Etched hits $5B valuation with $1B in orders for TSMC-made inference chips, raising $500M from top investors. The startup targets Nvidia's dominance.
Etched has booked $1 billion in contract orders for its transformer-inference chips, the startup announced Tuesday. The Nvidia competitor also revealed a $500 million round closed in December at a $5 billion post-money valuation, according to TechCrunch.
Key facts
- Etched raised $500M at $5B post-money valuation.
- Booked $1B in contract orders for inference clusters.
- TSMC manufactured the chip in early 2026.
- Total funding: $800M to date.
- Investors include Karpathy, Hinton, Fei-Fei Li.
The orders are for what Etched calls "frontier inference clusters" — full systems bundling its custom chips with racks and software, targeting the high-cost inference bottleneck for frontier AI models. TSMC manufactured the chip earlier this year, and Etched is now testing it with customers.
The $800 million total raised includes a cap table stacked with AI heavyweights: Andrej Karpathy, Geoffrey Hinton, Fei-Fei Li, Arthur Mensch, and Scott Wu, plus billionaires Stanley Druckenmiller and Peter Thiel. The round was led by Stripes, with participation from VentureTech Alliance, Jane Street, Hudson River Trading, and Two Sigma.
This is a dramatic reversal from 2023, when co-founders Gavin Uberti and Robert Wachen — both Harvard dropouts and Thiel fellows — struggled to get meetings. Per the founders, every major investor passed on a 30-page memo arguing AI would need specialized chips, not just GPUs. The company was operating month-to-month, near insolvency.
The inference chip market is now red-hot. Cerebras had the first breakout IPO of 2026. Groq raised $650 million. Hyperscalers Amazon, Google, and Microsoft build in-house chips. Even OpenAI just announced its first custom chip, built by Broadcom. Etched's bet — that transformer-specific silicon will beat general-purpose GPUs on cost and speed — now has $1 billion in customer validation.
But the orders are "under contract," not recognized revenue. The chips are still in testing. And Nvidia's Blackwell and Vera Rubin lines continue to dominate MLPerf benchmarks [as previously reported]. The real test comes when Etched's systems ship at scale and customers measure total cost of inference against Nvidia's installed base.
Key Takeaways
- Etched hits $5B valuation with $1B in orders for TSMC-made inference chips, raising $500M from top investors.
- The startup targets Nvidia's dominance.
What to watch
Watch for Etched's first customer deployment benchmarks — specifically latency and cost-per-token comparisons against Nvidia's Blackwell B200 and Vera Rubin NVL72 systems. Also track whether the $1B in contracted orders converts to recognized revenue by Q1 2027.
Source: techcrunch.com
Originally published on gentic.news
Top comments (0)