July 31, 2025
The objective of this white paper series is to contemplate the path forward for generative artificial intelligence. Part 1 reviewed scaling laws, a critical factor that drives AI infrastructure investment. To-date, scaling laws remain firmly intact driven largely by test-time scaling—the foundation for the newly popular reasoning models—and AI infrastructure investment has not abated. Part 2 examined the return on investment (“ROAI”) from GenAI. GenAI applications will only be broadly adopted if their cost can be justified by the returns. Indeed, the adoption of GenAI is growing as consumers and businesses learn how to wring productivity benefits from these tools. This paper, Part 3, explores the potential for GenAI to be leveraged in the physical world. Intelligence-infused machines—broadly referred to as embodied or physical AI—can perceive, reason and act by leveraging advanced hardware—sensors, cameras, batteries and actuators—in combination with sophisticated software—digital simulators and Vision Language Models. The rise of AI-enabled machines marks a profound transformation for the global economy, with significant implications for corporate strategy, the labor market and investors.