Hyperbolic's $30M Bet That GPU Capacity Becomes a Traded Market
Hyperbolic raised $30M in Series A-1 funding to turn fragmented GPU capacity into a market it can price, resell and eventually hedge. The hard part is defining what gets delivered.
Explore the foundational technologies and hardware driving the artificial intelligence boom. From next-generation data centers and custom silicon to advanced networking and energy solutions, discover the deep tech investments building the physical and digital layers that power the AI era.
Hyperbolic raised $30M in Series A-1 funding to turn fragmented GPU capacity into a market it can price, resell and eventually hedge. The hard part is defining what gets delivered.
ASI raised $325 million at a $3 billion valuation. Its edge is forecasting physical operations — air traffic, grid load, flight schedules. Its risk is that two federal contracts never become a platform.
Luminal's Form D shows $29.2M from 84 investors. The compiler-first startup publishes a 38% throughput lead over vLLM: a real number that is smaller than its own 3.2x claim.
Fireworks AI is reportedly shopping a round at $30-40 billion, roughly double its $17.5 billion July mark. At 40 times a July run-rate, the valuation assumes growth nobody has disclosed yet, and a credential incident still being contained.
Blacksmith went from a $60M valuation to $1B in twelve months by selling dedicated CI infrastructure. The round is a bet that validating AI-written code, not writing it, is now the constraint on shipping software.
A 15-person Dutch team raised €10 million to industrialise the light source inside co-packaged optics. The argument is about manufacturing yield, not optical power — and the 2028 volume claim is still two years from being testable.
Daytona closed a $60M Series B at a $1B valuation, eight months after its Series A. Its product is a virtual machine that lives for 60 milliseconds, and the price assumes the spread between per-second billing and owned capacity holds.
Per-token prices have collapsed. The infrastructure bill has not. A Microsoft Research lifecycle model shows why refresh timing, power topology and cooling decide whether cheap inference turns into a cheaper business.