Positron AI, a Reno, Nevada-based startup building AI inference chips as an alternative to Nvidia's dominant hardware, said Wednesday it has raised $875 million in a two-part financing that values the company at $5 billion — more than quadrupling its valuation in seven months.
The round splits into a $375 million Series C priced at a $3.5 billion pre-money valuation, co-led by NEA, Atreides Management, Valor Equity Partners, Andra Capital and Dylan Patel's SemiAnalysis Capital, and a Series C-1 of up to $500 million led by NEA and technology investor Jim Clark, according to the company's announcement. Sovereign and strategic backers including the Qatar Investment Authority, Hudson River Trading, Cisco Investments and Naver Ventures also participated. Just seven months ago, in February, Positron raised $230 million at a $1.06 billion valuation.
Betting on Memory, Not More Compute
Positron's pitch is architectural: rather than chasing the scarce, expensive high-bandwidth memory that constrains Nvidia's GPUs, its chips are built around commodity LPDDR5X memory, aiming for over 90% memory-bandwidth utilization for the specific task of running trained AI models rather than training them. The company's current-generation Atlas system is already deployed across more than 50 racks at Oracle Cloud Infrastructure, per Data Center Dynamics' earlier coverage of the company. The new funding is earmarked for taping out its next chip, Asimov, on Taiwan Semiconductor Manufacturing Co.'s N3P process by the end of this year, ahead of a production ramp targeted for the second half of 2027, along with building a 2-megawatt engineering data center and locking in LPDDR5X supply commitments.
"Speed matters in this market, both in how quickly we ship new generations of silicon and in how quickly they reach customers. Deploying Atlas at scale taught us an enormous amount about what inference customers actually need, and we have carried those lessons directly into Asimov and Titan."
Mitesh Agrawal, CEO, Positron AI
Asimov's successor system, Titan, will combine four to eight of the chips into a single node designed to support AI models with more than 16 trillion parameters and context windows exceeding 10 million tokens — specifications aimed squarely at the largest reasoning models now being deployed by frontier AI labs. The financing also adds Forest Baskett of NEA, Atreides' Gavin Baker, Thomas Jermoluk of the Jim Clark Office and SemiAnalysis founder Dylan Patel to Positron's board.
The raise lands amid a broader scramble by investors to back alternatives to Nvidia's inference stack as AI compute costs balloon, though Positron's near-quadrupling of its valuation since February — before Asimov has even taped out — reflects how far ahead of commercial proof points AI infrastructure funding has run this year.
The financing has drawn notice well outside the trade press: one independent business-desk account of the same round zeroed in on how far Positron's paper valuation has outrun any commercial proof of its unreleased Titan system, a gap investors are, for now, betting will close.
