Why Physical AI Is the Next Platform Shift
Encord Co-CEO Eric Landau reflects on his transition from a lucrative quant career to founding an AI startup, driven by a deep belief in AI's paradigm-shifting potential. He discusses Encord's slow, compounding path to product-market fit, the pivotal role of Physical AI, and the importance of embracing the emotional rollercoaster of startup life.
Eric Landau, Co-CEO of Encord, provided insights into his unique journey from high-energy particle physics to quant finance and ultimately to co-founding an AI data platform. He detailed the evolving landscape of AI development, Encord’s arduous path to product-market fit, and the burgeoning opportunities within Physical AI.
The Shifting Paradigms of AI and a Pivotal Career Change
Landau’s career trajectory intrinsically maps to the evolution of AI. He highlighted the “Bitter Lesson,” observing a historical shift in AI development from intricate, domain-specific feature engineering to a focus on scale: “just add more data, more compute.” His early work in particle physics involved highly specific data filtering, while his decade in high-frequency trading centered on crafting market-moving factors. This contrasts sharply with modern AI, which often entails “throwing data into a big pile of machine learning.” His decision to leave a highly lucrative quant career, notably during an exceptionally profitable month for his desk, was rooted in a profound belief that AI represented “the technological paradigm shift of our generation.” Despite early struggles, he found deeper meaning as a founder, driven by impact rather than financial gain alone.
Encord’s Gradual Ascent to Product-Market Fit
Encord’s journey to product-market fit (PMF) was not a singular event but a “slowly slowly compounding” process, feeling “a little bit better every day.” An inflection point occurred with ChatGPT’s release, which significantly altered the market’s perception, making customers “more open to the conversation” around AI. Landau recounts a moment of true PMF when sales closed organically, without the founders’ intense involvement or even full understanding of the client’s business, indicating strong market pull.
Building a sales organization was characterized by “trial and error,” with success only emerging after iterating through multiple sales teams. He stressed the uncanny accuracy of Y Combinator’s advice, even when founders initially overlook it. This experience underscored the importance of following one’s gut and making swift, decisive personnel adjustments.
Pioneering Physical AI: A Multimodal Future
Encord’s product strategy began with vision data, chosen as the “hardest modality,” with an overarching goal of building multimodal AI systems, reflecting the multimodal nature of human intelligence. This vision converged with the rise of “Physical AI,” which has emerged as the most significant application for multimodal AI, spanning robotics, autonomous vehicles, logistics, and manufacturing.
Landau underscored the immense market potential of Physical AI, stating that “80% of economic activity is manipulating or moving things in the real world.” He shared a provocative prediction: “in 5 to 10 years there will be more robots than people” operating in real-world environments. Encord identifies future bets through continuous engagement with “smart people,” customers, prospects, and other founders, synthesizing these diverse inputs into strategic directions. Their platform supports a wide range of applications, including specialized cases like “facial recognition for cows.”
Scale, Strategy, and Competition
Encord operates at a substantial scale, processing “multiple petabytes of all different types of multimodal data,” a volume exceeding that reportedly used for GPT-4 training. Their product suite encompasses data management, curation (likened to “finding a million needles in a billion haystacks”), annotation, enrichment, and evaluations. They maintain a facility in the Bay Area, which functions like “film sets” for collecting robot-specific data, complementing their ingestion of customer production data.
Geographically, Encord balances its London headquarters with extensive Bay Area operations, prioritizing proximity to customers, then talent, then investors. Landau values competition, asserting that “really good competitors because they push you, they make you better.” Encord differentiates itself through its deep specialization in Physical AI and its robust capability to manage data at multi-petabyte scale. For early-stage Physical AI companies, he recommends starting with open-source tools and adopting platforms like Encord’s when they reach a critical “inflection point where they need scale.”
Embracing the Entrepreneurial Roller Coaster
Concluding with founder advice, Landau described starting a company as a “roller coaster” of “low lows and high highs.” His counsel is to “just ride the roller coaster” and “try to have fun,” accepting the inherent fluctuations. He has learned to “hack your brain to enjoy the fires,” transforming challenges into exciting opportunities, and realizing one has “more control than you think over your own mental state” in navigating the entrepreneurial journey.