TLDRocket
Sign in

Etched’s valuation doubles to $21B in a month

TechCrunch Julie Bort

Etched just raised $700M and its valuation shot to $21B, up from $10.3B a month ago. Jane Street led the round after buying and testing Etched's AI chips itself.

Based on reporting by TechCrunch, Julie Bort — read the original for the full story.

Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error

Etched's valuation didn't just climb this month, it doubled. The chip startup was worth $5 billion back in December, jumped to $10.3 billion after a $300 million Series C in July, and now sits at $21 billion following a fresh $700 million raise led by Jane Street. That's roughly $11 billion added in a matter of weeks, which even in a market where AI numbers routinely defy gravity is a startling pace.

What's driving the enthusiasm, according to co-founder and COO Robert Wachen, is Etched's approach to inference — the work a system does after someone actually submits a prompt. Wachen breaks that process into two stages: prefill, where the system parses the prompt and its context, and decode, where it actually generates the output tokens a user reads. Etched built separate hardware for each stage rather than treating inference as one generic workload. Its prefill chip runs at low voltage so it can cram in more transistors without cooking itself, letting it chew through tokens faster. For decode, Etched built what it calls cluster-scale memory, a new memory and interconnect setup that lets a bunch of chips share a single fast, low-latency memory pool.

The pitch is simple: faster inference, cheaper inference. Etched sells this as complete packages it calls frontier inference clusters, its answer to what Nvidia markets as AI factories. Jane Street isn't just an investor here, either. The quant firm said in the funding announcement that it tested the chip, liked what it saw, and now has its own rack running in its datacenter — a pretty concrete form of validation for a hardware startup.

Etched is still shaking off a misconception baked into its own name. Early on, the idea was that each chip would be etched to run one specific frontier model, hence the name. That's not how it works anymore. The systems Etched ships today can run whatever frontier model a customer throws at them, which matters a lot given how quickly the model landscape shifts underneath any hardware bet.

The investor list backing this round reads like a checklist of the industry's usual suspects — Kleiner Perkins, Sequoia, Andreessen Horowitz, Peter Thiel, Tiger Global, Bain Capital Ventures, Blackstone, and several others alongside Jane Street. That breadth of backing, paired with a valuation that's nearly quadrupled since December, says a lot about how badly the market wants an alternative to Nvidia's grip on inference hardware.

My take — AI-written commentary, not fact-checked reporting

A company doubling its valuation in a single month on the strength of one customer actually running a rack is either a sign that inference-specific chips are about to matter enormously, or a sign that AI investors have stopped pretending valuations track anything but vibes. Probably both. The Jane Street endorsement is the real story here — quant funds don't buy hardware to be polite, so if the chips genuinely cut latency and cost, Nvidia should be paying attention rather than just renaming its racks.

Read more about this at: TechCrunch

Related stories

The daily briefing

Every AI story that matters, in your inbox by 8am.

TLDRocket reads all relevant sources, removes duplicate coverage, and summarises the day in two minutes. Follow companies and topics for alerts, or get the briefing in Slack. Free, no spam, unsubscribe anytime.