AI 导读
能源将很快成为 AI 的最大瓶颈，中国目前在建 36 座核反应堆，接近全球在建总量的一半。中国以每批 6 到 10 台相同机组进行批量化建设，平均建造时间约 6 年，全球平均约 9 年；2025 年新增约 315 GW 太阳能，约为美国已建太阳能总装机的 1.8 倍。
正文
Energy will be the biggest bottleneck in the very near future.
China has 36 nuclear reactors under construction right now, close to half of everything being built worldwide. They build in batches of 6 to 10 identical units, run one standardized design across sites, and source most components domestically. Average build time dropped to about 6 years. The global average sits closer to 9.
China also installed about 315 GW of new solar in 2025, roughly 1.8x the entire US solar fleet ever built, almost all of which went up in the last 15 years.
You see my point, right?
The biggest bottleneck will be energy- very soon. Gartner's 2026 forecast puts global data center electricity at 565 TWh, up 26% from last year. AI servers already account for 31% of that and pass conventional servers in 2027. What's worth noting is the constraint Gartner names: it's power, not chips. They project demand above 1,200 TWh by 2030 and warn the grid won't keep up. So the race quietly shifts from who has the best silicon to who can actually get the electricity to run it.在 X 查看被引用的帖子
来源：@kimmonismus · x.com