Physical AI: The Next Platform Shift Explained

Key Takeaways

• Physical AI dominates economic activity: 80% of economic activity involves manipulating or moving things in the real world, making physical AI applications essential.
• Multimodal data is critical: Video, sensor data, audio, and language combine to train systems for robotics, autonomous vehicles, and manufacturing.
• Scale is the competitive edge: Operating at the petabyte level of data—managing and curating massive datasets—separates leaders from followers.
• Market timing matters: ChatGPT's emergence accelerated adoption and validated long-term bets on AI infrastructure.
• Founder mindset shapes outcomes: Embracing the "rollercoaster" of entrepreneurship rather than fighting volatility leads to better decision-making and enjoyment.

Why Physical AI Is the Next Platform Shift

Physical AI represents the convergence of machine learning with real-world robotics and autonomous systems. Unlike previous AI applications focused on digital tasks, physical AI tackles the majority of global economic activity—manufacturing, logistics, autonomous vehicles, and consumer robotics. Encore, a data infrastructure company founded in 2021, has positioned itself at the center of this shift by building tools to manage, curate, and annotate the massive datasets these systems require. The company now handles multiple petabytes of multimodal data, far exceeding the scale used to train GPT-4.

The Data Infrastructure Challenge

The core challenge in physical AI isn't building robots—it's managing the data that trains them. Encore's approach involves three layers: data collection, curation and management, and annotation and enrichment. Data collection happens both in production environments and in Encore's Bay Area facility, where robot systems collect video, sensor data, audio, and language inputs across various tasks. Curation is particularly critical; as Eric Landau describes it, "finding a million needles in a billion haystacks." Not all collected data is useful; the real value lies in identifying which datasets improve model performance. This scalability challenge—operating reliably on petabyte-level datasets—remains a significant technical and operational hurdle that separates mature AI infrastructure providers from earlier-stage competitors.

Market Timing and Competitive Strategy

When Encore pivoted toward physical AI, the broader market hadn't yet caught up. The company began with computer vision, then transitioned to general multimodality, before recognizing that physical AI represented the fastest-growing application area. This wasn't pure foresight; it came from constant customer conversations and market sensing. While competitors have consolidated or exited, Encore welcomes competitive intensity, viewing strong competitors as forces that push the entire ecosystem forward. The company's differentiation lies in its ability to scale data operations and deep expertise in multimodal systems, not in being first-to-market.

Founder Insights: Embracing the Entrepreneurial Rollercoaster

Landau's transition from a decade as a quantitative trader to founding Encore during COVID illustrates a common founder challenge: accepting non-linear progress. The month he quit Wall Street, his former desk earned more in a single month than in the previous three years—a stark reminder of the financial sacrifice. Yet he emphasizes that starting a company is unavoidable a "rollercoaster with both extreme lows and highs." Rather than fighting these fluctuations, he advocates riding them and finding enjoyment in the journey. This mindset shift—realizing you have more control over your mental state than you think—becomes increasingly valuable as companies scale and face larger challenges. Early on, conviction in AI as a generational paradigm shift carried Encore through the "desert" years before ChatGPT validated broader market interest.

Conclusion

Physical AI is not a speculative future—it's already shaping industries from manufacturing to autonomous vehicles. Success in this space depends less on novel algorithms and more on the unglamorous work of data infrastructure: collection, curation, and annotation at scale. For founders entering this space, the lesson extends beyond product-market fit. It's about maintaining conviction, embracing uncertainty as part of the journey, and staying relentlessly focused on where customers and talent concentrate. Encore's growth demonstrates that solving hard infrastructure problems at the right time creates outsized value.

---

Original source: Why Physical AI Is the Next Platform Shift

powered by osmu.app