Interview, Fireside Chat
Why Physical AI Is the Next Frontier | Applied Intuition with a16z
- Applied Intuition forecasts that intelligence will eventually be deployed on a billion machines, potentially making physical-world companies larger than their digital counterparts, with a strategic pivot over the next 10 to 20 years from automotive manufacturers to mining operators, defense contractors, and industrial clients.
- The company projects that L2+ autonomous systems under $500 will become subsidized standard equipment by 2028 or 2029, leading to routine self-driving availability in all vehicles by the early 2030s, mirroring the ubiquity of mobile phones post-2007.
- Robotaxis are expected to become routine in the 200 largest U.S. cities by 2030, with full market ubiquity anticipated by 2032 or 2033, while fully driverless long-haul trucks may take longer due to the time required to achieve redundant hardware in high-volume production.
- A new platform named "Dana" is launching to enable developers, including students, to build autonomous systems in minutes rather than weeks, with the number of bots in healthcare, construction, defense, and agriculture expected to surge as development costs decline.
- Long-term economic shifts are predicted where self-driving trucking becomes free for buyers by the 2030s, and navigation-like features become standard, driven by demographic shortages in sectors like agriculture and mining, as well as labor shortages in trucking.
- Physical AI adoption is expected to follow a trajectory similar to the PC and mobile markets, with initial slow adoption followed by rapid scaling once costs drop and ecosystems mature, particularly in B2B sectors before becoming ubiquitous in consumer applications.
- Future video games are predicted to transition from traditional graphics tooling to world-model-based technology, potentially starting with titles like Grand Theft Auto VII, while "app store" models are expected to emerge for deploying niche physical robots.
- The industry anticipates "sovereign AI" trends requiring localized business strategies to navigate complex geopolitical landscapes, with global technology progress viewed as inevitable and essential to avoid ceding economic and security advantages to other nations.
- Safety remains the primary differentiator and barrier to entry, with the industry expecting that achieving fully closed-loop systems without human intervention will occur within a few years, though real-world edge cases like fog still require current human intervention.
- Applied Intuition expects to maintain a competitive moat through one of the world's largest proprietary data fleets and advanced synthetic data tools, which allow for safe training of systems in simulated environments to bridge the sim-to-real gap.
- While legacy manufacturers may initially deploy autonomy slower due to rigorous safety cultures, the market is expected to favor an "app store" or horizontal component provider model (analogous to the semiconductor industry) where companies like Applied Intuition supply intelligence to OEMs rather than building hardware themselves.
- Technological evolution is projected to move from modular approaches to end-to-end architectures for cost and scalability, with the "chip" analogy applying to physical AI where the distribution of intelligence through existing hardware manufacturers proves more successful than new entrants building their own hardware.
- Risks include significant safety validation challenges and the potential for rapid halts due to safety incidents, with the cost of failure in physical AI being substantially higher than in digital sectors, necessitating robust safety cases and regulatory compliance.
- The demographic trend of aging farmers and shrinking workforces in countries like Japan is expected to accelerate the adoption of automation in agriculture and long-haul trucking, driven by the necessity to address dangerous or undesirable jobs rather than just convenience.
- Efficiency gains in logistics and agriculture, such as potentially reducing transportation costs to 20 cents per mile, are predicted to lower the cost of food and goods, with the integration of AI into existing machinery (retrofitting) becoming a significant market segment for owners of expensive assets.
- The future of humanoids is expected to become more tangible in the midterm, with complex tasks like housekeeping and entertainment becoming viable applications, while the industry foresees a blur between digital and physical AI as foundational techniques unify but real-time performance remains a distinct constraint.