Etched Secures $5 Billion Valuation as AI Inference Race Intensifies

The startup confirmed it has successfully produced its first chips through TSMC and is currently testing "frontier inference clusters." These systems bundle custom-designed hardware and software specifically engineered to optimize inference, the process of running AI models after training. By targeting this specific bottleneck, Etched aims to undercut the power and cost requirements of general-purpose GPUs.

This growth follows an unannounced $500 million funding round closed in December, bringing the company’s total capital raised to $800 million. The investor roster includes VentureTech Alliance, Jane Street, and Hudson River Trading, alongside high-profile figures such as Geoffrey Hinton, Andrej Karpathy, and Peter Thiel. Founders Gavin Uberti and Robert Wachen, both Thiel fellows who dropped out of Harvard to launch the business, have pivoted from a period of near-insolvency in 2023 to a market hungry for specialized silicon. The current landscape is crowded with competition, as hyperscalers like Google and Microsoft develop internal alternatives, while rivals such as Groq and Cerebras aggressively capture market share in the race to accelerate AI deployment.