# Etched > Source: https://aiwiki.ai/wiki/etched > Updated: 2026-10-28 > Fact-checked: 2026-11-05 > Categories: AI Companies, AI Compute, AI Hardware > License: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/) - attribute to "AI Wiki (aiwiki.ai)" > Cite as: AI Wiki. "Etched." aiwiki.ai, 28 Oct 2026. https://aiwiki.ai/wiki/etched > From AI Wiki (https://aiwiki.ai), the free encyclopedia of artificial intelligence. Reuse freely with attribution. **Etched** is an American semiconductor startup building [Sohu](https://aiwiki.ai/wiki/sohu), the first [application-specific integrated circuit](https://aiwiki.ai/wiki/asic) (ASIC) designed to run only [transformer](https://aiwiki.ai/wiki/transformers) models, the architecture behind most [large language model](https://aiwiki.ai/wiki/large_language_model)s. Founded in 2022 by Harvard dropouts Gavin Uberti, Chris Zhu, and Robert Wachen, Etched hardwires the transformer directly into silicon, trading the flexibility of a general-purpose [GPU](https://aiwiki.ai/wiki/gpu) for raw [AI inference](https://aiwiki.ai/wiki/inference) speed: the company claims a single eight-chip Sohu server runs Meta's Llama 70B at over 500,000 [tokens per second](https://aiwiki.ai/wiki/tokens_per_second), equivalent to replacing roughly 160 [Nvidia](https://aiwiki.ai/wiki/nvidia) H100 GPUs.[1][2] The bet is explicit. "In 2022, we made a bet that transformers would take over the world," co-founder and chief executive Gavin Uberti said when the chip was unveiled in June 2024.[1] In January 2026 it raised about $500 million at a reported valuation of about $5 billion.[2][3][4] It then emerged from stealth in June 2026 with working silicon, raised a $300 million Series C at a $10.3 billion valuation in July and $700 million at a $21 billion valuation in August 2026, shipped its first rack to its first customer, Jane Street, and said it had more than $1 billion in customer contracts.[10][11][12]
Etched logo
Etched logo, from the official website etched.com. Trademark of its owner.
## Overview Etched launched with a single product, the Sohu chip, an inference [AI chip](https://aiwiki.ai/wiki/ai_chip) (ASIC) specialized for the transformer architecture that underlies most large language models, including those in the [GPT](https://aiwiki.ai/wiki/gpt) and [Llama](https://aiwiki.ai/wiki/llama) families; by 2026 it described its product as rack-scale "frontier inference clusters" built from its own chips.[1][13] The company's central argument is that a chip dedicated to one architecture can dedicate far more of its die area to transformer-specific compute than a general-purpose GPU, and that the resulting throughput and efficiency advantages justify the loss of flexibility. Etched describes Sohu in marketing as the basis for "the world's first transformer ASIC."[1][5] The original strategy was deliberately narrow. As Etched pitched Sohu in 2024, the chip could not run other model types, tying its commercial prospects to the continued dominance of the transformer;[8] by July 2026, however, the company described its hardware as architecture-agnostic.[10] The founders have framed this publicly as a calculated risk, arguing that the AI industry has converged on transformers to a degree that makes a specialized chip economically attractive. "We've hit a point in the evolution of AI where specialized chips that can perform better than general-purpose GPUs are inevitable, and the technical decision-makers of the world know this," Uberti said.[1][2] ## Who founded Etched and when? Etched was founded in 2022. Gavin Uberti and Chris Zhu both left [Harvard University](https://aiwiki.ai/wiki/harvard_university) to start the company; All three co-founders are [Thiel Fellowship](https://aiwiki.ai/wiki/thiel_fellowship) alumni. The third co-founder, Robert Wachen, is also a Harvard dropout. The early team recruited semiconductor veterans, most notably Mark Ross, chief technology officer of [Cypress Semiconductor](https://aiwiki.ai/wiki/cypress_semiconductor) in the early 2000s, who joined as chief architect and later became Etched's CTO, along with engineers drawn from companies such as [Broadcom](https://aiwiki.ai/wiki/broadcom).[1][4][6] The company was based in Cupertino, California at its Series A in mid-2024, when it employed roughly 35 people;[1][5] by 2026 it was headquartered in San Jose and had more than 400 staff.[3][11][13] ## How does the Sohu chip work? Sohu is an inference accelerator manufactured on [TSMC](https://aiwiki.ai/wiki/tsmc)'s 4-nanometer process. Each chip carries 144 GB of [HBM3E](https://aiwiki.ai/wiki/hbm3e) high-bandwidth memory, according to Etched.[9] Rather than memory bandwidth, Etched credited its claimed advantage to compute: it argued that modern transformer inference is compute-bound and said Sohu achieves over 90 percent FLOPS utilization, against about 30 percent on GPUs.[8] Etched markets Sohu around aggressive, company-supplied benchmarks. Its headline claim is that a single server containing eight Sohu chips can deliver about 500,000 tokens per second running Meta's Llama 70B model, which the company says is equivalent to replacing roughly 160 Nvidia H100 GPUs. Etched further asserts that Sohu is "an order of magnitude faster and cheaper than even Nvidia's next generation of Blackwell GB200 GPUs when running text, image and video transformers," a claim the company did not quantify further beyond saying one eight-chip Sohu server replaces 160 H100 GPUs.[1][8] These figures were company projections published before Etched had working silicon; no independent benchmark of them had been published as of November 2026, although first customer Jane Street said in August 2026 that it had tested the chip and was "pleased with the early results."[4][7][11] | Detail | Etched claim | | --- | --- | | Architecture | Transformer-only ASIC (inference) | | Process node | TSMC 4 nm | | Memory | 144 GB HBM3E per chip (~1.8x H100 SXM5 bandwidth) | | Headline throughput | ~500,000 tokens/sec on Llama 70B per 8-chip server | | GPU comparison | One Sohu server stated to replace ~160 [Nvidia H100](https://aiwiki.ai/wiki/nvidia_h100)s | | Claimed speedup | "An order of magnitude" faster and cheaper than Nvidia Blackwell GB200 | The same specialization that drives Sohu's claimed advantages also constrains it. Etched said in 2024 that because Sohu hardwires the transformer it could not run most traditional models, such as CNNs, RNNs, LSTMs, recommendation models, or older U-Net image models like Stable Diffusion 2, though it listed mixture-of-experts and other transformer variants as supported.[8][9] Some third-party analyses went further and argued it could not handle [mixture-of-experts](https://aiwiki.ai/wiki/mixture_of_experts) routing, [vision encoder](https://aiwiki.ai/wiki/vision_encoder)s, [diffusion models](https://aiwiki.ai/wiki/diffusion_model), or [state space models](https://aiwiki.ai/wiki/state_space_model);[7] by July 2026 Etched said its inference clusters were running large MoE models and the Mamba state space model.[10] ## Funding Etched raised its early capital in three principal stages, followed by two larger rounds in mid-2026. It closed a seed round of $5.4 million, reported by EE Times in 2023 at a valuation of about $34 million.[6] In June 2024 it announced a $120 million Series A co-led by Primary Venture Partners and Positive Sum Ventures, bringing total funding raised to about $125.4 million at the time. Named backers in that round included [Peter Thiel](https://aiwiki.ai/wiki/peter_thiel), [GitHub](https://aiwiki.ai/wiki/github) chief executive Thomas Dohmke, [Cruise](https://aiwiki.ai/wiki/cruise) and Bot Company co-founder Kyle Vogt, and Quora co-founder Charlie Cheever.[1] In January 2026 multiple outlets reported that Etched had raised approximately $500 million in a round led by the investment firm Stripes, with participation from Peter Thiel, Positive Sum, and Ribbit Capital. The round valued the company at about $5 billion and brought its total funding close to $1 billion.[3][4] In July 2026 Etched raised a $300 million Series C led by Sequoia, with a16z, Jane Street, Diffusion and SK Hynix, at a $10.3 billion valuation, and in August 2026 it raised $700 million led by Jane Street, with Kleiner Perkins, Sequoia, Andreessen Horowitz, Tiger Global and others, at a $21 billion valuation.[10][11][12] | Round | Amount | Date | Lead(s) | | --- | --- | --- | --- | | Seed | $5.4M | 2023 | (~$34M valuation) | | Series A | $120M | June 2024 | Primary Venture Partners, Positive Sum Ventures | | Later round | ~$500M | January 2026 | Stripes (~$5B valuation) | | Series C | $300M | July 2026 | Sequoia (~$10.3B valuation) | | Later round | $700M | August 2026 | Jane Street (~$21B valuation) | ## Why bet on transformers, and what are the risks? Etched's business is an explicit wager that the transformer will remain the standard AI architecture for years to come. A transformer-only ASIC offers no fallback if the field shifts toward a fundamentally different design; unlike a GPU, the silicon cannot be repurposed for another architecture. The founders have argued that the convergence of [frontier models](https://aiwiki.ai/wiki/frontier_models) onto the transformer makes this risk acceptable, and that the payoff is capabilities GPUs cannot reach: Uberti has said that "video generation, audio-to-audio modalities, robotics, and other future AI use cases will only be possible with a faster chip like Sohu."[1] Commentators have nonetheless noted that betting an entire company on architectural stability is unusual in a fast-moving field.[2] A second risk is execution and timing. Designing, taping out, and bringing a leading-edge accelerator to volume production is capital-intensive and slow, and Etched competes against Nvidia's entrenched ecosystem as well as other inference-focused challengers such as [Groq](https://aiwiki.ai/wiki/groq), [Cerebras](https://aiwiki.ai/wiki/cerebras), and [SambaNova](https://aiwiki.ai/wiki/sambanova). Until mid-2026 much of what Etched had published consisted of projected benchmarks rather than shipping silicon;[4][7] it delivered its first rack to a customer in July 2026, but independent performance data had not been published as of November 2026.[11] ## Is Sohu available yet? Etched partnered with TSMC's Emerging Businesses Group to manufacture the chip.[4] Reporting from April 2026 indicated that the chip was not yet publicly available for purchase or rental.[7] Later in 2026 Etched said its first A0 silicon had come back working from TSMC's N4P process, it shipped its first rack to Jane Street in July 2026, and it said it had begun production to fulfill more than $1 billion in customer contracts; its hardware is sold as rack-scale systems to customers rather than as a public cloud service.[11][12][13] At its 2024 Series A, Etched said unnamed customers had reserved tens of millions of dollars' worth of hardware and that it planned to offer early access through a developer cloud preview.[1] The $500 million raised in January 2026 was framed by the company as fuel for the continued push to bring Sohu to market against Nvidia.[2][3][4] ## References 1. Wiggers, Kyle. "Etched is building an AI chip that only runs transformer models." TechCrunch, June 25, 2024. https://techcrunch.com/2024/06/25/etched-is-building-an-ai-chip-that-only-runs-transformer-models/ 2. Bass, Dina. "AI Chip Startup Etched Raises $500 Million to Take on Nvidia." Bloomberg, January 13, 2026. https://www.bloomberg.com/news/articles/2026-01-13/ai-chip-startup-etched-raises-500-million-to-take-on-nvidia 3. "Etched Raises $500 Million to Challenge Nvidia." Yahoo Finance / Bloomberg, January 2026. https://finance.yahoo.com/news/etched-raises-500-million-challenge-115812959.html 4. "Harvard dropouts' Etched raises $500M at $5B valuation to challenge Nvidia." Tech Funding News, January 2026. https://techfundingnews.com/nvidia-rival-ai-chip-maker-etched-founded-by-harvard-dropouts-lands-500m-at-5b-valuation/ 5. "AI chip startup Etched.ai raises $120 million to take on Nvidia." Data Center Dynamics, June 2024. https://www.datacenterdynamics.com/en/news/ai-chip-startup-raises-120-million-to-take-on-nvidia/ 6. "Harvard Dropouts Raise $5 Million for LLM Accelerator." EE Times, 2023. https://www.eetimes.com/harvard-dropouts-raise-5-million-for-llm-accelerator/ 7. "Etched AI Sohu vs NVIDIA: Transformer ASIC vs General-Purpose GPU for LLM Inference (2026)." Spheron Blog, 2026. https://www.spheron.network/blog/etched-ai-sohu-vs-nvidia-transformer-asic-inference/ 8. Etched. "Etched is Making the Biggest Bet in AI." Company blog, June 25, 2024 (archived). https://web.archive.org/web/20250104005230/https://www.etched.com/announcing-etched 9. Etched. Company homepage, July 2024 (archived). https://web.archive.org/web/20240710214616/https://www.etched.com/ 10. "Etched Raises $300M at a $10.3 Billion Valuation to Scale Production of Frontier Scale Inference Hardware." Etched press release via StorageNewsletter, July 30, 2026. https://www.storagenewsletter.com/2026/07/30/etched-raises-300m-at-a-10-3-billion-valuation-to-scale-production-of-frontier-scale-inference-hardware/ 11. "Etched Raises $700 Million at a $21 Billion Valuation and Completes First Customer Delivery to Jane Street." Etched press release via StorageNewsletter, August 26, 2026. https://www.storagenewsletter.com/2026/08/26/etched-raises-700-million-at-a-21-billion-valuation-and-completes-first-customer-delivery-to-jane-street/ 12. Stanciuc, Ana-Maria. "Etched raises $700M at a $21B valuation led by Jane Street." The Next Web, August 18, 2026. https://thenextweb.com/news/etched-700m-series-d-21-billion-jane-street 13. Etched. Company homepage, accessed November 5, 2026. https://www.etched.com/