. Baker Hughes said on X that it had “secured awards” to provide “integrated power generation and cooling solutions to scale high-density AI compute”.

Turbines and coolers in one package

The cooling side comes from Chart Industries, which Baker Hughes bought in July. Chart will supply dry coolers for a closed-loop cooling system that Baker Hughes says is designed for high-density AI workloads. Baker Hughes booked the power order in the third quarter and the cooler order in the fourth.

That pairing is the point of the deal for Baker Hughes. Chief executive Lorenzo Simonelli said: “The AI revolution runs on energy, and Baker Hughes is building energy systems that will underpin the next era of industrial advancement.” The company describes itself as combining power generation, cooling, grid stability and energy management for AI customers.

A second AI order in the same quarter

The order also appears in Baker Hughes’ third-quarter results, published 45 minutes later, which describe it as “43 NovaLT and Frame 5 gas turbine packages”. The same results list a separate order of about 500 megawatts: 18 Frame 5 gas turbine generator packages for data centre projects of an unnamed “U.S. energy company”.

Its Industrial & Energy Technology (IET) division booked $5.9bn of orders in the quarter, up 42% on a year earlier and what the company called its second-largest order quarter. Group orders came to $10.8bn, and IET’s order backlog hit a record $39.4bn. Simonelli said demand “across data centers, space, LNG and gas infrastructure” supported raising the division’s full-year order guidance and its order outlook for 2026 to 2028 to “$50 billion plus”.

Why gas turbines are filling the AI power gap

The order shows how quickly AI builders are turning to gas turbines on their own terms, rather than waiting for the grid. Baker Hughes has not said whether this plant will feed a grid connection or run on its own. But the scale, 1.1GW from 43 units, is the size of a large power station dedicated to a single computing project.

It follows a run of similar deals. Earlier this month Venture Global signed frame agreements with GE Vernova and Baker Hughes for up to 16 gas-fired power islands. Crusoe raised $700m secured on gas generators to run its AI factories off-grid. Mitsubishi Heavy Industries and Ingeteam signed a US memorandum on data centre power. Behind them sit forecasts such as SemiAnalysis’s estimate that about 50GW of data centre capacity could be built in 2027 alone.

The fact that the customer is a tech company matters. Big tech firms have mostly bought electricity from utilities; this one is ordering the generating equipment itself. Industrial turbines such as the Frame 5 and NovaLT16 can be installed in blocks and added over time, which suits computing campuses that expand in phases. The cost is fuel and emissions exposure that a grid contract would have passed to a utility. Local resistance is also growing: Memphis recently voted for a 135-day pause on new data centre applications.

What Baker Hughes has not said

The release gives no delivery dates, no location and no price. It also leaves open whether the customer will own and operate the plant or use a developer. Those details will decide how quickly the 1.1GW reaches real chips. For now, Baker Hughes has a booked order and an IET backlog that, by its own count, is at a record.