A startup trying to beat Nvidia at the inference game just stepped into the spotlight. Etched, founded in 2022, has announced that it has booked $1 billion in contract orders for its AI chip systems, raised $800 million in total funding, and reached a $5 billion valuation after a $500 million round that closed quietly in December.
The company’s product is not just a chip. Etched sells what it calls “frontier inference clusters”: complete systems that bundle its chips with custom-designed racks and software. The pitch is that these systems run AI inference faster, more cheaply, and with better power efficiency than anything else on the market. TSMC manufactured the first version of the chip earlier this year, and Etched says it is currently testing those systems with customers.
Inference is the process that happens every time someone submits a prompt to an AI model. It is, right now, the single biggest cost and bottleneck for AI companies trying to serve users at scale. That is exactly why chips optimized specifically for inference are attracting serious attention from investors and big tech alike.
Etched’s investor list reflects just how seriously people are taking the inference problem. The cap table includes:
- VentureTech Alliance, Jane Street, Hudson River Trading, Two Sigma, and Ribbit Capital
- AI researchers Andrej Karpathy, Geoffrey Hinton, and Fei-Fei Li
- Arthur Mensch and Scott Wu as angel investors
- Billionaires Stanley Druckenmiller and Peter Thiel
CEO Gavin Uberti and president Robert Wachen, both Harvard dropouts and Thiel fellows, have been talking publicly about their chip plans since 2024. The “coming out of stealth” framing in Tuesday’s press release is a bit generous: the company has had media coverage and investor interest for some time. What is new is the scale. By 2024, Etched had raised just over $125 million. Now it has $800 million and a billion dollars in orders.
The contrast with the company’s early days is stark. On Patrick O’Shaughnessy’s “Invest Like the Best” podcast, the founders described 2023 as a near-death experience. They had a 30-page memo arguing that AI would eventually need specialized chips rather than general-purpose GPUs, and every major investor they pitched passed. The company was reportedly running month-to-month, close to running out of money entirely.
That was then. The current environment for AI chip startups looks nothing like 2023. Cerebras pulled off the first major chip-focused IPO of the year. Groq just closed a $650 million round. Amazon, Google, and Microsoft all build their own in-house AI chips. OpenAI recently announced its first custom chip, built with Broadcom. The race to solve inference at scale has every major player spending heavily, which gives a company like Etched a real opening if its hardware performs as claimed.
The key question now is whether Etched’s chips actually deliver what the company promises once real customer testing is complete. A billion dollars in orders signals strong confidence from buyers, but the inference chip market is getting crowded fast, and execution will matter as much as the fundraising numbers.
