Google Alum Raises $500M to Compete With Nvidia
Summary
AI chip startup MatX has secured over $500 million in funding to develop competitive hardware against Nvidia. CEO Reiner Pope shares insights on the unique features of their AI chips in an interview with Caroline Hyde and Ed Ludlow on “Bloomberg Tech.”
Key Insights
What makes MatX's AI chips different from Nvidia's existing products?
MatX is developing a chip that combines two distinct chipmaking approaches: high-bandwidth memory (HBM) used by Nvidia and Google for training AI models, with static random access memory (SRAM) used by other companies for faster inference processing. According to MatX CEO Reiner Pope, 'Our position is that it is actually possible to do both in the same product and you get a much better product as a result.' This hybrid approach aims to deliver higher throughput while achieving the lowest latency, potentially offering better performance for large language model workloads than existing alternatives.
Who are the founders of MatX and what is their background?
MatX was founded by Reiner Pope and Mike Gunter, both former Google semiconductor engineers. Pope worked on software for Google's chips and AI models, while Gunter was a hardware engineer for Google's tensor processing unit (TPU) chips. The two left Google in 2022 with the specific goal of creating a chip focused solely on running large language models, the technology that underpins AI chatbots.
Sources:
[1]