Chipmaker Positron nabs $875M to speed up inference with consumer-grade memory - SiliconANGLE

Chipmaker Positron nabs $875M to speed up inference with consumer-grade memory - SiliconANGLE

Positron raised $875M for inference accelerators using LPDDR5X instead of HBM, claiming 26x better cost-per-token than Nvidia Blackwell in tests. Infrastructure leaders gain a viable Nvidia alternative for LLM deployment, potentially reshaping datacenter AI procurement strategy.