AI developers seek used turbines amid six-year waits | Magica
AI power rush heats up used-turbine market as US delivery waits reach six years
Editorial Team
••📖6 min read
Data-centre developers are turning to second-hand turbines, refurbished machines and aeroderivative generators to accelerate power supplies. Lower equipment prices and shorter waits come with maintenance risks, scarce inventory and permitting obstacles.
New gas turbines take six years on average to arrive in the US, according to energy consultancy Wood Mackenzie.
Used and refurbished machines offer another route to power, with lower quoted equipment prices but uncertain maintenance and service risks.
AI infrastructure developer Crusoe has ordered 650 megawatts (MW) of aeroderivative generating equipment; that order does not establish that it bought second-hand turbines.
AI data-centre developers and utilities are searching for used gas turbines, restarting mothballed machines and turning retired jet engines into generators as US delivery waits for new turbines average six years. The search can advance individual projects, but buyers are competing for a limited stock of machines.
There is no reliable estimate of the secondary market's contribution to a global turbine market worth about $25 billion annually, the report says.
Jefferies estimates global gas turbine manufacturing capacity at around 87 GW in 2028, compared with expected demand of 129 GW. Source: Reuters via Euronext.
A manufacturing queue becomes a development constraint
Aurora Tenorio, a senior supply-chain analyst at Wood Mackenzie, said US delivery discussions were focused on 2030–31, while putting the average delivery time at six years. Those market estimates do not establish a delivery date for any particular project.
Gas turbines generate electricity by burning fuel, and their customer base has changed. A market that until a few years ago served almost exclusively traditional utilities now counts data centres and their operators among its major customers, according to the report.
Wood Mackenzie put global turbine orders at 110 gigawatts (GW) at the end of 2025 against annual manufacturing capacity of 60–70 GW in its April 2026 analysis. The November reporting still places worldwide orders above 100 GW. Order volumes measure demand for equipment, not power plants already operating.
NextDC is prioritising community support and contracted customers after Melbourne power delays forced a spending retreat. The risks extend to customer rights and access to debt, even as rivals take a 9.9% stake.
The cost exposure extends beyond the machine's purchase price. Wood Mackenzie estimates that turbines account for 20–30% of combined-cycle power-project costs, and a larger share for simple-cycle plants. Its assessment is that procurement constraints increasingly determine project viability alongside fuel economics.
The three dominant turbine suppliers—GE Vernova, Siemens Energy and Mitsubishi Heavy Industries—say AI-related orders have exceeded their production capacity. They are expanding, but adding factories and shifts does not immediately remove specialised production constraints.
Wood Mackenzie's April account described GE Vernova investing more than $160 million to increase annual large-frame turbine production from about 50 units to 70–80 by late 2026. Siemens Energy had moved key facilities to round-the-clock operations and announced a $1 billion US investment programme; Mitsubishi Heavy Industries planned to double manufacturing capacity through 2028. These were expansion plans, not evidence that the additional output had already arrived.
Wood Mackenzie identified specialised labour shortages, trade-related costs and hot-section components, particularly single-crystal blades made at scale by only a handful of suppliers, as constraints on throughput. November estimates from investment bank Jefferies suggest the imbalance could persist: roughly 87 GW of global manufacturing capacity in 2028 against expected turbine demand of 129 GW. For US data centres alone, the bank estimates nearly 17 GW of turbine demand in 2026, potentially more than tripling by 2028. These are forecasts of equipment demand, rather than completed power plants.
Lower purchase prices carry different risks
The immediate appeal of used equipment is both availability and price. The November equipment-price indications show why buyers seek alternatives:
Equipment category
Reported price per MW
Basis
Well-maintained used units
$1.5 million–$1.7 million
Origin Partners, which buys equipment for data-centre developers
Refurbished units
About $1.3 million
Origin Partners
New turbines
$2 million–$2.5 million
Aurora Energy Research
These are equipment-market indications from different sources, not matched bids for identical machines or estimates of a complete power plant's lifetime cost.
Prices can vary much more widely for smaller units. ReflowX, a marketplace connecting turbine owners and buyers, described a recent sale of two 14.5-MW machines made by Solar Turbines, a Caterpillar subsidiary, for $12.5 million. Dividing that price by their combined 29-MW capacity gives about $430,000 per MW. Its general manager said new equivalents would cost about $35 million and take five years to deliver. The buyer was not disclosed.
Lower purchase prices do not settle operating economics. Tenorio warned that overhauled turbines can have uncertain maintenance histories, limited remaining life and weak long-term servicing. Converting jet engines also raises questions about their suitability for continuous electricity generation.
Nor are all cheaper machines convenient substitutes for larger ones. Equipment supplier SVI Bremco said developers sometimes install more than 100 smaller turbines at one site, mixing brands and models when larger equipment is unavailable. That is a different operating configuration, not simply a cheaper version of the same purchase.
PROENERGY’s PE6000 aeroderivative engine. The company announced 13 generator sets for Crusoe; the announcement does not establish that they are second-hand. Image: PROENERGY. Source: PROENERGY via PR Newswire.
Crusoe's order shows a broader equipment alternative
Crusoe, an AI infrastructure provider that develops data-centre projects, has prior experience locating computing near energy supplies. In its account of its earlier business, the company describes turning oilfield gas that would otherwise be flared into electricity for modular data centres at the oilfield.
On April 16, 2026, PROENERGY, a US manufacturer and supplier of fast-start generating equipment and plant services, announced a contract for 13 PE6000 aeroderivative generator sets for Crusoe, each capable of producing 50 MW. Their combined capacity is 650 MW.
Aeroderivative turbines use aircraft-engine technology. The November report describes the units as based on GE Aerospace jet engines; PROENERGY's announcement calls the Crusoe purchase a manufactured equipment solution and says its equipment is made in-house. Neither source establishes that these particular units are second-hand or converted retired engines. The order illustrates the wider search for alternatives to the conventional turbine queue.
Crusoe said the equipment would support upcoming hyperscale data-centre projects, with speed central to its purchasing rationale. PROENERGY said it would test-fit and quality-check the equipment at its Sedalia, Missouri, manufacturing facility before shipment. The contract price was not disclosed, and the companies did not provide additional information for the November report.
Gas engines are another substitute. Finland's Wärtsilä said in the November report that it had booked about 3 GW of orders from US data centres since spring 2025 for generating units that use gas engines rather than turbines. The competition to supply power therefore extends beyond finding and refurbishing old turbine inventory.
The next test is dependable, permitted power
The consequences also reach electricity customers outside the AI industry. The Tennessee Valley Authority, a federally owned US utility, said prices for new, used and refurbished equipment had risen significantly over the previous two years. The National Rural Electric Cooperative Association, representing customer-owned nonprofit utilities, said members faced unprecedented lead times and rising prices, with some turning to refurbished turbines.
For developers, the unresolved question is how much earlier an available machine can produce dependable, permitted electricity at a particular site. Delivery and installation dates, remaining operating life, service support, air permits and grid access will determine whether a purchase actually advances an AI facility's opening.
Meta is expanding an existing system that places suspected teenagers into Instagram’s protective settings despite adult birthdays. Its global rollout follows legal settlements, but the announcement provides no numerical accuracy results.
Developer Iain Kay’s Linux script turns available RAM into parody emails after Sam Altman’s remarks about local AI. Its own documentation preserves Altman’s clarification—and explains why its measurements cannot test his utilization claim.
Oppenheimer upgraded AMD to Outperform with a $600 target as it forecasts 135% data-center growth in 2027. The bullish call faces tests in supply, customer infrastructure and cash generation.
Statnett has paused new data center grid reservations pending technical safety rules, while the government plans licensing for existing and future facilities. The proposals would determine both connection requirements and which computing uses receive priority.
Google’s TPU 8tx targets confidential AI workloads, while TPU 8i Max combines two chips for larger contexts. Both reuse existing silicon, putting the competitive test on system design, software and delivered cost.
Reported app code points to a paid Galaxy AI tier with higher usage limits and new features. Samsung says core tools will stay free, while prices, allowances and a commercial launch remain unconfirmed.
Citi cut US equities and technology to neutral and upgraded Europe outside the UK, citing lower exposure to AI turbulence and benefits from falling oil prices while retaining a positive market outlook.
ByteDance is rolling out Seedream 5.5 with a 100-edit demonstration, native transparent output and a lower reported editing price. Its predecessor and Google already offered parts of that workflow; production reliability remains unproven.
DeepSeek says its open-source agent framework reached 1.54 million daily active users through its official API. Its model-agnostic framework is gaining plugins, but DeepSeek’s own terms caution against treating it as ready for production.
Animoca Brands co-founder Yat Siu will become CEO on December 1 while remaining executive chairman. His AI agent strategy comes alongside a target to clear overdue financial reporting during 2027, with no relisting date announced.
Amazon’s Fauna Robotics hiring spans manipulation, companionship and cloud operations. Sprout brings a developer platform and research customers, but Amazon has yet to disclose a household product, price or launch date.
Linux creator Linus Torvalds says eBPF fixes may now be arriving faster than new reports, offering a limited sign of relief from AI-driven maintenance work. Most other kernel areas probably have not reached that point, and the release note supplies no counts to establish a wider turnaround.
Dialpad and Ada plan to combine a communications network with customer-service AI agents. Reported terms suggest a sharp reset from Ada’s 2021 valuation, while the product test is whether the merger improves on automation and human handoffs Dialpad already advertised.
Nvidia plans to back three Korean AI startups spanning enterprise software, language models and robotics. Its investment amounts remain undisclosed, limiting what the announcement reveals about the financing it will provide.
Independent directors and outside counsel will investigate reports involving Nvidia’s founder and CEO, who has admitted infidelity. The board says it has no reason to believe national security or company secrets were compromised.
Apple reportedly wants Siri to pursue goals without prompts at every step. A possible 2027 debut is a reporter’s assessment, while Apple experiments with its iOS 28 development process after earlier Siri delays.
Apple led Nvidia by roughly $615 billion in market value at the November 5 close. Nvidia’s rapid sales growth contrasts with financing and construction risks for AI infrastructure, while Apple faces rising costs and uses Nvidia chips in its own cloud.
Councils, transport officials and MPs’ staff report more demanding AI-assisted correspondence. Official records show pressure alongside successful appeals, while regulators seek to protect access and contain excessive submissions.
Anthropic has reportedly taken over commitments for Caterpillar generators ordered by OnePWR, gaining access to scarce equipment and installation capabilities as it seeks greater control over data-center power.
