Press

Positron Announces Partnership with Altera for Silicon Powering Faster LLM Inference

Positron has chosen Altera's Agilex line for its faster memory bandwidth speeds and availability at scale.

Building the New Hardware for AI Model Inference

Thomas Sohmers outlines Positron's unique approach to building the next generation of AI accelerators, including focusing on memory capacity and avoiding introducing a new compiler by ingesting model files directly.

The Gigawatt Gamble: The Economics for Global AI Inference Deployment with Positron AI

Thomas Sohmers explains that running modern transformer models is consuming more energy and water every day. Adam Smith's theory of the "invisible hand" of the free market opens the window for lower cost competition to the GPU giant in terms of serving models more efficiently, in terms of both power and cost.

Positron Ranked #3 on The Information’s 50 Most Promising Startups for 2024

The Information, a renowned news outlet for the digital age, with a knack for breaking impactful news stories on Artificial Intelligence, ranked Positron #3 on its ranking list of the 50 Most Promising Startups for 2024.