Positron.ai

Positron AI is a hardware company at the forefront of accelerating artificial intelligence, with a mission to make advanced machine learning more accessible and efficient. Founded in the spring of 2023, the company is dedicated to providing solutions that offer superior performance per dollar and enhanced energy efficiency. All of Positron's products are designed, fabricated, and assembled in the United States, emphasizing a commitment to domestic manufacturing and supply chain security. The team at Positron brings together over 400 years of combined experience in AI, systems, silicon, and cloud technologies. The company was established to address the growing problem of GPUs becoming a financial burden for organizations deploying AI. The core mission is to deliver inference that is highly efficient, affordable, and proudly American-made, with the ultimate goal of making GPUs an optional component in the AI technology stack.

The company's first-generation product, Atlas, is the world's first accelerator designed specifically for LLM-inference. Developed and shipped within 18 months of the company's inception, Atlas was created to tackle the cost and energy constraints that are hindering growth in the AI sector. Positron is already developing its second-generation system, Titan, which aims to be even faster and more efficient. By leveraging the insights gained from Atlas, Titan is being designed to unlock virtually limitless context and enable the concurrent operation of numerous models and AI agents, supported by terabytes of memory per accelerator. Positron's innovative hardware architecture is engineered to resolve the power, memory, and scalability challenges inherent in legacy infrastructures, thereby offering the lowest total cost of ownership for transformer models. The company's systems are compatible with Hugging Face transformer models and serve inference requests through an OpenAI API compatible endpoint, ensuring seamless integration into existing AI workflows.