Hardware's slow burn: Why physical product cycles demand ruthless prioritization

May 17, 2026 · Lenny's Podcast
🎧 PodShort 99 min squeezed to 3 AI SprinklerAS Sales Tech
Episode artwork
Caitlin Kalinowski
(Former) Head of Robotics and Hardware Division at OpenAI
Lenny's Podcast
99 min squeezed to 3
Full episode from Lenny's Podcast
Quotable Moments

What you can do behind a keyboard with AI is going to saturate. When that happens, the next frontier is the physical world.

Caitlin Kalinowski(Former) Head of Robotics and Hardware Division at OpenAI
01:26

In hardware, we only get to compile our code 'quote-unquote' like four or five times [ever].

Caitlin Kalinowski(Former) Head of Robotics and Hardware Division at OpenAI
10:21

I think there's probably more change in war than there is in consumer electronics in the next two years, for example.

Caitlin Kalinowski(Former) Head of Robotics and Hardware Division at OpenAI
01:46
Key Insights
  • The acceleration in what AI can do behind a keyboard is going vertical and will eventually saturate. The next frontier for innovation is the physical world, encompassing robotics, manufacturing, and industrialization.
  • There will likely be more transformative change in warfare within the next two years than in consumer electronics, driven by advancements in drones and military technology.
  • Unlike software development where code can be compiled daily, hardware development allows only about four or five 'compilations' (major builds) in its entire lifetime due to the physical nature and permanence of the product once released.
  • For robots to be accepted and non-threatening, their design should prioritize qualities like apparent softness, reactiveness, and clear signaling of intent, a skill that studios like Pixar and Disney excel at.
  • The fundamental technologies and lessons learned from developing VR, such as orienting objects in space and depth sensing, are proving to be highly useful and are now being applied directly to the field of robotics.
  • To ensure national security, the United States needs to re-industrialize significantly and establish independent supply chains for critical components, especially for military technology, as current allies may not remain so in the future.
  • When building hardware, it's crucial to define goals (KPIs) early and stick to them, as changes midway through development can effectively burn a lot of the initial investment due to the long and irreversible nature of physical production cycles.
  • The most effective hardware architects start design by identifying and solving the hardest, riskiest problems first, rather than beginning with the easiest or most familiar parts.
Metrics Mentioned
  • 10 billion dollars (Meta spent this amount on VR development.)
  • 70 degree field of view (A feature of the Orion AR glasses, which provides a binocular experience.)
  • 100,000 drones (A hypothetical visual scenario from Mark Andreessen about drones coming out of China.)
  • 6x price increase (Memory prices have reportedly gone up by 6 times.)
  • 4-5 times (The limited number of times hardware code can be 'compiled' or a major build can be done in its entire lifecycle.)

RevBots.ai View:

  • AI Sprinkler teams bolting hardware onto SaaS stacks will hit wall: physical cycles don't compress
  • Tab Hoppers should note: hardware requires 10x upfront KPI discipline vs software pivots
  • ARM-stage orgs will treat supply chains like code repositories: version-controlled and globally distributed
  • VR's spatial computing lessons directly transfer to robotics - watch for talent pool crossover
🎧Full Episode:Lenny's Podcast →