AI inference startup Etched raised $800M from investors including Jane Street and a TSMC-linked venture firm, and says it has signed sales contracts worth $1B
Nvidia rival says it has raised $800 million and has $1 billion in contracts. — AI chip startup Etched said it has raised $800 million …
BloombergDina Bass
Context & Ripple Effects
Etched’s financing has escalated sharply from its 2024 Series A for Sohu, a transformer-only inference chip, to a reported $500 million round in January and now an $800 million raise. The company’s claimed $1 billion in sales contracts adds a commercial marker to what had primarily been a funding story.
The move sits alongside large financings for other Nvidia challengers, including MatX, whose round was led by Jane Street, and inference-focused Baseten. Jane Street’s appearance across these deals underscores investors’ interest in the inference layer as a distinct AI infrastructure opportunity.
First-order effects
Etched gains substantial capital to pursue delivery against its reported contracts and to build out a specialized alternative to Nvidia for transformer-model inference.
The backing brings Jane Street and a TSMC-linked venture firm into Etched’s investor base, strengthening the startup’s financing and semiconductor-ecosystem credibility.
Second-order effects
Etched’s reported contract pipeline raises the execution bar for competing inference-chip startups such as MatX: fundraising alone will be less differentiating than converting demand into deployed systems.
Nvidia faces a more credibly funded specialist challenger in a workload segment where Etched is explicitly narrowing its product scope to transformer inference.
Third-order effects
If specialized-chip vendors can turn contracted demand into shipped deployments, AI infrastructure may become more segmented between broad-purpose accelerators and workload-specific inference hardware.
The concentration of large rounds among inference companies suggests capital is increasingly underwriting alternatives to incumbent AI hardware, though manufacturing and customer-delivery execution will determine whether that funding translates into durable competition.
The trend: AI infrastructure investment is broadening from general-purpose training hardware toward heavily funded, specialized inference platforms designed to challenge Nvidia in specific workloads.
Every Ribbit investment starts with one question: can we explain the “why” on a napkin? As @Etched announces its latest funding round and shares more of what they've built, here's what our Etched napkin says. @UbertiGavin @robertwachen [image]
Nvidia is worth $5T because the world realized inference is the new oil. If Etched's chips are meaningfully better than Nvidia's for inference... What is Etched worth? This is a worldview-shattering event.
Chips are a substrate for computation that the layer of intelligence we're all building relies on. Like every layer of the stack, it requires dedication, hard work, taking risks and innovating. Etched Sohu has all those in spades and I'm impressed every time I hear about
personal update: I joined Etched. I first met @robertwachen in early 2023 when I was a @neo scholar and he was a scholar finalist, one of my several interviewees. I still remember our first zoom. Within two minutes of meeting I knew Rob was not an ordinary 19 year old Harvard
Etched is coming out of stealth with $800M raised, $1B+ in customer contracts, first racks shipping this summer, and claims of SOTA inference throughput, latency, and power efficiency. But holy, look who backed the funding. The who-is-who if AI reseracher and VC. [image]
It's wild how quickly Etched designed and got the chips out, all within 2 years. They went deep, hardcoding attention into silicon and getting very high MFU. This kind of hardware tailored made for LLM inference is soon gonna bring cost of intelligence down 10x
Etched is going to be the first chip to market that was designed after ChatGPT. Sohu is a full stack solution for serving frontier inference fast and cheap. Etched is coming to market with the perfect product at the perfect time. I suspect frontier inference is going to become
Three years ago this was a handful of us and a bet that felt obvious to us and crazy to most: the world was going to need vastly more inference than anyone was building for, and the systems to serve it didn't exist yet. Today it's 400+ of the best engineers I've ever worked
Bringing our first rack to life has been nothing short of exhilarating and grueling. @UbertiGavin @czhu1729 and I hibernated in San Jose for three years building the team, solving thousands of problems, and convincing the world to believe in us. I'm excited to finally start
Introducing Low-Voltage Inference (LVI) for high throughput workloads. Today, AI chips can't scale FLOPs without thermal throttling. As FLOPs utilization increases, AI chips draw more power and downregulate clock speed. This often results in sustained inference throughput under […
Our inference systems are built to push the entire pareto curve on frontier models, including many-trillion parameter MoEs, long context, and agentic workloads. We've co-designed new chips, packages, PCBs, cold plates, interconnects, & more. Today, we're sharing two
We're a team of 400+ engineers from NVIDIA, TPUs, Broadcom, SK Hynix, TSMC, & more. We're backed by Jane Street, HRT, Two Sigma, and Jump, with strategic investment from VentureTech Alliance. We're excited to deepen our partnership with the world's leading fab. Our Series B was
We're coming out of stealth. We've built our first racks after a successful A0 tapeout, $1B+ in customer contracts, and $800m raised. Early customer tests show us achieving SOTA throughput, latency, and power efficiency on inference workloads. Our first racks ship this summer. [i…
Three years ago, two Harvard dropouts set out to build a better AI chip than the largest companies in the world. Almost everyone I called at the time said it was impossible. Today, Etched (@Etched) comes out of stealth with $800M total raised, $1B in signed customer contracts, [v…
Breaking news for people who want to look hot, be young and not die. A few years ago, two college dropouts told me they could accelerate longevity by building a faster AI chip. I invested, and they just pulled it off. What it means: > 10x more throughput (tokens per