Who Is Buying Used Power Outside America: Failing Grids, Reconstruction and the 50 Hz Queue

The turbines that left Asia

The most important secondary transaction of 2026 did not happen in Texas. It happened at the Port of Pagbilao in the Philippines, where two Siemens F-class gas turbines and a steam turbine had sat in crates since a 650 MW LNG-to-power project stalled a decade ago. In June an American buyer paid $350 million for them, nearly four times what the owner had invested, and the equipment is now on its way to Indiana. Southeast Asia is short of exactly this kind of machine. Wood Mackenzie counts 53 GW of new gas capacity targeted across six Southeast Asian markets by 2030, of which only 11 GW has turbines secured and 14.9 GW is expected to actually reach commercial operation. Vietnam alone targets 29.4 GW and will get about 3.7.

Yet the Pagbilao machines went west, not into that gap. The Philippines runs at 60 hertz, so the equipment was portable to the American grid, and an American buyer with a tenant and a 2028 deadline outbid a region that needs the power at least as badly. That is the first thing to understand about the international market: the United States is now pulling 60 Hz iron out of every country that has it. The second thing is that most of the world does not run at 60 hertz, and the 50 Hz half of the market has its own buyers, its own queue and, for the moment, its own prices.

Quick takeaways

• The world splits by frequency. The Americas, Saudi Arabia, the Philippines, Korea, Taiwan and half of Japan run at 60 Hz; Europe, Africa, the Middle East, South Asia and most of Southeast Asia run at 50 Hz. Geared machines (Frame 5, 6B, SGT-800, aeroderivatives) can cross the line. Direct-drive heavy frames cannot.
• The 50 Hz queue is as long as the American one. Siemens Energy's backlog is 69 GW with lead times of three years or more; Mitsubishi quotes 2031; Chinese OEMs are the only ones offering a heavy frame in about a year.
• Latin America is the 60 Hz pressure gauge. Venezuela's thermal fleet runs at 13 percent of capacity, Cuba's deficit hit a record 2,380 MW in July 2026, Ecuador has failed nine thermal tenders, and Brazil just re-contracted 7.6 GW of existing thermal plants rather than wait for new ones.
• Powerships are the market's thermometer. Karpowership has 45 vessels and more than 8 GW deployed in twenty countries, and the places paying for floating generation are the places that will pay for relocated turbines.
• Reconstruction is a 50 Hz order book nobody can fill: Ukraine needs $71 billion of generation, Syria has signed for 4 GW, Iraq faces an 11 GW summer shortfall, and the OEM deliveries behind those plans run to 2031.
• The supply side is the mirror image: fleets idled by fuel, payments or demand that never came. Turkey has been exporting gas plants since 2015, and a 240 MW combined cycle moved from Antalya to Tashkent in two years.
• Europe's data center build on gas is real but slower, and Asia's is about to hit the wall Wood Mackenzie describes. Both will be used-equipment buyers by 2028.

Two frequencies, one queue

Every gas turbine spins a generator at a speed fixed by the grid it feeds: 3,000 rpm for 50 hertz, 3,600 for 60. For the big direct-drive frames, a 9E or a 9FA, an SGT5-4000F or a GT26, that speed is designed into the rotor and cannot be changed without rebuilding the machine. For geared units, where the turbine runs at its own speed and a gearbox steps it down, the change is made at the gearbox: Frame 5s and 6Bs, Siemens' SGT-800, Ansaldo's AE64.3A and every aeroderivative on the market are sold in both frequencies. Siemens Energy's own product page lists the SGT-800 in 50 and 60 Hz packages, and Ansaldo told POWER its AE64.3A gearbox handles either, which is how a 50 Hz Italian machine ended up with eight orders for a Texas data center. Genloop keeps a country-by-country frequency reference for exactly this question.

The consequence is that the secondary market is two markets with a narrow bridge between them. The American buyers described in the companion piece are bidding for 60 Hz frames, 60 Hz steam turbines, and anything geared. Everything else, which is most of the world's installed base, is spoken for by buyers at 50 hertz.

The 50 Hz queue is dominated by Siemens Energy, Mitsubishi and Ansaldo, and it looks like the American one. Siemens Energy's gas turbine backlog reached 69 GW in June, with 15 GW booked and 6 GW shipped in a single quarter, and its chief financial officer puts lead times at three years or more across the company. It is adding capacity, from 80 medium frames a year to about 100 by 2028 and from 35 large frames to 50 in 2027. The order mix tells you where the 50 Hz demand sits: about 40 percent United States, 35 percent Europe, with Poland and Turkey named, and the rest split between the Middle East and Asia; Reuters reported that Middle East customers and data center operators together make up about half of the company's gas turbine orders. Mitsubishi's Asia Pacific chief said in late October 2026 that a large frame ordered now arrives around 2031, with 2028 and 2029 almost full. Ansaldo's orders rose 24 percent in 2025 and it too warns that delivery times are lengthening.

The exception is Chinese. Dongfang's G50, a 50 MW F-class, shipped its first export unit to Kazakhstan in August, about nine months after contract, and is quoted at roughly thirteen months against five to eight years for a Western large frame. Shanghai Electric won its first overseas heavy-duty order in September, a 500 MW-class unit for Sarawak, with availability as early as 2028. Harbin is building Saudi Arabia's newest combined cycle plants around Siemens turbines, and replaced Russian equipment at Kazakhstan's largest coal plant when the Russian financing never arrived. Chinese capacity is small, around ten G50 units a year rising to twenty, and booked through 2027. But for a buyer in Central Asia, Iraq or Africa who needs a frame before 2029, the choice is now between a Chinese new unit and a Western used one, and the used one has a thirty-year operating record.

Sixty hertz, south of the border

Latin America is where the 60 Hz shortage is most acute, and it is the best read on how much pressure is behind the price, because its grids draw on the same pool of machines as Texas and most of them cannot pay Texas prices. Some of these markets are open to foreign suppliers and some are not; all of them show what a grid in deficit will do.

Venezuela has about 36,000 MW of nameplate capacity and, by the best outside estimates, less than 13,000 MW that actually works. Its thermal fleet runs at roughly 13 percent of capacity, and rationing outside Caracas runs eight to twelve hours a day against a deficit estimated at 3,000 MW. What changed this year is the politics. Since the change of government in January, Washington has issued licenses allowing Chevron and others to negotiate new oil investment, Chevron has committed more than $7 billion to double production, and the National Assembly has advanced a bill opening generation to private concessions for the first time since 2010. In September GE Vernova signed agreements with Corpoelec and PDVSA, witnessed by the US Energy Secretary, to bring 1 GW online within 24 months and 5 GW over the four years after that. Read that timeline against GE Vernova's own 2031 delivery schedule and the arithmetic is plain: a first gigawatt inside two years cannot come off a new production line. It can only come from rehabilitating the existing fleet, which GE Vernova describes as legacy machines that have gone years without OEM parts, and that is what the Planta Centro restart is: a Chinese contractor returning a 600 MW oil-fired unit to service in November. Whatever one thinks of the politics, Venezuela is the clearest demonstration anywhere that when the queue is closed, the existing fleet is the only fleet.

Ecuador is the opposite case: a government ready to buy that the factories cannot serve. Seventy percent of its electricity is hydro, and when the 2024 drought emptied the reservoirs the country went dark for up to 14 hours a day at a cost its finance minister put at 1 to 1.5 percent of GDP. The government signed seven emergency generation contracts worth $706 million. None of the contracted engines ever produced power, the contracts were terminated and arbitration followed. Nine thermal tenders totalling 871 MW have failed since 2024. When Celec tried to buy 355 MW of combined-cycle turbines for Pascuales, GE offered delivery in mid-2029 and Ansaldo, invited as the sole bidder, did not bid. In May the energy minister said what every small buyer has learned privately: the major OEMs do not want to sell to Ecuador, new units are being offered for around 2032, and the government is now targeting roughly 500 MW in 2026 from abandoned and idle equipment, including 1970s-era turbines fitted with new controls, while leasing powerships to survive each dry season.

Cuba's grid collapsed seven times in 2026. In July the state utility forecast a record deficit of 2,380 MW against 3,200 MW of peak demand, with 850 MW available and 106 distributed plants out of fuel; the Turkish powerships that once supplied more than 600 MW had dwindled to two before a Russian crude delivery allowed two more to restart. Mexico moved 150 MW of portable turbines to Yucatán and 128 MW of mobile turbogas units to Baja California Sur in 2026 while planning 5.8 GW of new combined cycle, and CFE has run international tenders for temporary plants. The Dominican Republic plans more than 2,500 MW of tenders toward 2028, including 900 MW of existing thermal to be contracted for late 2027. Guyana's powership contract expired in May, and the operator declined a 30-day extension while asking for a rate increase from 7.6 to 9.5 cents a kilowatt-hour.

The most telling Latin American signal is Brazil, which has the region's deepest capital markets and still chose old iron. The March 2026 reserve capacity auction contracted 18,977 MW, of which 7,609 MW came from 35 existing thermal plants at discounts near zero, alongside 8,866 MW of new builds that will take years. Wärtsilä won a ten-year contract in the same auction by converting three existing heavy-fuel engines at a 50 MW plant in Pernambuco to gas rather than shipping new ones. When the buyer with the most options takes the existing plant at full price, the market is telling you what existing plants are worth.

The powership is the thermometer

If you want to know where turbines will be bought next, look at where diesel and LNG barges are being paid for now. Karpowership operates 45 powerships with more than 8,000 MW in some twenty countries, and in 2026 it ordered four new 300 MW combined-cycle vessels from a Korean yard for 2028 delivery and signed with Seatrium in Singapore to convert five more at up to 600 MW each. Nobody builds a $400 million floating power plant for a market they expect to clear.

The pattern in that table is the pattern of the used-turbine market. Every one of those contracts is a government paying a premium for megawatts it could not otherwise get, and several of them, Gabon and Guyana most visibly, are now trying to replace the ship with a plant on land. A relocated gas turbine is what replaces a powership. The countries at the top of the table are the ones to watch.

Reconstruction, and the order book that cannot be filled

The 50 Hz world has its own version of the data center boom, and it is reconstruction. Ukraine's generation fell from more than 37 GW before the invasion to under 14 GW by the end of 2024, all fifteen of its thermal plants have been hit, and the World Bank's latest damage assessment puts the energy sector's ten-year need at $90.6 billion, $71 billion of it for generation. What Ukraine is buying today is small and distributed, because small and distributed is hard to hit: about 1.8 GW of gas engines and turbines have been commissioned since the invasion, with another 532 MW under construction, financed in part by the EBRD, whose wartime support passed €10.5 billion in June. The large frames come later. DTEK and GE Vernova have signed a memorandum for a 650 MW combined cycle at Burshtyn, about €900 million, with contracts in 2027 and power in 2031. That is the OEM timeline. Anyone who can put a refurbished 50 Hz frame on the ground in Ukraine in 2028 is selling into a market with no competition.

The Middle East is the other reconstruction bid, and it is larger. Syria's capacity fell from about 9.5 GW before 2011 to under 2 GW that runs; a Qatari-led consortium signed concessions in November for four combined cycle plants totalling 4,000 MW, GE Vernova and Siemens Energy are negotiating to supply them, and Azerbaijani gas is already flowing to Aleppo. Qatar has proposed two 825 MW plants for Lebanon, where the state utility delivers power a few hours a day. Iraq lost more than 3 GW overnight in March when Iranian gas flows stopped, faces a summer shortfall of roughly 11 GW against 40 GW of demand, and is advancing a strategic agreement with GE for up to 24,000 MW while chartering powerships for the summer and taking delivery of Chinese G50s. Kuwait, which has added no major capacity since 2016 and lost units to drone strikes in July, is tendering 900 MW of gas turbines at an existing plant. Add it up and the region is asking for something like 30 GW of 50 Hz frames from factories that are sold out through 2030. The gap between what has been signed and what can be delivered before 2031 is the secondary market's opening, and it is measured in gigawatts.

The mirror image

Every shortage above has a mirror image: a market with turbines it cannot run. Fleets built for demand that never arrived, plants idled by fuel they cannot buy or bills they cannot collect, projects financed on run hours the grid no longer needs. That iron is the supply side of the international secondary market, and it is what the buyers above are competing for.

The clearest precedent is Turkey, which has been exporting gas plants for a decade. As early as 2015 Turkish independent producers were moving turbines to West Africa, Iran and Kazakhstan because domestic run hours had fallen to a third of what their financing assumed. Aksa's 240 MW combined cycle plant was dismantled at Antalya and rebuilt in Tashkent between 2019 and 2021, and Aksa now operates 740 MW in Uzbekistan. Relocation at that scale is routine engineering, and the firms that do it treat frequency conversion as a standard scope item. The constraint has never been whether a plant can move. It is whether a buyer will pay for it to, and in 2026 the buyers will.

Beyond the acute cases are the markets whose problem is structural. South Africa's first gas-to-power tender, 2,000 MW on offer, closed in late October 2026 with no preferred bidder because none could secure gas, and its 3,000 MW Richards Bay plant is targeted for 2031, which is why its energy analysts expect load shedding to return around 2030 when 5,260 MW of coal closes. Egypt hit a record 40,000 MW peak in August and is importing LNG by the cargo to keep the lights on. Kazakhstan imported 2.5 billion kWh from Russia last year and plans 4.5 GW of combined cycle across eleven regional centres; Uzbekistan rationed evening power in December after a rotor failure on a brand-new turbine. These are 50 Hz markets with real demand, gas, and governments willing to sign.

Europe's data centers, slowly, and Asia's wall

Europe's data center story is slower and quieter, and the slowness is regulatory rather than commercial. Dublin has had a de facto connection moratorium that runs to 2028, and the Irish regulator's answer has been to require new data centers to bring their own generation, which in practice means gas: seventeen planned Irish campuses have applied for gas connections with a combined peak equal to six 500 MW combined cycle plants, and the first 110 MW on-site microgrid, built on Wärtsilä dual-fuel engines, is due to energize at the end of this year. In Britain more than a hundred data centers plan to self-generate on gas, a 300 MW gas-powered campus in Buckinghamshire was cleared in March to proceed as nationally significant infrastructure, and engine makers describe a pipeline of European projects at up to 100 MW each. In Frankfurt grid operators are telling applicants to wait beyond 2032, and the Netherlands has 15 GW of data center connection requests queued against a ten-year wait. Siemens Energy says roughly 35 percent of its turbine orders now come from Europe and that customers are paying premiums for speed. The European developers who are pushing through are accepting longer timelines than their American peers, and the ones who want to shorten them are starting to look at the same place everyone else is.

Asia is where the wall is closest. Malaysia's data center pipeline is 7,500 MW across 56 projects against 850 MW of actual load at the end of 2025; analysts expect Johor's campuses to absorb more than 90 percent of peninsular generation additions over the next five years, and the national gas tender has drawn bids for roughly 3.5 GW of new plants from YTL, Petronas, TNB and Genting. Some OEMs are asking Malaysian bidders for a third of the turbine price as a down payment to hold a slot, and the first Chinese heavy-duty frame in the country is now on order for 2028. India's data center pipeline exceeds 8 GW with gigawatt campuses proposed at Jamnagar and Visakhapatnam. Vietnam's 18 LNG-to-power projects have one on schedule and a trade ministry warning of shortages from 2027. Every one of these is a 50 Hz market, every one of them is discovering that the factories are full, and the Wood Mackenzie numbers at the top of this piece are the result. The developers who cannot wait will do what Americans did two years ago: find the machine that already exists.

What this means

If you own idle 50 Hz equipment, your buyers are not in Texas, but they are real and they are closer than they were a year ago. They are the reconstruction programs that have signed for gigawatts they cannot receive before 2031, the Latin American grids that are paying powership rates for megawatts, the African and Central Asian utilities with gas and demand and no factory slot, and, by 2028, the European and Asian data center developers who are a cycle behind the Americans. If your equipment is geared, a 6B, an SGT-800, an aeroderivative, you have both markets, and the American one will pay more. If it is a direct-drive 50 Hz frame, price it for the 50 Hz buyers and understand that their ability to pay is rising as the queue lengthens.

If you are buying, the supply exists, and it is not on a listings page. It is in the surplus fleets and the stalled projects described above, and it takes owner-level work to reach. Prices, when the equipment has records and a shop behind it, are converging on what a new machine costs, because the new machine arrives in 2031 and the used one does not.

The international market has always been where used turbines went to find a second life. What is new is that the lives are being bid for, by governments and by data centers, at prices that would have been laughed at in 2023. That is not a forecast. It is a description of a port in the Philippines in June.