Analysis
I keep looking at Atoms' $1.7 billion Series A and thinking about the $100 million next to it, from Uber, and concluding I'd rather be holding Uber's position.
Atoms has to build the stack, staff the safety team, run the fleet, and absorb whatever regulatory and liability exposure comes from actually putting vehicles on public roads. Waymo has spent over a decade and multiple billions doing exactly that. Tesla's Cybercab drew a federal safety probe within hours of its first Austin deployment. Every operator in this category is underwriting a safety case in real time, in public, with its own balance sheet.
“If Atoms fails, Uber has lost $100 million and moved on to the next supplier.”
Uber underwrites none of that, per TechCrunch's reporting on the round. It writes a check into Atoms, another into whichever AV supplier launches next, and it aggregates demand across all of them -- the same playbook it already runs with human drivers, just applied to a fleet Uber never has to own, insure or maintain. If Atoms works, Uber captures distribution economics on top of it. If Atoms fails, Uber has lost $100 million and moved on to the next supplier.
This is not a new insight about platforms versus asset owners -- it is one of the oldest arguments in venture. What's different in 2026 is how much capital is flowing to the asset-owning side anyway. a16z put $1.7 billion behind Atoms specifically, and the market keeps pricing operators as if the fleet is the valuable asset rather than the demand it serves.
Room for disagreement: The aggregator argument assumes AV technology commoditizes the way ride-hailing supply did, and that is not obviously true yet. If autonomy turns out to be a genuine, durable technical moat -- the way Waymo's decade of supervised miles suggests it might be -- then the operator who actually cracks the safety and reliability problem captures nearly all the value, and Uber's diversified bets look like a hedge against being wrong rather than a smart allocation. Anyone confident a single AV stack pulls meaningfully ahead of the others should own that operator directly, not the company writing small checks into several of them.
