The physical substrate of the AI era.
Compute demand has turned data centres into an infrastructure story: land, grid connections, cooling water and transformers now gate how fast AI capacity can grow.
According to a source, the new financing would more than triple the AI infrastructure startup's valuation from four months ago.
Source: TechCrunch · Read the original
The next generation of computing presents mounting heat challenges with every new launch.
Source: IEEE Spectrum · Read the original
The set of techniques — quantisation, batching, caching, distillation and specialised silicon — that reduce the cost and latency of running a trained model.
Moving heat away from servers using liquid — either cold plates attached to chips or full immersion in a dielectric fluid — instead of moving air.
AI capacity planning has become an electricity problem. Interconnection queues, transformer lead times and cooling water now determine build schedules more than chip supply does.
SignalBuyers with round-the-clock load are paying for generation they can schedule. That is pulling capital toward nuclear, geothermal and multi-day storage rather than additional intermittent capacity.
Huburb Daily
Five developments that matter, why each one matters, plus a company and technology worth watching.