Positron AI, a rising player in the AI hardware space, has announced that it has raised $51.6 million in an oversubscribed Series A funding round to accelerate the development of its purpose-built inference hardware platform. The funding marks a significant step toward scaling high-performance and energy-efficient AI processing for enterprise applications.
The round was led by Celesta Capital and Intel Capital, with additional participation from M12 (Microsoft’s Venture Fund), Kleiner Perkins, Human Capital, and GSR Ventures. The capital infusion brings Positron AI’s total funding to over $70 million, including prior seed investments and grants.
Building a Foundation for AI Inference at Scale
Founded in 2021, Positron AI focuses on creating domain-specific inference engines tailored to real-world enterprise use cases. While training large AI models remains compute-intensive and typically reserved for major cloud providers, the inference stage—where trained models are deployed for tasks like language processing or image recognition—is increasingly critical across industries.
The company’s proprietary MatrixStream™ architecture is designed to address the performance bottlenecks associated with general-purpose GPUs and CPUs in inference workflows. According to Positron AI, its chips can deliver up to 10x better throughput-per-watt compared to existing inference processors.
“While there has been massive innovation in AI training hardware, inference has largely been an afterthought,” said Arjun Kapoor, CEO and co-founder of Positron AI. “We’re building hardware and software that reimagines inference from the ground up—optimized for latency-sensitive, real-time applications.”
Enterprise-Grade, Developer-First
Positron AI’s inference solutions are targeted at enterprise developers building AI products for finance, healthcare, logistics, and retail. The company’s hardware integrates seamlessly with existing ML frameworks like PyTorch and TensorFlow, and supports APIs and tools that ease deployment in private and hybrid cloud environments.
Kapoor added that the company is “on a mission to democratize scalable inference by making deployment as easy and efficient as training has become in recent years.”
A Surge in Interest Amid AI Infrastructure Boom
The Series A round comes amid surging demand for AI infrastructure, as enterprises adopt generative AI, large language models (LLMs), and multimodal AI at scale. While companies like NVIDIA dominate the training GPU market, a growing number of startups—including Positron AI, Groq, and Tenstorrent—are vying for leadership in the inference layer.
“With AI workloads growing exponentially, there’s a clear need for specialized solutions that can handle inference cost-effectively and sustainably,” said Sriram Viswanathan, Founding Managing Partner at Celesta Capital. “Positron AI is uniquely positioned to fill that gap.”
Plans for Expansion and Talent Growth
The newly secured funds will be used to accelerate product development, expand customer pilots, and scale engineering and go-to-market teams across the U.S., India, and Southeast Asia. Positron AI is also hiring for roles in silicon design, ML engineering, and customer success.
According to company officials, the first production versions of its MatrixStream processors are expected to be available for enterprise evaluation later this year, with general availability scheduled for early 2026.
Positioned at the Core of AI Inference Transformation
As the AI landscape shifts toward real-time and edge computing, Positron AI’s focus on inference aligns with the needs of businesses seeking to reduce latency, cut compute costs, and operate sustainably. By tailoring hardware for inference workloads, the company hopes to unlock new efficiencies in AI deployment across verticals.
The funding also reflects continued VC confidence in AI infrastructure companies, especially those offering differentiated, domain-optimized solutions.