Etched Doubles Valuation to $21 Billion in a Month

Etched, a startup developing specialized chips for AI models, has doubled its valuation to $21 billion in just one month. The company announced that Jane Street, a global quantitative trading firm, has installed the first AI cluster system Etched has shipped and was so impressed with its performance that it led the next major funding round. This news was reported by TechCrunch on August 18, 2026. The new valuation reflects a surge in investor confidence in Etched's approach of building application specific integrated circuits (ASICs) for large language model inference, as opposed to the general purpose GPUs that have dominated the market. Etched is previously known for its chip called Sohu, which is specifically designed to run transformer architectures, the foundation of generative AI models like GPT.

This approach promises significantly higher efficiency and speed compared to GPUs for inference workloads, but it carries risks because it only supports one type of model architecture. Jane Street is known as one of the most intensive users of high performance computing technology in the world. The company's decision to install Etched's system in their infrastructure is a strong signal that specialized AI chips are starting to be accepted in real production environments, not just research labs. The funding round led by Jane Street marks a new chapter for Etched, which had previously raised substantial funds from technology investors. With a valuation of $21 billion, Etched now ranks among the most valuable AI startups, although its revenue has not been widely disclosed.

This valuation surge comes amid intense competition in the AI chip market, which is dominated by Nvidia. Several startups, including Etched, Cerebras, and Groq, are attempting to challenge Nvidia's dominance with different architectural approaches, ranging from wafer scale engines to specialized inference chips. Etched's success in attracting Jane Street as both a customer and investor shows that the market is beginning to see real value in specialized AI chips. The cluster system installed at Jane Street is Etched's first commercial delivery, making it a significant milestone in transitioning from chip design to a product operating in the field. TechCrunch reported that Jane Street was so impressed with the system that it decided to lead the funding round.

This creates a unique dynamic where the first customer is also the lead investor, providing a strong signal about product quality to other potential investors. For the broader AI industry, the delivery of Etched's cluster system to Jane Street marks a concrete step toward diversifying AI computing infrastructure. Until now, almost all training and inference of large models have run on Nvidia GPUs, and the presence of a proven alternative in production environments could reshape the competitive landscape. Etched has not disclosed technical details of the cluster system installed at Jane Street, including the number of chips and the performance levels achieved. However, Jane Street's decision to invest after using the product indicates satisfactory results in real world testing.

The $21 billion valuation places Etched on par with other major AI startups. The company was founded by Gavin Uberti and Chris Zhu, who previously worked in the AI industry and decided to build chips specifically optimized for transformers. Etched's approach has clear trade offs. The Sohu chip is designed only for transformer architectures, so if the industry shifts to fundamentally different new model architectures, the chip could lose relevance. However, as long as transformers remain the dominant standard, the efficiency advantage could be a major differentiator. Jane Street itself is not a company that typically leads investments in chip startups. Its primary focus is quantitative trading and liquidity provision across global financial markets.

Their decision to lead Etched's funding round demonstrates deep conviction in the technology. This new funding round comes just one month after Etched's valuation was around style="background-color: #ffffff;"0.5 billion. The doubling in such a short time reflects very strong market momentum for AI infrastructure startups amid the explosion in demand for AI computing. Going forward, Etched needs to prove that the cluster system installed at Jane Street can be replicated for other customers. Production capacity, supply chain, and software support are operational challenges that must be addressed to convert this high valuation into real revenue. The presence of Jane Street as an investor could also open doors for adoption in the financial sector, which requires very high speed AI inference.

Algorithmic trading and real time risk management are use cases that demand low latency, an area where Etched's specialized chips could excel. With this new valuation, Etched has the capital to expand its team, increase production, and build a software ecosystem. Competition with Nvidia, which has the advantage of CUDA and a mature ecosystem, remains a significant hurdle, but the adoption evidence from Jane Street provides a strong foundation of credibility. Industry observers will continue to monitor Etched's progress as an indicator of whether specialized AI chips can become a serious alternative to GPUs on a commercial scale. The first system delivery and a customer led funding round give the startup rare momentum in this segment.