Uber Is Spending Ten Billion Dollars on Something It Cannot Stop

Uber is cutting 3,300 roles, about 10% of its global staff. Bloomberg reported it yesterday. It is the company’s largest workforce reduction since 2020.

The number is the story everywhere. The sentence underneath it is the one worth reading twice. Dara Khosrowshahi said the restructuring makes Uber “simpler and faster,” and that the savings get reinvested in “the autonomous future.”

That is not a layoff story. It is a capital allocation story, and Uber has been unusually direct about saying so.

What the summer actually looked like

Three reductions, and they are not all pointing the same direction.

On June 3, Uber cut 23% of its People and Places division. That is the unit that runs human resources, recruiting, workplace facilities and culture. Nikki Krishnamurthy departed as chief people officer. Jill Hazelbaker was elevated to president and chief corporate affairs officer, now overseeing safety operations in addition to people and places. A spokesperson told Bloomberg the reductions were unrelated to artificial intelligence.

On July 22, Uber cut 10 percent of its community operations team, their customer service arm. Megha Yethadka, vice president of global community operations, wrote that the organization “has become too complex and siloed,” and that AI tools showed promise but structural fragmentation was slowing their potential. Bloomberg noted this was the first time Uber directly attributed a headcount reduction to its AI agenda.

Yesterday, the 3,300, aimed at several management layers.

It is tempting to read those three as one plan to fund their robotaxi bet. Two of them support that reading. The first one does not. Uber even said so. So, one round explicitly not about AI, one round explicitly about AI (and the first time they said it out loud) and one round about the shape of the organization with the reinvestment openly named.

The figure that deserves to sit next to the layoffs

Uber has committed more than ten billion dollars to robotaxi partnerships, with investments in Avride, Lucid, Nuro and Rivian.

Now read that against the three dates.

On June 29, Uber and Waymo ended their robotaxi pilot in Phoenix, after three years. Uber said it would announce a new partner there.

Waymo has told Uber it intends to offer robotaxis on its own app in Austin and Atlanta starting in January 2028, alongside the existing Uber offering rather than instead of it. That statement came from Uber.

The contract between them terminates in May 2028. The Financial Times has reported that Waymo is looking for a way out. Waymo has not commented, and I am not trying to characterize their intent.

Where the exposure actually sits

So here is the chessboard. Uber is spending over ten billion dollars building a platform for a capability it does not own, with a flagship partner that has already exited one market, has already named the month it starts competing directly, and holds a contract that runs out four months after that.

None of that is a technology problem. All of it is a strategy problem.

I wrote last week that you do not own something unless you can stop it. This is the same test with more zeros. Uber can stop paying. It cannot stop Waymo from launching its own app in January 2028, and it cannot extend an agreement that expires in May. Uber owns the spending decision. It does not own the outcome.

There it is, the vendor dependency problem, at scale. This is important to recognize, because most of us are making a smaller version of the same decision. Almost nobody is buying robotaxis. A great many companies are buying an AI capability from a vendor who also sells it to their competitors, on terms they did not write, with a renewal date nobody has looked at since signature.

The next questions

There is a real lesson here about judging AI investment, and it is not that spending equals seriousness.

Ask what you control at the end of the spending. What happens at renewal. Whether the capability is separable from the vendor at all. Who can turn it off. And what your position looks like on the day the vendor becomes the competitor.

Uber has answered those questions in public, for ten billion dollars, with a due date. That is more than most companies can answer privately about the tools they are currently building dependencies on.

That is the part I would actually give Uber credit for. “Reinvest in the autonomous future” is a stated goal with a funding line behind it. They also have a number you could check against later. Most AI budgets I have seen do not have one.

The exposure is real. Make sure it is an exposure somebody chose on purpose.

Originally posted on LinkedIn.