Atoms' robotaxi denial carries an exception for Uber's ridesharing business

The denial is narrower than its tone. Atoms told the FT it is an industrial software company with "no plans to enter the saturated robotaxi market" ^[11], and then said Uber is a partner that may use Atoms technology for its ridesharing business if that proves helpful ^[12]. The first statement is about operating a fleet. The second is about supplying the software that drives one, and the FT describes Atoms following the model Wayve and Nuro use, where you build the robotaxi software and leave the vehicles to somebody else ^[14]. Under that arrangement a company can honestly decline to enter the robotaxi market and still ship robotaxi software, because the customer does the entering.

The money is not ambiguous. Uber's cheque is $100m ^[1] inside a $1.7bn round led by Andreessen Horowitz ^[2], about 6 percent of the round ^[18]. Ben Horowitz, who joined the board, told an a16z event it was "certainly the biggest cheque I've ever written" ^[15]. Put the stake next to the bill it echoes: the FT puts Uber's cost of the Google trade secrets litigation over Levandowski's files at close to $350m ^[3], roughly three and a half times what Uber has now put into the company that employs him ^[19].

Payroll is harder evidence than a category label. Levandowski co-founded Waymo and ran Uber's self-driving programme ^[3], and he arrived at Atoms in March through its acquisition of Pronto, which automates heavy industry and mining equipment ^[5], a purchase that fits the industrial software description exactly. Eric Meyhofer, who took over Uber's robotaxi development after Levandowski left, now runs Lab37, the arm automating commercial kitchens, and according to the FT he has hired several dozen former Uber staff and recruited from Zoox, Tesla and Waymo ^[6]. Of the more than 2,000 people at Atoms, most sit in the food division, with a growing number on autonomous vehicles ^[7]. Gautam Gupta, Uber's former finance chief, is CFO ^[10].

Uber's own spending looks like a buyer hedging suppliers rather than a company with a single autonomy plan. It launched London's first robotaxi service with Wayve last week, though not by the legal route Britain built for it ^[8], and agreed in August to put more than 2,000 Pony.ai robotaxis on European streets ^[9].

For anyone weighing a contract with Atoms, the question is not which account is true but which one would survive being written down. A supplier with no plans in a market gives up nothing by scoping the licence to industrial automation, or by granting exclusivity in ridesharing to somebody other than Uber. A supplier with plans will want the carve-out Atoms has already published, and that carve-out names its beneficiary ^[12]. Take the sentence you were told and ask for it as a term; the version that comes back on paper is the one that governs.