AI giants need more computing power, but massive data centers take too long to build. So companies are increasingly looking at 20–30 MW facilities that can be deployed much faster.
⏺ Anthropic is considering facilities in the UK and Northern Europe.
⏺ OpenAI is looking at Northern Europe and the US.
The reason is simple: training AI models requires enormous computing capacity, while running already-trained models - inference - can be handled by smaller clusters.
According to JLL, inference capacity could already exceed training capacity by 2027.
AI infrastructure may therefore expand not only through massive hyperscale data centers, but also through smaller, faster-to-deploy facilities.
📈 Trade stocks and other markets with TonTrader 🚀
🔗 Chat • X • TonTrader