Why we’re at the beginning of the AI hardware boom | Caitlin Kalinowski (ex–OpenAI, Meta, Apple)

Caitlin Kalinowski frames the coming hardware boom as the next stage of AI's development. As AI systems increasingly saturate the tasks people can perform behind a keyboard, she expects the frontier to move into the physical world: robotics, manufacturing, and industrialization. She connects that shift to a broader security argument, saying that countries need to rebuild the ability to manufacture at scale because military technology may change faster than consumer electronics and current allies or supply chains may not remain dependable. Her comments also preview the leadership themes of the conversation: Sam Altman's pressure to think at much larger scale, Steve Jobs's unwavering quality bar, and the challenge of making robots feel socially legible rather than threatening.

The discussion treats VR less as a failed endpoint than as a technological bridge. Work on spatial positioning, SLAM, depth sensing, and human visual perception remains useful in robotics, autonomous vehicles, drones, and manufacturing. Kalinowski believes AR glasses can become part of the future because they could provide information without isolating people from one another, although waveguides, MicroLED displays, manufacturing yields, cost, and unobtrusive input remain unresolved. Her account also explains why VR has struggled socially: putting a screen over the face makes it harder to maintain human connection, even when the experience itself is technically impressive.

The first hardware lesson is that physical products have far fewer opportunities for correction than software. A mass-produced device may get only a handful of major design iterations, so teams must resolve reliability, tolerance, yield, and component variation before the final production build. Kalinowski contrasts this conservative process with software's frequent compilation and updates, then asks why robotics is suddenly attracting attention: the field is becoming central to the next phase of AI even though hardware has historically been a less glamorous and less highly paid career. The core challenge is to make a product that survives real-world variation across millions of parts, not merely a prototype that works once.