Tip Of AI
Discover what you can do with AI
Etched is a provider of AI hardware and software clusters designed to deliver best-in-class inference performance for large models. It co-designs chips, racks, software, and manufacturing methods to optimize throughput, latency, cost, and power efficiency in both prefill and decode workloads. It integrates features such as Low Voltage Inference to run math blocks at under half the voltage of most AI chips, enabling multiple times the floating point operations per second density without thermal throttling. The Cluster Scale Memory architecture provides an ultra-low-latency shared memory pool across chips, combining HBM and SRAM hybrid designs for high bandwidth and low decode latencies. Advanced packaging, power delivery networks, tiling and scheduling algorithms, and proprietary interconnects further enhance performance and scalability. The company has raised seven hundred million dollars at a twenty one billion dollar valuation and eight hundred million across four additional financings. Its team of over four hundred engineers includes experts from NVIDIA, Google, TSMC, and other leading semiconductor firms. Early customer tests report state of the art throughput, latency, and power efficiency, and the first rack scale products are set to ship in summer to fulfill over one billion dollars in contracts.