Positron AI Raises $875M at $5B Valuation to Challenge NVIDIA in Inference

NEA and Atreides lead the Reno chipmaker's Series C as its Atlas systems run inside Oracle Cloud and next-gen Asimov nears tapeout.

Positron AI, a Reno, Nevada-based chipmaker building hardware for running AI models, has raised $875 million across Series C and Series C-1 financing at a $5 billion post-money valuation.


NEA, Atreides Management, Valor Equity Partners, Andra Capital, SemiAnalysis Capital and Netscape co-founder Jim Clark led the round. DFJ Growth, the Qatar Investment Authority and a long list of others participated, along with strategic investors Hudson River Trading, Cisco Investments and Naver Ventures.


How the Round Breaks Down


The financing includes a $375 million Series C priced at a $3.5 billion pre-money valuation, plus a Series C-1 of up to $500 million. Thomas Jermoluk is joining the company's board.


A Memory-First Bet


Training AI models has made NVIDIA the most valuable chip company in the world, but inference, the work of actually running those models for users, is becoming the larger and faster-growing workload. Inference performance is often limited less by raw compute than by how much memory a chip can access and how quickly.


Positron is designing around that constraint. Its first-generation Atlas system is already running at scale, with more than 50 racks deployed inside Oracle Cloud Infrastructure. Other customers include Parasail, Jump Trading and i3d.net.


Its next-generation Asimov chip is set to tape out at the end of 2026 on TSMC's N3P process, with production planned for the second half of 2027. Each Asimov chip will support between 288GB and 2,304GB of memory, using lower-cost LPDDR5X memory, SiliconANGLE noted. A multi-chip system called Titan will combine four to eight Asimov chips.


"Positron is making the boldest memory-first bet in AI hardware," said NEA partner Forest Baskett, pointing to Atlas already running at scale inside Oracle's cloud.


Where the Money Goes


Positron will use the funding to tape out Asimov, build a 2-megawatt-plus engineering data center and emulation platform, ramp Titan production and lock in LPDDR5X memory supply commitments.


"Speed matters in this market," said CEO Mitesh Agrawal, adding that lessons from Atlas have been carried directly into Asimov and Titan.


Why It Matters


An $875 million round makes Positron one of the best-funded challengers in AI chips, a category where hyperscalers and model developers are actively looking for alternatives to NVIDIA to control cost and supply.


The investor mix, spanning venture firms, sovereign wealth, trading firms and networking giant Cisco, signals broad demand for cheaper inference. The real test comes next year, when Asimov has to move from tapeout to production on schedule.


Source


PR Newswire — "Positron AI Raises $875 Million at a $5 Billion Valuation to Bring Its Next-Generation Inference Silicon to Market"

0 Comments