软银71亿美元季度亏损,或给AI天然气雄心封顶
SoftBank’s billion-dollar loss may have put a cap on AI’s natural gas ambitions
- 1软银周二公布71亿美元季度亏损,过半来自发电设备相关损失,涡轮机定金被认定大部分失去价值。
- 2软银两年前起为13吉瓦以上涡轮机付定金,超出需求近50%,这笔押注规模超过30亿美元。
- 3燃气轮机订单排队已至2030年代初,孙正义曾押注以巨额资金锁定部分AI电力市场。
- 4行业人士认为数据中心更宜接电网配大电池调峰,燃气市场或慢于软银预期,但难现大规模取消。
软银想靠超额订货锁定部分AI供电市场,这笔巨额亏损折射出燃气抢购潮可能见顶,行业开始重新权衡电网与储能路线。
- 影响:数据中心供电方案的权重在变,电网接入加大型电池可能比抢购燃机更划算;天然气市场扩张或比软银预期更慢,但AI电力需求仍在,供电紧张短期难解;为锁定稀缺产能付出的不可退还定金是真实成本,需计入能源预算。
- 参考:企业在规划算力用电时可考虑把电网接入加储能方案与自建燃机做一次成本对比;涉及长周期设备采购时,可先评估不可退还定金的敞口上限,再把产能排期写进合同条款。
- 依据:软银官方公布71亿美元季度亏损,过半与发电设备有关;13吉瓦超额订货为媒体报道口径;电网加电池建议来自Ascend Analytics分析人士,属行业观点而非实测数据。
SoftBank might not be the first to invest in a hot trend, but when it does, it goes big.
Witness its $30 billion bet on OpenAI in early 2025 or its announcement in February that its SB Energy subsidiary would spend $33 billion on a 9.2-gigawatt natural gas power plant. If you can’t be first, be biggest, I guess.
That philosophy apparently spilled over into procurement, where orders were also supersized. Starting about two years ago, SoftBank reportedly put down deposits on more than 13 gigawatts of turbines, almost 50% more than it needed.
The decision to over-order undoubtedly helped secure turbines and speed up SoftBank’s projects.
“SB Energy did not have a choice except to overbook turbines in the future if they wanted some turbines right away,” Peter Gardett, CEO of Noreva, told TechCrunch via email.
The energy world was caught off guard by the swiftness and magnitude of AI’s rising power demands. Queues for turbine orders now extend into the early 2030s. As natural gas was shaping up to be a go-to power source for the AI boom, CEO Masayoshi Son bet that by deploying his investment firm’s significant war chest, he could corner a portion of the market. If the AI world wanted power for its data centers, it would have to come to him.
Now, that bet appears to have fizzled. SoftBank announced a $7.1 billion quarterly loss on Tuesday, with over half of that stemming from losses related to power-generating equipment. The investor had paid turbine manufacturers nonrefundable deposits to secure orders, and now it considered much of those worthless.
The move was also a costly bit of price discovery, Gardett said. Rather than overpay and potentially still be short of turbines, SoftBank overpaid and over-ordered.
“This charge is just that price discovery finally happening,” Gardett said.
The magnitude of that misplaced bet — over $3 billion — reflects some of the electricity mania that has overtaken the AI world. Faced with years-long waits to connect to the grid, many hyperscalers and developers have turned to natural gas as a sort of energy security blanket, something that is proven and familiar.
As AI data centers spend more time in operation, their operators are getting a better feel for what’s required to keep them operating. Natural gas might seem like a safe bet, but there are still risks, including winter curtailments when supplies dip and demand spikes, Brent Nelson, senior managing director of markets and strategy at Ascend Analytics, told the Equity podcast Wednesday.
In light of that, Nelson said the grid might still be their best bet. Most data centers would be best served by a grid connection, albeit a smaller one than they might normally request. To make up for the shortfall, large batteries are a cost-effective choice, Nelson said. They’d help shave peak loads or ride out requests from the utility to curtail electricity use.
Data center operators might be coming to that same conclusion, leading the natural gas market to develop more slowly than SoftBank expected.
SoftBank’s natural gas-related losses are unlikely to have a domino effect, leading to mass cancellations of gas power plants. AI’s power demand appears to be too large for that. For now, natural gas is riding high, though SoftBank may have found its ceiling.
When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.
Tim De Chant is a senior climate reporter at TechCrunch. He has written for a wide range of publications, including Wired magazine, the Chicago Tribune, Ars Technica, The Wire China, and NOVA Next, where he was founding editor.
De Chant is also a lecturer in MIT’s Graduate Program in Science Writing, and he was awarded a Knight Science Journalism Fellowship at MIT in 2018, during which time he studied climate technologies and explored new business models for journalism. He received his PhD in environmental science, policy, and management from the University of California, Berkeley, and his BA degree in environmental studies, English, and biology from St. Olaf College.
You can contact or verify outreach from Tim by emailing [email protected].
来源:TechCrunch · AI · techcrunch.com