Fahad Manzur
AI chip startup Etched has raised another $700 million at a $21 billion valuation, in a round led by quantitative trading firm Jane Street, TechCrunch reported on August 18, 2026. The new valuation is roughly double the $10.3 billion the company reached in a round reported in late July 2026, meaning Etched has doubled its paper value in under a month. According to TechCrunch, the company has booked more than $1 billion in orders and has started shipping chips to customers.
What Etched actually builds
Etched makes server racks filled with processors designed specifically for inference, which is the computing work that happens when an AI model responds to a user query. That is a deliberate narrowing. Most AI hardware sold today is general purpose enough to handle both training a model and running it. Etched bet that inference would eventually dwarf training in total compute demand, and that a chip built only for that job could beat a flexible one on speed and cost per response.
For a long time that was a contrarian position, and the company drew plenty of skeptics. The reporting this week suggests the bet is now producing revenue rather than just enthusiasm, since the orders and shipments are what changed between the July round and this one.
Why Jane Street leading matters
The detail that stands out in this round is who led it. Jane Street is not a traditional venture investor chasing AI exposure. According to TechCrunch, the firm installed Etched’s first shipped AI cluster system and was impressed enough by it to lead the round afterward. A customer who tests hardware in production and then writes the largest cheque is a stronger signal than a generalist fund taking a position on a narrative.
Etched’s investor list also includes Kleiner Perkins, Sequoia Capital, Andreessen Horowitz, Peter Thiel, Tiger Global, Bain Capital Ventures, Neo, Stripes, Primary, Positive Sum, Diffusion, Argo and Blackstone, per the same report. That is an unusually crowded cap table for a hardware company, which reflects how much capital is currently chasing anything that reduces the cost of running AI models.
The context around the number
The $21 billion figure is worth putting next to a comparable transaction. TechCrunch notes it tops the $20 billion Nvidia paid in December to license technology and hire the leadership of rival inference chip startup Groq. In other words, the market is now valuing an independent inference chip company above what the dominant player paid to absorb a competitor’s technology and team.
An earlier TechCrunch report from July 2026 covered the $300 million round that took Etched to $10.3 billion, which gives a clear picture of how fast the repricing happened.
Why this matters if you work in tech
Most readers will never buy an inference chip. The reason to track this story anyway is that the price of running AI models is the single biggest input cost behind almost every AI tool people use for work, from writing assistants to image generators to customer support bots. When purpose-built inference hardware gets cheaper and faster, that cost falls, and the tools built on top of it get cheaper or better or both.
There is a second reason. A serious independent challenger in inference silicon puts pressure on the incumbent supplier that most AI companies currently depend on. Competition at the chip layer usually shows up months later as competition at the API pricing layer, and that is where freelancers, agencies and small businesses actually feel it.
The caution is the obvious one. A $21 billion valuation is a bet on future orders, not a verdict on delivered performance. Etched has started shipping, but it has not yet operated at the scale its valuation implies. The next real test is whether the $1 billion order book converts into deployed, working clusters at customers other than the one that just led its round.
