Exclusive: Positron raises $230 million in Series B to take on Nvidia’s AI chips

Semiconductor startup Positron has secured $230 million in Series B funding, TechCrunch has learned exclusively. The organization plans to use the capital to accelerate the deployment of high-speed memory chips, a key component of chips used for AI workloads, people told TechCrunch.

Investors in the round also include Qatar Investment Authority (QIA), the country’s sovereign wealth fund that is increasingly focused on building AI infrastructure, the people said.

The Reno-based startup’s Series B comes as hyperscalers and AI companies push to reduce their dependence on longtime leader Nvidia. Those companies include OpenAI, which despite being one of Nvidia’s largest and most important customers, is dissatisfied with some of the company’s latest AI chips and has reportedly been looking for alternatives since last year.

Meanwhile, through QIA, Qatar is accelerating its broader commitment to so-called “sovereign” AI infrastructure, a priority reiterated at the Web Summit Qatar in Doha this week. Sources told TechCrunch that the country sees computing power as essential to remaining competitive on the global economic stage, positioning itself as the Middle East’s leading AI services hub and increasing interest in startups like Positron.

This strategy is already taking shape through major initiatives, including a $20 billion AI infrastructure joint venture with Brookfield Asset Management announced in September.

Positron’s funding brings the three-year-old startup’s total raised to just over $300 million. The startup raised $75 million last year from investors including Valor Equity Partners, Atreides Management, DFJ Growth, Flume Ventures, and Resilience Reserve.

The company claims its first-generation chip Atlas, manufactured in Arizona, can match the performance of Nvidia’s H100 GPU with less than a third of the power. Positron’s focus on inference (the computing required to run AI models for real-world applications) rather than training large-scale language models positions the company as demand for inference hardware soars as companies shift their focus from building large-scale models to deploying them at scale.

Sources told TechCrunch that beyond memory capabilities, Positron’s chips also excel in high-frequency and video processing workloads.

TechCrunch has reached out to Positron for more information.