AI Infrastructure Arms Race: Inference + Chips (Together AI, Fireworks, SambaNova, Etched, General Compute, Infinity)
Key Questions
What funding did Together AI secure?
Together AI raised $800M Series C at an $8.3B valuation with $1.15B in bookings and 500MW compute capacity. Backers include Aramco, NVIDIA, and Vista.
How does Fireworks AI compare to Together AI?
Fireworks AI raised $1.5B Series D at $17.5B valuation with $1B ARR and 5x growth, processing 40T tokens per day. It directly competes with Together AI in the inference space.
What is Infinity's focus and recent raise?
Infinity raised $15M seed at $100M valuation to build chip-agnostic AI inference optimization software. It partners with d-Matrix and aims to reduce NVIDIA's software advantage.
What chip-side investments are highlighted?
SambaNova raised $1B at $11B valuation with plans for a 2027 IPO, while Etched seeks $800M at a potential $20B valuation despite no revenue. Positron is also mulling a $750M round at $5B valuation.
What trends are emerging in AI infrastructure?
Open-weight inference, custom chip investments, and enterprise on-premise deployments are accelerating. Valuation inflation and bubble warnings are also noted in the sector.
Together AI $800M Series C at $8.3B val ($1.15B bookings, 500MW compute, Aramco/NVIDIA/Vista). Fireworks AI $1.5B Series D at $17.5B val ($1B ARR, 5x growth, 40T tokens/day) — directly competes with Together AI. Ollama $65M with 8.9M devs, 85% Fortune 500. Chip side: SambaNova $1B at $11B val (General Atlantic, IPO 2027); Positron mulls $750M at $5B; Etched $800M (seeking $20B val, no revenue). General Compute $400M loan backed by SambaNova chips. New: Infinity $15M seed for inference software layer (chip-agnostic optimization, d-Matrix partnership). Open-weight inference shift accelerating. Custom chip investment and enterprise on-premise AI deployment. Valuation inflation and bubble warnings emerging.