Positron AI raises $875 million at a $5 billion valuation for memory-first inference chips
The Reno company's valuation is five times its February figure; its Asimov chip uses commodity LPDDR5X memory to sidestep HBM and packaging shortages.
Positron AI announced an $875 million Series C financing on Thursday at a $5 billion post-money valuation. The Reno, Nevada company builds inference hardware designed around memory capacity and bandwidth rather than raw compute, arguing that serving agents, assistants, and copilots is now constrained by memory and power more than by training. The round was co-led by NEA, Atreides Management, Valor Equity Partners, Andra Capital, Dylan Patel's SemiAnalysis Capital, and Jim Clark, the founder of Silicon Graphics and Netscape. Quartz reports the raise is structured in two tranches: a $375 million Series C priced at a $3.5 billion pre-money valuation, and a follow-on Series C-1 of up to $500 million anchored by NEA and Clark. Additional investors include DFJ Growth, Qatar Investment Authority, Hudson River Trading, Cisco Investments, and Naver Ventures. Positron's prior round was $230 million in February at a $1 billion valuation. Forest Baskett of NEA, Gavin Baker of Atreides, Thomas Jermoluk from Jim Clark Office, and Dylan Patel will join the board. The money will fund the tapeout of the next-generation Asimov chip and the production ramp of Titan systems, following a 50-plus-rack Atlas deployment at Oracle Cloud Infrastructure. Asimov pairs Positron's compute architecture with between 288 GB and 2,304 GB of commodity LPDDR5X memory per chip, avoiding constrained high-bandwidth memory and advanced packaging supply. The company says its systems realize more than 90 percent of available memory bandwidth.