Travis Kalanick built Uber into the most disruptive logistics company on the planet, but an unfortunate chain of events saw him part ways early Now he's back with a much bigger thesis on robotics; his next venture is "Atom":
i) A physical-world automation company making "gainfully employed robots"; specialized industrial machines designed for specific jobs as opposed to humanoids
ii) Three verticals:
a) Food (digitized manufacturing and logistics, evolved from CloudKitchens)
b) Mining (autonomous mineral extraction)
c) Transport (robotic vehicle platforms - strong insights from Uber here!)
iii) The core framework is "digitize the physical world" in three steps:
a) understand current physical state
b) predict future state
c) control future state
the same logic that powered Uber's real-time dispatch, is now being applied to entire industries
iv) Betting hard on the full physical AI stack: sensors, compute, AI models, manufacturing, real estate, energy, chemistry
v) Firmly anti-humanoid; the argument is that a specialized pancake machine beats a robot arm flipping pancakes every time when throughput and economics matter
Why shouldn't robots be made into the most efficient form for their task? The core counter is to build trust with us, but is it necessary?
Here's why I'm paying attention:
i) This is the guy who already digitized transportation at global scale once; the pattern recognition for what it takes to coordinate physical-world networks at millions of transactions per day is not something you can fake or replicate easily
ii) CloudKitchens laid the R&D foundations for this venture; he spent years building the operational infrastructure, real estate portfolio, and manufacturing automation that now form the foundation of Atoms
iii) Real estate as a strategic asset for physical AI is a deeply under-appreciated moat; Travis has been quietly acquiring and developing properties while every other robotics founder obsesses over the model layer
iv) The macro timing is right; AI compute costs are cratering, physical-world AI models are maturing, and the gap between "software automation" and "physical automation" is the largest remaining unlock in the global economy
Check our @khalaresearch article on "Autonomous Software Factories"
AI embodying physical form is the next evolution this tech, while I think we're early there are signs of high potential companies emerging
You just need to pay attention