Etched closes $300M Series C at $10.3B valuation, doubling in seven months as inference chip orders hit $1B

Etched, the AI chip startup founded by three Harvard dropouts in 2022, closed a $300 million Series C round on July 23 at a $10.3 billion valuation. Sequoia led the round, with Andreessen Horowitz, SK Hynix, Jane Street, and Diffusion Capital joining alongside earlier investors, according to TechCrunch. The round reportedly carries the highest valuation ever recorded for a Sequoia-led Series C.

The figure matters beyond the headline number. In December 2025, Etched was valued at $5 billion when it closed a $500 million raise. Seven months later, it has doubled. That pace, combined with $1 billion in orders already booked and first full systems in active client testing, signals that Etched has moved from a credible concept to a real infrastructure vendor, one that procurement and IT operations teams should now be actively evaluating.

Inference without GPUs: what Etched is actually selling

Etched's core bet is specialization. While general-purpose GPUs dominate AI compute today, Etched has built chips designed specifically to accelerate the inference step, running a trained model to generate outputs, rather than training. According to VentureCapital.com, the company has also developed two new components alongside the chips to further accelerate the inference process.

TechCrunch reports that Etched claims its hardware can speed up inference on any AI model with no GPUs required. That is a direct procurement implication: if the claim holds under production conditions, enterprise teams running inference-heavy workloads, customer-facing AI, large-scale document processing, real-time decisioning, have a potential GPU-alternative to benchmark against their existing stack.

The company says it successfully manufactured its homegrown chips last month and that first full systems are now being tested by clients. That sequencing matters for enterprise buyers: the hardware exists, it is not vaporware, and the company has enough demand to be scaling operations rather than chasing a first customer.

The investor roster signals long-term conviction

Beyond Sequoia and Andreessen Horowitz, the participation of SK Hynix, one of the world's largest memory chip manufacturers, is operationally significant. SK Hynix's involvement suggests Etched's memory architecture is credible enough to attract a strategic partner with deep semiconductor manufacturing expertise, not just financial capital.

Individual backers include Peter Thiel, Andrej Karpathy, Dylan Field, and Amjad Masad, according to both VentureCapital.com and TechCrunch. Karpathy, a former OpenAI and Tesla AI lead, is a particularly notable technical validator given that his endorsement carries weight with enterprise buyers assessing chip credibility.

Jane Street's participation adds a different signal: the quantitative trading firm runs some of the most inference-intensive compute workloads in financial services. Its investment suggests Etched's performance claims have been stress-tested against a real, demanding use case, not just benchmark conditions.

What this means for enterprise AI infrastructure teams

For IT leaders and infrastructure architects, the immediate question is not whether to invest in Etched but whether to put it on the evaluation list. With $1 billion in orders booked and live client testing underway, the company has enough scale to support a meaningful pilot or proof-of-concept conversation. Teams currently locked into GPU procurement cycles should watch Etched's pricing and performance benchmarks as they emerge from client testing.

Etched's funding also underscores a broader shift in enterprise AI infrastructure: the market is no longer treating GPU availability as a fixed constraint to work around. Specialized inference chips are attracting serious institutional capital precisely because GPU costs and lead times remain pain points for scaled AI deployment. Etched is not the only player here, but a $10.3 billion valuation backed by Sequoia and SK Hynix gives it a credibility floor that newer entrants cannot yet match.

The company is currently scaling up operations, according to VentureCapital.com. The next concrete milestone to watch is when client testing results move from private to published: performance numbers from production deployments will be the real arbiter of whether Etched's inference-without-GPU claim holds at enterprise scale.