Etched closed a $300 million Series C at a $10.3 billion valuation, led by Sequoia with participation from Andreessen Horowitz, SK Hynix, Jane Street and Diffusion Capital. The financing makes the semiconductor startup one of the most highly valued private challengers to Nvidia in AI inference, the computation used to run trained models and generate answers.
The valuation has doubled in roughly seven months. Etched raised $500 million at a $5 billion valuation in December and now describes the new transaction as the highest valuation achieved in a Sequoia-led Series C. The rapid step-up reflects investor demand for alternatives to incumbent AI hardware, but it also raises the performance threshold that future revenue must support.
Commercial evidence is beginning to emerge. The company says it has manufactured its first chips, placed initial complete systems with clients for testing and booked $1 billion of orders. Bookings show demand but are not the same as recognised revenue or delivered equipment, making conversion from customer commitments to installed systems a central financial milestone.
Etched will use the proceeds to expand production and customer deployments. It employs about 400 people, operates a two-megawatt data centre and has opened an 80,000-square-foot, ten-megawatt facility in Milpitas near its San Jose headquarters for production and prototyping. Those fixed investments move the business from chip design toward industrial execution.
The product architecture targets the two stages of inference. Etched says its prefill chip processes prompts at lower voltage, reducing heat and allowing denser transistor use. For the decode stage that generates output tokens, it has developed cluster-scale memory and interconnect technology designed to pool memory across chips with low latency.
Management argues that the systems are not limited to one model family. It says they can run mixture-of-experts models such as DeepSeek and Qwen as well as non-transformer designs such as Mamba. That compatibility claim is strategically important because hardware tied too closely to one architecture could lose value if model design changes.
The principal risk is scale. Access to the hardware has so far been concentrated among investors and early customers, and private demonstrations do not replace independent performance data, manufacturing yields or repeatable delivery. Etched must secure components, produce full racks, meet reliability targets and compete with Nvidia and other specialised inference developers while demand is said to exceed supply.
The next tests can be measured directly: mass production from the Milpitas facility, conversion of the $1 billion order book into shipments and revenue, broader customer benchmarks and continued compatibility as models evolve. The $10.3 billion valuation prices in a successful transition from promising silicon to dependable infrastructure; the Series C finances that transition but does not complete it.
