Elon Musk spent roughly $10B on AI training hardware in 2024 | Hacker News Reader

Elon Musk spent roughly $10B on AI training hardware in 2024

tomshardware.com

6 pointsrbanffy7 comments

He has to. At least for the self-driving taxis, if the competition (Waymo) gets to the superhuman level of driving safety before Tesla, then all the gamble will be for nothing. If Waymo succeeds, then it's very likely car ownership would plummet. Why would you own your own car, when a robotaxi can reliably show up in a matter of minutes and could cost, let's say, $1/mile?

One reason would be ... because people like to own things.

I doubt car ownership would really plummet. In a lot of big cities, people don't own a car anyway.

And in small or more distant places, robotaxis may not be viable economically - too few people, too long distances, need for automated cleaning, charging etc.

Car ownership can definitely become a past fad.

How many people still own their music library vs subscribing to Spotify?

I think the rampant success of streaming services and subscription services shows how little people really care about ownership.

Our country (the US) was not designed for public transportation. Renting a car is fine if you never want to leave your suburb. Many people where I am at commute 100mi to the bay for work.

I think Elon's bet on driverless vehicles is stupid. It just won't work outside cities. Car culture in America isn't going anywhere.

> Car culture in America isn't going anywhere

I’m sure that’s what record companies were saying about selling music circa 91

What a completely idiotic parallel to try and draw.

That's at least 3 product generations and tens of billions in capex away. Uber provides 28 million rides per day. Waymo is somewhere past 100,000 per week and growing fast, but that's still something like 0.05% of Uber's volume.

Waymo is spending billions to expand right now, which shows confidence in the tech and that overhead for remote monitors, charging depots, cleaners, and CSAs is sustainable. If Waymo can scale up 40X That could get them to 2% of Uber, but let's be generous and say Waymo availability can be a fairly aggressive 3X of an Uber driver, so that's 6%. If my math is correct, assuming very little in free cash from from operations, that's plausible with the recent $5B investment announcement.

I also expect pricing will remain a bit above Uber until some tipping point is reached in terms of building the vehicles and sensors scales up. Waymo would go from hundreds, maybe low thousands of vehicles now to tens of thousands.

Reply on news.ycombinator.com