Episode summary
The Two Harvard Dropouts Who raised $800M to take on NVIDIA
Invest Like the Best
· · 1 hr 33 min · Hosts: Patrick O'Shaughnessy · Guests: Gavin Uberti, Robert Wachen
Summary
Etched’s founders argue that inference, not training, will be the dominant market in AI, and that whoever produces the most tokens at the best cost will be the most valuable company in the world. They say the industry has been underestimating how much hardware design, power delivery, and interconnect performance matter once models are deployed at scale. Their core claim is that older AI chips were built for a pre-ChatGPT world, so a new architecture can win by optimizing specifically for modern inference workloads rather than general-purpose flexibility.
They describe Etched as a full inference system, not just a chip: the rack, board, power, interconnect, and production process all matter. The founders say they bet on low-voltage inference to avoid thermal throttling and on cluster-scale memory to treat many chips as a single memory pool. They argue that better chip-to-chip communication and much lower latency can materially improve tokens per second and concurrency, especially as models and agents move toward giant-scale serving.
Much of the episode is also about how they built the company: extreme vertical integration, very fast decision-making, and recruiting a mix of top “legends” and scrappy operators. They say they sent engineers to Bangalore, ran day and night shifts, built mock thermal chips and FPGA simulations before silicon returned, and cut the time from chips arriving to running inference from months to 40 days. The founders present that pace as a strategic advantage and a necessity in a supply-constrained market.
Key takeaways
- The founders argue inference will become the biggest AI market, and that token production will determine which companies are most valuable.
- Etched’s technical bet is low-voltage inference, which they say lets them avoid thermal throttling and run at under half the voltage of other AI chips.
- Their second bet is cluster-scale memory, using a custom interconnect to cut chip-to-chip latency and raise effective bandwidth across a rack.
- They say the company built around extreme vertical integration, including its own racks, production flow, and test infrastructure, to move faster than incumbents.
- The founders claim this approach let them go from silicon return to running inference in 40 days, versus much longer for other AI chip launches.
Notable quotes
“We know inference is going to be the biggest market in the world. Whoever produces the most tokens is going to be the most valuable company in the world”
“We're not just building a chip. We're building a full inference solution. And that means a rack”
“we think all AI chips in the future are going to be low voltage chips”
“What we are able to do is add way, way more bandwidth and a much lower latency from chip to chip to our interconnects”
“we were able to do it in 40 days. And it's because by the time the chip came back, everything was boring”
How the episode opens
We know inference is going to be the biggest market in the world. Whoever produces the most tokens is going to be the most valuable company in the world. — We we had people quit. — Yeah. — That people literally were like, "This problem is unsolvable." And uh best of luck, guys.
You kind of have to be sick in the head to join our company. You're going to convince your family to move to San Jose for the semiconductor company run by two what 24 year olds now going against the biggest companies in the world with a design that they're saying is not going to be like 10% better, but it's going to be 10 10x better. we need $100 million to do this.
If we do this, we think this could be one of the most important companies of all time.
You're sitting in that moment, you think, holy crap, we can't afford this. And I began looking at like, huh, how hard is it to go back to Harvard? All right, gentlemen.
It's been three years or so, Gavin, since you and I last did this, which is nuts. And at the time, I was just wildly intrigued by your story and what you were going to build. I didn't know a lot about chips at the time.
I was considering investing in the company. And so, I was calling everyone I could conceive of that could give me an opinion or something. And at the time, basically, the consensus was these kinds of companies are not built by young people.
that the semis world uh the best companies are founded by 40 50-year-old people that have had a whole career's worth of experience, have learned all the problems, have done have shipped multiple chips.
From the episode’s captions: roughly the first two minutes, lightly cleaned. Keep watching on YouTube →
Topics
- ai chips
- inference hardware
- venture fundraising
- semiconductor strategy
- ai infrastructure
Companies and tickers mentioned
- NVDA
- Etched
- Nvidia
Watch the episode
Watch “The Two Harvard Dropouts Who raised $800M to take on NVIDIA” on YouTube
Summary, takeaways and quote selection by Octagon. Quotes are short excerpts; the full conversation is in the episode.
More from Invest Like the Best
Invest Like the Best, this episode and its artwork belong to Colossus | Investing & Business Podcasts. Octagon is not affiliated with the show; this page is an independent summary.
