SAN JOSE, Calif., July 23, 2026 (GLOBE NEWSWIRE) -- Etched, the company building frontier inference clusters, today announced $300M in new financing at a $10.3B valuation less than one month after emerging from stealth. The latest round was led by Sequoia, with participation from a16z, Jane Street, Diffusion, and SK Hynix and represents the highest valuation ever for a Sequoia-led Series C. The company is using this latest round of funding to aggressively scale production and customer deployments. In addition to the Taiwan factory established earlier in the year, Etched also recently opened a new 80,000 square-foot facility 15 minutes from their office to further expedite production and prototyping. They expect to use the 10 MW Milpitas site to bring up an NPI lab, an in-house SMT line, and expand deployments.
Etched was founded on the belief that inference is the defining workload of the AI era. AI will not deliver on its economic and societal promise until inference becomes dramatically faster, cheaper, and more plentiful than the status quo can provide. Etched was built to close the gap by improving the economics and power efficiency of AI compute so that inference can operate at the scale of humanity, not just the scale of today's data centers.
Etched’s hardware is architecture-agnostic, supporting models of any shape and arbitrarily large parameter sizes. Today, the company’s inference clusters are running large MoE models including DeepSeek and Qwen as well as non-transformer designs like Mamba, a state space model with a fundamentally different architecture from transformers. For customers, this means a single system that can serve whatever the frontier produces next without needing to bet their infrastructure on a single model design winning.
“Inference is on the path to becoming the largest market in the world, and purpose-built compute will power the majority of the world’s inference. We have spent years studying the space and waiting for a company with the technical ambition, operational excellence, and execution required to define it. Etched is that company," said Sonya Huang, Partner at Sequoia Capital. "What the team has accomplished in less than three years is extremely rare in the semiconductor world and a testament to the company and the culture they have built. We could not be more thrilled to partner with them as they build the defining company in inference hardware.”
Etched is developing two novel computing technologies at the core of its rack-scale system, each targeting one of the two bottlenecks that cap how much intelligence the world can afford to run: power and memory. Together, these innovations deliver more tokens per dollar, more tokens per watt, and much lower latency for frontier AI workloads.
- Low Voltage Inference (LVI) is a new paradigm in AI computing that unlocks dramatically better power and cost efficiency. Thermal limitations have traditionally constrained compute performance, throttling clock speeds before chips can reach peak FLOPs. LVI enables significantly higher FLOP density within the same power envelope as existing hardware, using deep hardware and systems co-design to deliver best-in-class throughput.
- Cluster Scale Memory (CSM) is a hybrid SRAM/HBM memory subsystem designed around an entire cluster rather than a single chip. CSM enables dramatically faster memory access by creating a massive shared SRAM pool across the entire scale-up domain, connected by a proprietary ultra-low-latency, high-bandwidth interconnect that delivers cluster-wide memory performance. CSM improves latency while avoiding the cost, reliability, yield, thermal, and compute tradeoffs of SRAM-only and 3D DRAM chips.
Demand for Etched's inference systems continues to outpace supply as customers move from evaluation to production deployment. To meet this demand, Etched is rapidly expanding its team, now 400 people, with engineers from NVIDIA, Broadcom, Google TPU, SK Hynix, and HFT quant trading firms. The team is in person and works out of their San Jose office which also has an NPI prototyping facility and 2 MW datacenter running Etched hardware 24/7.
"Now is the time to be aggressive. Our chips work, people want them, and it’s time to ship,” said Gavin Uberti, co-founder and CEO of Etched. “The infrastructure required to serve frontier AI sustainably and economically was never going to come from incremental improvements to existing hardware. This round reflects a growing industry conviction that the challenge demands a new entrant willing to rebuild the stack from first principles and deliver frontier inference without the trade-offs between speed, cost, and scale that have hampered this market until now."
“We tend not to celebrate fundraises. We have a lot of work to do to get to Gigawatt scale. We’re excited to partner with the best AI infrastructure investors in the world to get there faster,” said Rob Wachen, co-founder and President of Etched. “The team is working around the clock with our early customers to bring our first product to life. We’re super excited.”
About Etched
Etched builds frontier inference clusters designed to make AI inference dramatically faster, cheaper, and more abundant. Through deep co-design across the entire stack, Etched's rack-scale systems deliver best-in-class throughput and latency. Etched has raised over $1B from investors including Sequoia, a16z, SK Hynix, VentureTech Alliance, Peter Thiel, Jane Street, HRT, Jump Trading, Two Sigma, Ribbit Capital, Stripes, Radical Ventures, Primary, and Positive Sum. To learn more, visit etched.com.
Contact:
media@etched.com
