AI chip startup Etched books $1B in orders, hits $5B valuation

On Tuesday, AI chip startup Etched issued a progress report revealing it has closed a $500 million funding round at a $5 billion post-money valuation. The transaction, which closed in December, brings the company’s total capital raised to date to $800 million. Alongside the funding, Etched announced it has booked $1 billion in contract orders for its systems, which are powered by chips manufactured by TSMC earlier this year.

The startup is currently testing its first product with customers. The systems, which Etched calls “frontier inference clusters,” are bundles that include its chips, custom-designed racks, and software. According to Etched, these systems are designed to help frontier models run inference faster, more cheaply, and with better power efficiency than rivals. Inference is the process of running an AI model after a user submits a prompt, which currently represents the largest cost center and operational bottleneck for artificial intelligence companies operating at scale.

The $500 million round was led by Stripes. The company’s investor base includes:

• Institutional investors: VentureTech Alliance, Jane Street, Hudson River Trading, Two Sigma, and Ribbit Capital.
• Individual and angel investors: Andrej Karpathy, Geoffrey Hinton, Fei-Fei Li, Arthur Mensch, Scott Wu, Stanley Druckenmiller, and Peter Thiel.

Etched is entering a highly competitive market for AI hardware. It faces competition from established chip companies like Nvidia, as well as specialized startups like Cerebras and Groq, the latter of which has raised $650 million. Additionally, hyperscalers—large cloud providers like Amazon, Google, and Microsoft—are building their own in-house AI chips, while OpenAI recently announced its first custom chip built by Broadcom.

Founded in 2022 by CEO Gavin Uberti and president Robert Wachen, Etched initially struggled to secure backing. During 2023, the founders struggled to get investors interested in their thesis that AI would eventually require specialized chips rather than general-purpose graphics processing units (GPUs), leaving the company reportedly operating month-to-month and close to running out of cash. However, by 2024, the founders began talking to TechCrunch about their chip plans, and the company had raised more than $125 million as investor interest in specialized inference hardware intensified.

Why it matters

Etched is betting that specialized “frontier inference clusters” can outperform general-purpose GPUs, directly addressing the massive cost and efficiency bottlenecks currently facing AI companies running inference at scale.