LOS ANGELES, 13 March 2026 — Travis Kalanick has brought his post-Uber businesses into public view under a new name, Atoms, with an industrial-automation thesis built around specialised machines. The group is targeting food production, mining and transport rather than asking one humanoid form to perform every physical job.

The launch turns a collection of operating companies and emerging robotics projects into a single story about physical productivity. It does not yet prove that one technology platform can span kitchens, quarries and freight yards. Those environments share a need for automation but differ sharply in safety, duty cycle, materials and economics.

A holding company becomes an automation platform

Reuters reported that Kalanick launched Atoms by expanding and renaming City Storage Systems, the company he began building after Uber. The new organisation divides its ambition into Atoms Food, Atoms Mining and Atoms Transport.

The structure matters because it begins with businesses that already handle physical operations. Food infrastructure provides immediate workplaces for automation, while mining offers repetitive movement in controlled industrial territory. Transport is described more broadly as a wheelbase for robots, leaving the eventual products and customers less defined.

Three bright industrial scenes show a compact food-assembly robot, an autonomous open-pit mine hauler and a low robotic cargo wheelbase without branding
Each environment rewards a different physical design even if sensing, prediction and fleet software can share common ideas.

Specialisation narrows the engineering problem

  • a kitchen robot can be built around known ingredients, containers and cleaning routines;
  • a mine hauler can optimise for payload, dust, gradients and marked operating zones;
  • a freight platform can sit low beneath pallets instead of imitating human posture;
  • sensors and software can be reused while actuators and frames remain job-specific;
  • commercial value can be measured against a defined shift, route or production station.

The food division provides an operating base

Atoms lists CloudKitchens, Otter, Lab37, Picnic and ProFood among its food businesses. Together they cover delivery-kitchen property, restaurant software, robotic preparation, workplace meals and production infrastructure. That portfolio offers data and physical sites where machinery can be designed around real workflows.

It also creates an integration challenge. A property network, an order-management system and a preparation robot have different unit economics. Calling them one automation stack is credible only if shared technology lowers operating cost or improves output across the businesses rather than merely placing them under one brand.

Mining offers a clearer job for autonomy

The group's site presents Pronto AI as an autonomous-haulage system for mines and quarries. These sites have large vehicles, repeated routes and valuable safety gains from removing people from hazardous operating zones. A machine designed for hauling can devote its entire geometry and control system to that duty.

At launch, reports described Kalanick as close to acquiring Pronto rather than confirming that the contemplated transaction had already closed. The distinction is important. A displayed strategic direction and a completed ownership transfer are different facts, even when the technology fits the new group's mining thesis.

A saturated hand-drawn industrial scene contrasts a food arm, mine hauler and low cargo robot with a small multi-jointed humanoid struggling with mismatched tools
Specialisation trades general flexibility for fewer mechanical compromises and a more direct route to measurable work.

The wager runs against the humanoid fashion

General-purpose humanoids promise to use spaces and tools built for people. Their appeal is flexibility: the same body might move between tasks. The cost is that walking, balancing, dexterous handling and reasoning must work together reliably before the machine can complete even a simple industrial shift.

Atoms argues for what it calls gainfully employed robots—machines suited to a job and capable of producing enough value to justify their cost. This is not proof that humanoids have no role. It is a capital-allocation choice: solve a constrained, high-frequency task first and widen the platform only after the economics work.

Understanding, prediction and control form the software layer

The company describes its operating sequence as understanding the current physical state, predicting what happens next and controlling the outcome. Sensors create the observation, models estimate motion or demand, and machinery performs the action.

The useful common layer may therefore be less visible than the robots themselves. Fleet management, simulation, maintenance data and safety monitoring can transfer between divisions even when a bowl assembler and a haul truck share few mechanical parts. That is the strongest route to a platform rather than a loose portfolio.

Kalanick returns to physical networks

In the United States, Kalanick is best known for scaling Uber's ride network. Reuters noted that he resigned as chief executive in 2017 and left the board in 2019. Atoms brings him back to a public operating narrative after years spent building City Storage Systems and food infrastructure.

There is continuity in the focus on dispatch, utilisation and physical movement, but industrial robotics carries different liabilities. A failed app transaction is inconvenient; a failed control decision around heavy equipment, hot food or workers can cause physical harm. Safety cases and site-level service will matter as much as software velocity.

The launch now needs evidence from productive shifts

Atoms arrives with a broad mission and an unusually concrete design preference. Specialised machines can reach deployment faster when their jobs are repetitive and valuable. The portfolio also gives the company places to test systems without waiting for an external customer to redesign operations around a prototype.

The next evidence should be operational: hours worked without intervention, cost per meal or tonne moved, equipment availability, safety events and customer renewals. A robot earns the description “gainfully employed” only when it repeatedly performs useful work at a total cost below the alternative. The launch defines that test; field results must now answer it.