General Compute Partners with Positron AI
Atlas joins our inference offering in Q1 2027, with a commitment to Titan, powered by Asimov.
General Compute is partnering with Positron AI to expand our dedicated inference offering. We will purchase Atlas systems and have committed to purchasing next-generation Titan systems.
Atlas capacity in Q1 2027
Atlas will be available to our customers beginning in Q1 2027. The deployment builds on Positron's experience integrating its hardware alongside NVIDIA systems in existing data centers.
A path to Titan and Asimov
Titan will use Asimov, Positron's custom inference chip. Its LPDDR-based architecture combines terabytes of memory with bandwidth for responsive generation, aiming to hold larger models and longer context in memory.
The partnership supports matching each stage of inference to suitable hardware: GPUs process prompts, while memory-first accelerators generate tokens. Customers get dedicated capacity through General Compute with a single contract and shared SLAs.