Etched is trying to turn AI inference into a dedicated hardware market, and its latest update gives the startup more numbers to put behind that pitch.
The company says TSMC successfully manufactured its chip earlier this year, and Etched is now testing its first product with customers. That product is not just a chip. It is a full system built around the chip, including racks and software designed for frontier model inference.
$1 billion in orders for full systems
Etched says it has already booked $1 billion in contract orders for its product. The company describes that product as full systems powered by its chips, rather than standalone components.
Those systems are called “frontier inference clusters.” In Etched’s framing, each cluster brings together the chips, custom-designed racks, and software. The purpose is to help frontier models run inference faster, more cheaply, and with better power efficiency than rival options, according to Etched’s claims.
That focus matters because inference is the step that happens after a user submits a prompt. For AI companies serving customers at scale, the source article describes inference as both the biggest bottleneck and the biggest cost center. A startup that says it can improve that part of the AI workflow is therefore operating in one of the most closely watched areas of the market.
The company is still in the testing phase with customers for this first product. That makes the $1 billion order figure important, but it also keeps the focus on execution. Etched has moved from a chip plan to manufactured silicon and customer testing, yet the broader question is how its systems perform as customers evaluate them.
A larger financing story
Etched also disclosed that it has raised a total of $800 million to date. The most recent financing was an unannounced $500 million round that closed in December, according to the company.
That round valued Etched at a $5 billion post-money valuation. Stripes led the $500 million round, and the startup’s investor list includes VentureTech Alliance, Jane Street, Hudson River Trading, Two Sigma, Ribbit Capital, and Stripes.
The cap table also includes angel investment from several high-profile AI figures: Andrej Karpathy, Geoffrey Hinton, Fei-Fei Li, Arthur Mensch, and Scott Wu. Billionaires Stanley Druckenmiller and Peter Thiel are also included.
Taken together, those names show that Etched has moved well beyond a narrow chip startup audience. It has attracted backers from venture, trading, AI research, and major technology investing circles. The reason is straightforward from the source: investors are paying attention to companies that promise to reduce the cost and difficulty of AI inference.
From difficult pitches to investor demand
Etched was founded in 2022 by CEO Gavin Uberti and president Robert Wachen. The two co-founders dropped out of Harvard and became Thiel fellows to build the company, as Uberti previously told TechCrunch.
Although the company’s press release framed the latest announcement as Etched “coming out of stealth,” the startup has not been entirely quiet. Uberti and Wachen have been talking to TechCrunch about their chip plans since 2024.
By 2024, Etched had already raised more than $125 million and was on investors’ radar. But the source describes a very different moment in 2023. On Patrick O’Shaughnessy’s “Invest Like the Best” podcast, the founders said they had struggled to get investor interest even after preparing a 30-page memo arguing that AI would eventually need specialized chips rather than only general-purpose GPUs.
Every major investor they pitched passed, according to the source. The company was reportedly operating month-to-month and close to running out of cash during those early days.
That contrast is central to the Etched story. In one phase, the startup was trying to convince investors that specialized AI chips would be necessary. In the current phase, the market is actively looking for AI chip companies that can speed up inference.
The AI chip race is widening
Etched is not alone in pursuing this opportunity. The source describes a funding environment in which investors are chasing AI-related companies, especially chip technology aimed at faster inference.
Competitor Cerebras had the first breakout IPO of the year. AI chip maker Groq just raised $650 million. Hyperscalers Amazon, Google, and Microsoft all build their own in-house AI chips. OpenAI also announced its first custom chip, built by Broadcom.
That activity shows why Etched’s progress report is about more than a single startup milestone. The company is entering a crowded and strategically important field where specialized hardware, full-stack systems, and power efficiency are all part of the competitive argument.
Etched’s claim is that its frontier inference clusters can make inference faster, cheaper, and more power efficient than rival approaches. The company now has manufactured silicon, customer testing underway, $1 billion in contract orders, $800 million in total funding, and a $5 billion post-money valuation.
The next stage is less about whether investors understand the pitch and more about how customers respond to the product. For Etched, the market has changed dramatically since its early fundraising struggles. Now the company has to show that its specialized systems can deliver on the inference problem that made investors interested in the first place.