Etched Raises $800M for AI Inference Chip and Systems

Emerging from stealth, Etched has announced a significant market entry, securing a substantial $800 million in funding and revealing over $1 billion in customer contracts. The company's breakthrough centers on a novel rack-scale inference system, designed to accelerate the computational demands of advanced artificial intelligence models. This substantial capital infusion, with the latest tranche closing at $500 million in December at a $5 billion post-money valuation, underscores strong investor confidence in Etched's vision for AI infrastructure.

The funding round saw participation from a formidable list of investors, including VentureTech Alliance, Peter Thiel, Jane Street, Hudson River Trading, Jump Trading, Two Sigma, Stripes, Ribbit Capital, Radical Ventures, Primary VC, and Positive Sum. Notable AI luminaries such as Andrej Karpathy, Geoffrey Hinton, Fei-Fei Li, Stanley Druckenmiller, Arthur Mensch, Scott Wu, and others also contributed, signaling a deep alignment with the future of AI development. A key element of this funding is a strategic foundry partnership with a leading global semiconductor manufacturer, enabling Etched to ramp up production of its specialized inference systems.

Etched's core innovation lies in its co-designed inference clusters, engineered from the ground up to address the escalating need for faster, more cost-effective, and abundant AI inference. The company's first-pass silicon success on TSMC's N4P process, achieved in under three years since its seed funding, marks a critical milestone. Their current rack-scale systems are already demonstrating performance with prominent models like DeepSeek, Qwen, Mamba, and Llama, with a design philosophy to support models of any size and complexity. This focus on optimizing the entire stack aims to dramatically reduce the cost and power consumption associated with running large AI models.

The urgency for such infrastructure is palpable. The global AI market is experiencing exponential growth, with AI chip revenue alone projected to reach hundreds of billions of dollars in the coming years. As AI applications permeate every sector, from healthcare to finance and autonomous systems, the bottleneck of inference processing becomes increasingly critical. Etched's co-founder and CEO, Gavin Uberti, highlighted this gap, stating, "The infrastructure needed to serve those models in a sustainable and economically viable way simply did not exist." The company's ambition is to enable "humanity-scale" AI inference.

With a team of over 400 professionals, many drawn from industry giants like NVIDIA, Broadcom, and Google's TPU program, Etched is prioritizing production and scalability. The company has established a Taiwan factory and a comprehensive San Jose facility encompassing data center, test house, and prototyping capabilities. This integrated approach allows for tight control over design, validation, and manufacturing. Co-founder Rob Wachen emphasized this operational focus, noting, "Production is the product. We are living through one of the largest infrastructure buildouts in history, and the companies that matter will be the ones that can translate technology into systems that can be manufactured, deployed, and operated at massive scale." The company is targeting gigawatt-scale production by 2027.

The market for AI inference hardware is highly competitive, with established players and numerous startups vying for dominance. However, Etched's substantial pre-launch traction, evidenced by significant customer contracts and a strong investor syndicate, positions it as a formidable contender. The ability to deliver high-throughput, low-latency inference at scale is paramount for the next wave of AI innovation, and Etched appears to be making a decisive move to capture this critical segment of the AI infrastructure market.