| Entity | Qatar Investment Authority (QIA); Positron AI |
|---|---|
| Date | 10 Sep 2026 |
| Instrument | Series C financing of $875 million; post-money valuation USD 5 billion |
| Region | U.S. & Allies |
Sovereign capital backs next-generation AI inference silicon designed to bypass HBM and CoWoS bottlenecks.
On 10 September 2026 the Qatar Investment Authority announced its participation in Positron AI’s $875 million Series C financing that values the AI inference hardware company at USD 5 billion.
The Qatar Investment Authority has taken a position in Positron AI’s $875 million Series C round, valuing the Reno-based inference hardware company at USD 5 billion. The financing was co-led by NEA, Atreides Management, Valor Equity Partners, Andra Capital and Jim Clark, founder of Silicon Graphics and Netscape, together with other investors. QIA had already participated in Positron’s Series B in February 2026.
Positron designs hardware and software that target the cost and energy constraints of AI inference—the stage in which trained models are run. As demand for AI agents, assistants and copilots expands, inference infrastructure faces limits in memory capacity, memory bandwidth and power draw. The company’s memory-first architecture is intended to ease those constraints, reduce dependence on High Bandwidth Memory and Chip-on-Wafer-on-Substrate supply chains, and permit deployment in both air-cooled and liquid-cooled data centers.
Proceeds will fund development and commercialization of the next product generation. This includes the tape-out of Asimov, Positron’s next-generation silicon, and the production ramp of Titan, its multi-terabyte-memory inference system. Atlas, the current shipping product, is already deployed at hyperscaler scale. Asimov is scheduled to tape out in late 2026 with production in the second half of 2027; the chip will power Titan systems aimed at long-context and next-generation workloads.
QIA framed the investment as consistent with its focus on companies driving the next phase of AI adoption and long-term growth. The announcement was issued jointly from Doha and Reno.
For founders building inference or accelerator hardware, the round signals that sovereign capital is prepared to underwrite architectures that deliberately step away from the constrained HBM and CoWoS ecosystem. For enterprise buyers, the presence of Atlas at hyperscaler scale and a clear silicon roadmap reduce execution risk relative to pure-play software inference solutions. For investors, the $5 billion valuation and repeat QIA participation establish a reference point for later-stage AI infrastructure deals that emphasize energy efficiency and supply-chain resilience over raw compute density.
Sovereign participation at this scale in a pure-play inference hardware company indicates that capital is shifting toward infrastructure layers that can operate outside the tightest memory and packaging bottlenecks. The dual participation by QIA across Series B and C, combined with a co-lead group that includes Jim Clark, raises the bar for subsequent raises in the category and places competitive pressure on vendors still dependent on conventional HBM/CoWoS supply.