Engraved, the AI chip startup founded by three Harvard dropouts in 2022 has closed a $300 million Series C funding round at a $10.3 billion valuation, co-founder and COO Robert Wachen tells TechCrunch.
The round was led by Sequoia, with Andreessen Horowitz, SK Hynix, Jane Street and Diffusion Capital also participating, along with other, older investors. Other supporters of the company include names like Peter Thiel, Andrej Karpathy, Dylan Field, Amjad Masad and others.
Etched was previously valued at $5 billion in December when it raised a $500 million round, doubling its valuation in about seven months. The company says this is the highest valuation ever for a Sequoia-led C Series. Last month, Etched announced that it had successfully manufactured its homegrown chips, that its first complete systems were being tested by customers and that it had already booked $1 billion worth of orders.
Etched was released at a time when the idea of building a chip specifically for AI models based on transformer technology (the architecture behind most modern AI systems, including ChatGPT and Claude) was is considered wild if not eccentric. The company still battles the perception that its products — which are sold as complete systems, not just chips — include chips designed to run only specific LLMs.
Not so, explains Wachen. The systems can run any AI model, including Mixture of Experts models such as DeepSeek and Qwen—an architecture that divides tasks into specialized submodels rather than relying on one large model—as well as transformerless designs such as Mamba, which is built on a different underlying architecture known as the state-space model. (Interestingly, the idea of etching parts of a particular AI model directly into silicon to boost performance is no longer considered far-fetched. Google is reportedly pursuing the same idea with Frozen v2 chip for Gemini.)
But Etched’s claim to fame today is that it designed two new elements from scratch to speed up inference — the calculation process that happens after a user submits a prompt.
“Inference is built in two stages,” Wachen says, “prefilling and decoding.” The “prefill phase” involves understanding the prompt, including the context. It is mathematical and computationally intensive. The “decode” phase generates the output tokens (the actual response that the user sees). It requires less computation but needs huge amounts of memory.
Etched has created a prefill chip that works “dramatically” faster, he promises, “running at a much lower voltage than any other AI chip. We call this low-voltage inference.’ The lower voltage produces less heat, which allows the chip to pack more transistors.
For the decoding process, Etched created a new type of memory and “interconnect technology that we call cluster-scale memory. It allows multiple chips to connect together and use shared memory storage at very, very fast, low latency,” he says. The result, Etched promises, is high speeds but lower costs.
Because the startup started before most of the tech world (except Nvidia) understood the specialized computing needs of artificial intelligence, founders CEO Gavin Uberti, Wachen, and Chris Zhu have faced many skeptics who continued to doubt even after the company announced that its first batch of silicon had been successfully manufactured by TSMC.
A lot of that comes from how few people had access to the systems. So far, access has been limited to investors and early customers. In fact, that’s how Etched landed her list of celebrity investors in the first place — by showing them private demonstrations in her office.
“Andrej Karpathy from Anthropic, Noam Brown from OpenAI, Geoffrey Hinton, as well as all the investors in the funding round — these are all people who have actually tried the hardware and are very excited about it,” says Wachen.
Still, it’s been a long, hard road with a lot to go before rack systems are mass produced and delivered. The trio famously dropped by Harvard to start Etched, not knowing how to raise cash (much less the loads they’d need) or how to hire.
“We had no idea how difficult it would be,” says Wachen. “I think we still have to be humble about what it takes to get to scale.”
Wachen remembers landing in the Bay Area after telling his parents he was dropping out of school to start a startup, without arranging an office or an apartment. He slept on the floor of a friend’s unfurnished house.
“I remember staying in my friend’s house that was going to be sold, using a towel as a blanket,” he laughs. The founders eventually built the servers needed to run the chip design tools in an early employee’s garage, and “every time he needed a reboot, he’d call his wife and she’d go and hit the reboot button.”
Today, there are 400 people buzzing in one office, and Etched operates a 2 megawatt data center there. And it just opened a new 80,000-square-foot, 10-MW facility in Milpitas, below its San Jose headquarters.
“We’re running tokens in our lab today, working with some of the biggest AI companies in the world,” he says.
Wachen also has a blanket now, and a mattress “and one more pillow. Multiple pillows,” he jokes.
Most importantly, he and his co-founders never let the naysayers stop them. “It’s come a long way. It’s a very, very different world. But I think when you really believe something is possible, and you just work at it for a long time, you can do it.”
Note: This story has been updated to include reference to the new 10 MW facility.
When you purchase through links in our articles, we may earn a small commission. This does not affect our editorial independence.
