I wonder how long Google will continue to subsidize this at a substantial loss? Estimated $30–40 billion spent in the last decade that only really pays off if they dominate the market.
So one year of revenue buys ~2500 cars at those prices, which is roughly the size of their fleet (~3000 according to Wikipedia). It seems plausible that newer cars will be cheaper as designs get optimized, economies of scale hit and what used to be really expensive cutting-edge hardware becomes commoditized and goes down in cost over time.
They certainly also need support including contractors that assist cars that get need human input, maintenance etc. and the electricity for the cars isn't free either, but just based on these numbers, it sounds like they are likely close to being profitable if you ignore R&D.
If you assume $10 a ride, and a car giving 3 rides an hour for 12 hours a day, that's $360 in revenue per car per day, close to the $320 you'd get from $350M/3000/365. That means each car pays for itself in about a year (ignoring all other costs, of course).
Based on this and the assumption that cars last for more than 2 years, I'd guess that Waymo is only "unprofitable" (not sure how this works in accounting terms) due to ongoing R&D and expansions and there really isn't much more to "subsidize".
I don't believe Google breaks it down but everyone assumes Waymo accounts for roughly 50% of $7.52B in "other bets" losses. And that's just 2025, losses in Q4 2025 are almost 2x Q4 2024. Cost factors are improving but they continue to shovel money into the R&D furnace.
I bet they've hit operating breakeven a couple years ago. If they hadn't they wouldn't have been expanding. Expanding while you have an operating loss means the loss would be expanding alongside the service. I'm not seeing that in the numbers.
They have the money to do so, and investors are aware that it is a long term play. Waymo is already dominating the market for all intents and purposes.
I don't think Waymo needs to dominate the market to succeed. They just need to scale up (time)x(number of vehicles) enough to amortize the R&D costs of the self driving capability. Paying a driver is a big chunk of a taxi/Uber's costs, so eliminating that leaves a lot of room to maneuver.
Having lived on military bases that is a false sense of security. That's one gate guard away from a problem. They make mistakes. There are far better options he can afford.
Kalshi: Sequoia Capital, Paradigm, Andreessen Horowitz (a16z), Y Combinator, Charles Schwab, Henry Kravis (KKR), and CapitalG (Alphabet).
Polymarket: Intercontinental Exchange (ICE - parent of the NYSE), Founders Fund (Peter Thiel), Vitalik Buterin, and 1789 Capital (Donald Trump Jr.).
PredictIt: Primarily supported by Aristotle International (a political tech firm) and historically Victoria University of Wellington.
Donald Trump Jr.: Prominent investor in Polymarket through his firm, 1789 Capital, and serves as a paid strategic advisor to Kalshi.
The current CFTC Chairman, Michael Selig, created a 35-member panel to draft new regulations for prediction markets. This panel includes the CEOs of the platforms they regulate, such as Shayne Coplan (Polymarket) and Tarek Mansour (Kalshi).
These companies were blatant about operating illegally in the US. I remember being shocked to see the live billboard polymarket ads in Philadelphia on the eve of the election. I guess there are too many scandals these days for this to stand out.
This tracks my usage exactly. It was like Mad Libs - in that moment it was THE MOST FUN but after a while it became just a novelty bordering on... creepy. Now I feel kind of guilty for having exposed so many friends to what looks like a data gathering scheme.
The struggle to "compete with China" on price is not new. Tesla's failings have everything to do with it's failures in execution, quality and brand image. Apple does quite well in China.
It aligns with Emil Michael, the Pentagon’s chief technology officer who said "I need someone who’s not going to wig out in the middle"
Both statements give the impression Anthropic is looking to be directly inserted into the kill chain command structure which as far as I understand is not accurate.
reply