← Back to stories Spacious industrial warehouse interior featuring machinery and network of pipes.
Photo by Sonny Vermeer on Pexels
钛媒体 2026-03-09

Europe and the U.S. scramble for gas turbines as AI drives a power squeeze — China steps in

A new grid bottleneck meets an old technology

A global power crunch is converging on one unlikely protagonist: the gas turbine. As hyperscale AI data centers proliferate and run around the clock, their voracious, always-on electricity needs are colliding with the variability of wind and solar. The result? Grid operators in Europe and the United States are racing to secure fast-ramping, highly reliable generation. Why gas turbines now? They start in minutes, stabilize frequency, and fill gaps when renewables dip — while emitting less CO₂ than coal.

It has been reported that the world faces a structural shortfall in heavy-duty turbines: global “intent” orders could exceed 80 GW in 2025, versus roughly 50 GW of deliverable capacity, leaving a 30 GW-plus gap. Backlogs at Siemens Energy, GE Vernova and Mitsubishi Heavy Industries reportedly stretch to 2028, with delivery times lengthening to five to seven years and prices jumping more than 20%. Chinese media estimate that a single U.S. hyperscale site can draw power comparable to a mid-sized city; one oft-cited example puts Google’s Oregon data center near 3.5 billion kWh annually. In a world where milliseconds matter for AI uptime, the market is voting for dispatchable power.

China’s late start turns into a window

Enter China’s turbine makers. After decades of catch-up under technology transfer limits and export controls on high-temperature materials and core designs, Dongfang Electric (东方电气) has emerged as the domestic standard-bearer. According to Chinese reports, the company shipped a domestically designed heavy-duty turbine platform (G50, around 50 MW) in 2022 and achieved stable commercial operation on a demonstration unit in 2023 — a symbolic break from reliance on imported cores. In 2025, Dongfang Electric reportedly secured an export order to Kazakhstan for three G50 units worth about 1.5 billion yuan, with production slated for 2026, marking a first for Chinese-made heavy-duty turbine exports.

The broader ecosystem is moving in tandem. Shanghai Electric (上海电气) focuses on heavy-duty assembly and large power-plant integration, while Harbin Electric (哈尔滨电气) brings depth in main and auxiliary systems, especially for overseas projects. Up the chain, suppliers are maturing in critical alloys, blades, and forgings — including firms tied to China’s aero-engine complex such as AECC Aviation Power (航发动力). Together, they are building the full-stack industrial base that long eluded China. It has been reported that Dongfang Electric’s domestic share has risen above a third, and its order book now extends into 2027.

Geopolitics and the gas reality

Geopolitics looms large. Europe’s post-Ukraine-war energy calculus hinges on LNG and flexible backup as Russian pipeline gas recedes; the U.S. faces grid bottlenecks and slow permitting even as it leads in AI infrastructure. Meanwhile, technology transfer in advanced turbine materials and design has historically been restricted for China — a constraint that, paradoxically, pushed sustained indigenous R&D. As Western OEMs juggle backlogs, Chinese manufacturers are reportedly expanding in emerging markets and eyeing more stringent U.S./EU arenas, where certification, export controls, and sanctions risk remain hurdles.

The climate trade-off is clear: more gas turbines can stabilize high-renewables systems and keep AI online, but they also lock in gas supply and methane-management challenges. In the near term, however, the market signal is unmistakable. With Europe and the U.S. scrambling for firm capacity, and Western order books choked, China’s turbine sector — led by Dongfang Electric (东方电气) and flanked by Shanghai Electric (上海电气) and Harbin Electric (哈尔滨电气) — has found a rare alignment of industrial readiness and global demand. Can a once-closed technology club accommodate a new entrant just as AI forces the grid to reinvent itself?

AI
View original source →