SoftBank Cancels $3.7 Billion in Gas Turbine Orders Amid AI Market Uncertainty
SoftBank abandons four gigawatts of turbine reservations as AI-driven energy demand projections face scrutiny.
1Strategic Shift in Energy
SoftBank Group is taking a $3.7 billion loss after canceling reservations for approximately four gigawatts of natural gas turbines. Initially, the conglomerate secured rights to 13 gigawatts of capacity to power future AI data centers. This move marks a significant reversal for Masayoshi Son’s strategy to capitalize on surging electricity demands.
2Turbine Supply Chain Constraints
Global demand for gas turbines has depleted supply, with major manufacturers like GE Vernova reporting order books filled through 2031. SoftBank’s attempt to corner this market failed to yield the expected leverage. Consequently, developers are increasingly turning to less efficient alternatives, including diesel generators and repurposed jet engines, to meet immediate power needs.
3Ohio Data Center Project
Despite the cancellations, SoftBank remains committed to a 9.2-gigawatt power plant in Pike County, Ohio. This facility, backed by Trump and intended to support an eight-gigawatt data center leased to OpenAI, involves $33 billion in construction costs. Nvidia has reportedly entered discussions to guarantee $250 billion in financing for this massive project.
4Addressing Default Allegations
SoftBank’s CFO, Yoshimitsu Goto, explicitly denied claims made by Elon Musk regarding a default on turbine blade contracts. The company maintains that the decision to unwind these specific orders was entirely voluntary. They assert that the original reservations were a precautionary measure against anticipated equipment shortages that did not materialize as expected.
5Market Sentiment and Volatility
The broader AI sector is experiencing significant market turbulence, with major firms like Alphabet, Meta, and Microsoft seeing monthly declines. Nvidia’s stock fell 11% over five days following reports regarding CEO Jensen Huang. These financial pressures suggest that the initial AI boom is transitioning into a period of complex economic uncertainty.
6Efficiency Concerns in Infrastructure
Public filings indicate the Pike County facility may not utilize combined-cycle turbine configurations, which are typically more efficient. If built without this technology, the plant's carbon intensity will likely exceed that of most modern natural gas installations. This raises questions about the long-term environmental impact of massive AI-focused energy infrastructure projects.
7Broadening Sector Weakness
Signs of cooling demand are evident across the AI supply chain. Memory and chip makers like Micron and Applied Materials have seen valuations drop by over 25%. Similarly, GE Vernova and Oklo have experienced nearly 20% declines. These trends indicate that investors are reevaluating the sustainability of rapid, AI-driven capital expenditure cycles.
8Future Outlook for Data Centers
While Anthropic prepares for a major IPO, the overall economic trend for data centers appears to be peaking. The failure to sell excess turbine reservations suggests that energy-producing asset demand may be weaker than previously projected. Stakeholders must now navigate a landscape where AI technology improvements contrast with cooling infrastructure investment enthusiasm.