Untether AI, a startup focused on AI inference chips, has shut down, with its engineering team moving to AMD. Despite their high-performance technology for small neural networks, Untether struggled to find a significant market niche amid the rise of large generative models like ChatGPT and StableDiffusion. Founded in 2018, their aim was to enable edge AI inference through distributed servers rather than deploying chips in individual devices. While they initially targeted areas like automotive use and smart city applications, they fell behind as the industry’s focus shifted to large language models (LLMs). Their efforts in vision inference and partnerships for autonomous vehicle systems could not compete with the growing importance of LLMs, which required different design considerations. Ultimately, Untether’s technology lacked the flexibility to adapt to this new landscape, highlighting the critical need for AI chip startups to anticipate future developments in AI.
Source link
