By Ljubisa Bajic Many believe AI is the real deal. In narrow domains, it already surpasses human performance. Used well, it is an unprecedented...

Hacker News is a community-driven platform for sharing and discussing technology news, startups, and programming-related topics. Through user submissions and comments, Hacker News offers insights into emerging technology trends, industry developments, and entrepreneurial ventures. Readers can participate in discussions, share their insights, and stay informed about the latest advancements in technology and innovation.

Hacker News

Taalas, a hardware startup, argues that AI's path to ubiquity is blocked by high latency and cost, drawing a parallel to how ENIAC gave way to practical computing. Their solution: custom silicon that hard-wires specific AI models directly into chips, merging storage and compute on a single die to eliminate the memory-compute bottleneck. Their first product is a silicon-hardwired Llama 3.1 8B achieving 17,000 tokens/sec per user — roughly 10x faster than state-of-the-art GPU solutions — at 20x lower build cost and 10x lower power consumption. Built by a 24-person team spending only $30M, the platform is now available as a chatbot demo and inference API. A mid-sized reasoning LLM on HC1 is expected in spring, with a frontier model on their second-gen HC2 platform planned for winter.

The path to ubiquitous AI