Why Physical AI Is the Next Platform Shift
Eric Landau, co-founder of Encord, discusses his career trajectory from particle physics to quantitative trading to founding a data infrastructure company for AI. He explains how Encord pivoted from general vision AI to focus on Physical AI, which he believes will be a massive market as robotics proliferates across manufacturing, logistics, and autonomous vehicles.
Summary
Eric Landau traces his career through the evolution of AI approaches, illustrating how the field has shifted from hand-crafted feature engineering and domain expertise to the "bitter lesson" of simply scaling data and compute. After working in particle physics and then spending 10 years as a quantitative trader earning significant income, he quit during COVID in March 2020—coincidentally when markets experienced extreme volatility and his desk was making record profits—to co-found Encord with a belief that AI represented a generational technological paradigm shift.
Encord spent several years in the "desert" before finding product-market fit, which came gradually rather than as a sudden inflection point. The ChatGPT release in late 2022 accelerated market interest, though Landau emphasizes this was more about opening conversations than creating a binary threshold. The company's first major sales win with a company they didn't recognize signaled true product-market fit.
Landau discusses the challenges of building a sales organization as a technical founder, admitting to hiring and firing sales teams multiple times before success. He emphasizes the value of YC advice but acknowledges that founders often make mistakes despite knowing better, sometimes taking 9 months to act on gut feelings about bad hires.
Encord initially started in vision AI with a thesis that AI systems would naturally become multimodal, then expanded to serve the burgeoning Physical AI market—robots, autonomous vehicles, manufacturing, and logistics applications. The company operates a data collection facility in the Bay Area, manages petabytes of multimodal data (exceeding what was used to train GPT-4), and focuses on data curation, selection, annotation, and evaluation.
On competition and distribution, Landau notes that many early competitors pivoted or consolidated, and he actually welcomes strong competitors as they push the company to improve. Encord targets the inflection point where customers move from POC to production scale rather than serving early-stage builders.
The conversation concludes with Landau offering advice to founders: embrace the roller coaster nature of startups, recognize that mental state is more controllable than it appears, and find ways to enjoy the journey through inevitable ups and downs.
Key Insights
- The AI field has undergone a fundamental shift from the 'old way' of careful feature engineering and domain expertise to the 'bitter lesson' approach of simply scaling data and compute without hand-crafting systems.
- Encord experienced gradual product-market fit through continuous daily improvements rather than a sudden breakthrough moment, with the market starting to move toward them faster after ChatGPT rather than a binary threshold event.
- The company's genuine product-market fit signal came when they closed a sale from a company they didn't even know existed, meaning they were no longer personally involved in every sale.
- Landau credits three iterations of hiring sales teams before achieving success, and acknowledges keeping one poor hire for 9 months despite knowing it wasn't a fit, illustrating the difficulty of acting on gut feelings as a founder.
- Physical AI represents an enormous market opportunity because 80% of economic activity involves manipulating or moving things in the real world, with experts predicting more robots than people within 5-10 years.
Topics
Transcript
[0:00] Starting a company is a roller coaster. So, there's like low lows and there are high highs. And [music] my view is that you should just ride the roller coaster. Like people try to mitigate it, but just ride it. And try to have fun. >> Please welcome Eric Landau who is the co-founder of Encord who is building the data layer for physical AI, right? >> Yep. >> Welcome Eric. >> Thank you. Thanks for having me. >> [applause] [0:30] >> So, as I was going like preparing the the chat and looking at your background, I was actually pretty surprised. Like it's kind of like a very surprising that you did big data particle physics. And then…
Full transcript available for MurmurCast members
Sign Up to AccessMore from Y Combinator
What Big Tech Missed And How Startups Can Still Win
Alex Lebrun, CEO of Amilabs, discusses his journey building multiple AI startups and explains why world models—AI trained directly on sensory data rather than text—represent the next frontier beyond large language models. He argues that being early in technology requires specific founder strengths, and shares lessons on ambition, fundraising, and the challenge of managing a $1.2 billion seed round.
New Operating Systems for the Physical World
Traditional software for physical-world industries like construction and fleet operations hasn't evolved in 20 years, but AI agents, robots, and wearable-equipped humans now create a new operating system opportunity. These new systems will manage human and robot labor together, capturing end-to-end work data that incumbents lack, while operating in industries that spend 10-100x more on labor than software.
Opencode CEO: Blocked, 20X Growth in 6 Months, Building the Coding Agent for the World
Jay Vineet, CEO of OpenCode, discusses how his open-source coding agent reached 13 million monthly active users and $40M annualized revenue in just 8 months by offering free access to coding agents globally and a cheap subscription tier for real work. He shares how being blocked by Anthropic's Claude Code actually accelerated growth, the company's unique insights into model usage patterns, and his 16-year journey building the same legal entity through multiple failed startups before finding product-market fit.
How Photoroom Trained Themselves To Dream Bigger
PhotoRoom founders discuss how YC transformed their ambition from a vague dream to concrete billion-dollar thinking, and share tactical strategies for European founders to overcome cultural barriers to ambitious goal-setting. They emphasize that depth in focus and hiring ambitious people are keys to achieving exponential growth.
The Best Time to Build in Crypto
Y Combinator is increasingly bullish on crypto despite current market downturns, expecting more startups to integrate crypto infrastructure for capital raising and payments. The firm argues that bear markets attract serious builders focused on real innovation rather than speculative gains, and highlights emerging opportunities in stablecoins, trading infrastructure, and financial rails.