Technology news and discussion on TalkFuse.
AI chip startup Etched defies skeptics, hits $10.3B valuation from big-name investors
Etched, an AI chip startup founded by three Harvard dropouts in 2022, has closed a $300 million Series C funding round at a $10.3 billion valuation, doubling its worth in roughly seven months. The round, led by Sequoia with participation from Andreessen Horowitz, SK Hynix, Jane Street and Diffusion Capital, comes as the company shakes off scepticism that greeted its early bet on building chips tailored specifically for transformer-based AI models, an approach many initially dismissed as unworkable.
The company says the valuation marks the highest ever for a Sequoia-led Series C, up from $5 billion when it raised $500 million last December. Last month Etched confirmed it had successfully manufactured its chips via TSMC, with first full systems now being tested by clients and roughly $1 billion in orders already booked. Co-founder Robert Wachen said the systems, unlike earlier assumptions, can run any AI model, including Mixture of Experts designs such as DeepSeek and Qwen and non-transformer architectures like Mamba, thanks to newly designed prefill and decode components that boost inference speed while cutting costs. Access to the systems has so far been limited to investors and early customers, including high-profile figures such as Andrej Karpathy and Geoffrey Hinton, who saw private demonstrations before backing the company.
In short
- Etched raises $300m Series C at $10.3bn valuation, led by Sequoia
- Valuation has doubled in about seven months since December
- Chips already manufactured, tested by clients, $1bn in orders booked
Steelman: both sides, in good faith
The strongest fair case each way — we don't pick a winner.
The case for
Etched has demonstrated remarkable technical and commercial progress that justifies investor confidence: it has successfully manufactured advanced chips, secured $1 billion in pre-orders from credible customers, and attracted backing from the world's most selective venture capital firms alongside industry giants like SK Hynix. The company's founders have solved what sceptics claimed was impossible—building flexible chips that efficiently run multiple model architectures—which represents genuine innovation addressing a critical infrastructure bottleneck in AI deployment. The confidence of prominent researchers like Karpathy and Hinton, who have reviewed the technology firsthand, suggests this is not speculative enthusiasm but recognition of solved technical challenges with clear commercial applications.
The case against
A $10.3 billion valuation for a company with no proven revenue, relying entirely on pre-orders and investor backing, represents substantial risk in an extraordinarily crowded market. Numerous well-funded chip startups have failed despite impressive initial metrics, and Etched faces entrenched competition from Nvidia, major cloud providers and other well-capitalised rivals with massive scale advantages. The valuation's doubling in seven months on the basis of demonstrations and pre-orders rather than operational results suggests valuation driven more by investor enthusiasm for AI than by demonstrated business fundamentals, raising concerns about whether the company can deliver returns proportionate to its valuation amid intense competitive and technological pressures.
Explained simply
Etched is an artificial intelligence startup founded in 2022 by three Harvard dropouts. The company designs computer chips optimised specifically for running AI models, arguing that specialised hardware tailored to particular AI needs would be faster and more efficient than using general-purpose processors.
When Etched first announced this approach, many industry sceptics doubted whether it would work. Creating chips designed for specific AI technologies seemed risky, as critics argued the hardware would be inflexible and become outdated as AI models evolved, making the venture commercially unviable.
The company has now successfully manufactured its chips and demonstrated that they can run a range of different AI model types, not just those originally envisioned. This proof of concept is significant because it suggests that specialised hardware for AI may be more practical and valuable than industry sceptics initially believed.
AI Americas Art Business Celebrity Companies Culture Entertainment Technology World