Key Takeaways
- • Positron AI secured $875 million in Series C funding at a $5 billion post-money valuation, with NEA, Atreides Management, Valor Equity Partners, Andra Capital, SemiAnalysis Capital, and founder Jim Clark as lead investors.
- • Positron's Atlas inference hardware is already deployed at hyperscaler scale, while next-generation custom silicon Asimov will tape out in late 2026 with production commencing in H2 2027.
- • The company's architecture eliminates dependency on constrained HBM and CoWoS supply chains while supporting flexible deployment in air-cooled or liquid-cooled data centers at varying rack densities.
- • Foley & Lardner LLP, led by partners Natasha Allen and Louis Lehot, represented NEA in the transaction.
- • Positron's product roadmap targets long-context and next-generation AI workloads through Titan, a multi-terabyte-memory inference system powered by Asimov silicon.
Summary
Positron AI, an AI inference hardware specialist focused on reducing computational costs and energy consumption for model serving, closed an $875 million Series C funding round that values the company at $5 billion. The financing was co-led by NEA, a global venture capital firm with more than 270 portfolio company IPOs and 450+ mergers and acquisitions to its credit, alongside Atreides Management, Valor Equity Partners, Andra Capital, Dylan Patel's SemiAnalysis Capital, and Jim Clark, founder of Silicon Graphics and Netscape. Liberty Global and the Qatar Investment Authority participated alongside additional investors DFJ Growth.
Positron's existing product, Atlas, is already deployed at hyperscaler-level production today, addressing a critical market need for cost-efficient AI inference. The company's technical differentiation centers on energy-efficient architecture that enables deployment flexibility across varied data center configurations without reliance on supply-chain-constrained components such as HBM (high-bandwidth memory) or CoWoS (chip-on-wafer-on-substrate) packaging. This supply chain independence represents a significant operational advantage in an environment where semiconductor component scarcity remains a material constraint for hyperscale AI deployment.
The capital injection will accelerate Positron's path to next-generation silicon. The company's custom Asimov chip is scheduled to tape out in late 2026, with mass production targeting the second half of 2027. Asimov powers Titan, a multi-terabyte-memory inference system engineered to handle long-context workloads and emerging generational AI model architectures. This product roadmap directly addresses the evolving infrastructure demands of hyperscalers deploying increasingly sophisticated language and reasoning models.
Foley & Lardner LLP provided legal representation for NEA in the transaction, with the deal led by partners Natasha Allen and Louis Lehot, supported by associates Saige Gallop, Sarina Rogers, and Sarah Waste. The transaction underscores investor confidence in Positron's technical approach to the inference hardware market, where competition for alternative architectures to existing dominant solutions continues to intensify.
Positron's existing product, Atlas, is already deployed at hyperscaler-level production today, addressing a critical market need for cost-efficient AI inference. The company's technical differentiation centers on energy-efficient architecture that enables deployment flexibility across varied data center configurations without reliance on supply-chain-constrained components such as HBM (high-bandwidth memory) or CoWoS (chip-on-wafer-on-substrate) packaging. This supply chain independence represents a significant operational advantage in an environment where semiconductor component scarcity remains a material constraint for hyperscale AI deployment.
The capital injection will accelerate Positron's path to next-generation silicon. The company's custom Asimov chip is scheduled to tape out in late 2026, with mass production targeting the second half of 2027. Asimov powers Titan, a multi-terabyte-memory inference system engineered to handle long-context workloads and emerging generational AI model architectures. This product roadmap directly addresses the evolving infrastructure demands of hyperscalers deploying increasingly sophisticated language and reasoning models.
Foley & Lardner LLP provided legal representation for NEA in the transaction, with the deal led by partners Natasha Allen and Louis Lehot, supported by associates Saige Gallop, Sarina Rogers, and Sarah Waste. The transaction underscores investor confidence in Positron's technical approach to the inference hardware market, where competition for alternative architectures to existing dominant solutions continues to intensify.
Go deeper than the headline
Research every firm in this story, track the ones that matter and get the daily briefing — free for 7 days.