Hacker News Reader

Comment by simonw | Hacker News Reader

I think you're putting way too much weight into what one person said in unprepared remarks at the 27 minute mark in a 32 minute podcast conversation.

View on HN

Unprepared remarks are the best source of info frankly.

That "one" person is the COO of Uber. And the other one - the one based on whose statement about burning through yearly AI budget in the first few months - the whole discussion sprung up internally at Uber in the first place is the bloody CTO of that huge company. So yes, their words do have A TON OF WEIGHT. Thats why they are in such important positions, arent they? They're not quite the Derek from the pub, casually commenting on how Liverpool will fare this season.

I think the way people reacted to those statements was entirely out of proportion to what was said.

I repeat: a CTO saying that they spent their entire AI budget for 2026 when that budget was clearly set in 2025 before anyone knew what those November models + harnesses were capable of is entirely unsurprising. Any analysis that doesn't also point out the difference between 2025 and 2026 era coding agents is either ignorant or deliberately misleading.

Yes, but that's irrelevant, because the COO uses that to base his core argument - that all that jackshit 1800 code changes per week that the CTO boasts about, mean absolutely nothing in terms of value. It means they are spending a lot on it, to gain as he diplomatically said "perhaps 20% more" - and I wonder 20% of fucking what - it's a ride-sharing app, what could they be possibly building on top of it with all that token crap?

You have to try pretty hard to get to "all that jackshit 1800 code changes per week that the CTO boasts about, mean absolutely nothing in terms of value" from what he said on that podcast.

(We still don't even know what Uber's planned AI budget for 2026 was. They didn't reveal that when asked - in https://www.theinformation.com/newsletters/applied-ai/uber-c... it says "He wouldn’t disclose exact figures of the company’s software budget or what it spends on AI coding tools").

I don't have to try at all - I think anyone who spent as much as an internship, let alone years at a modern tech corp would have no trouble distilling the absolutely clear message - we are spending too much for too little value. And actually wtf am I explaining myself? Every major tech outlet interpreted it like that too. It's not that hard Simon.

I think that both you and the other major tech outlets interpreted that poorly.

Oh really? How about the king of non-tech-outlets and digestible, shallow bites for the not-reading-books-middle-class, the Business Insider?

https://www.businessinsider.com/uber-coo-andrew-macdonald-ai...

You know their business is literally correct interpretation of the C-Suite statements.

I think Business Insider are not a particularly high quality publication - they're prone to clickbait - and that headline is an example of why I think that.

Hah, I just checked their homepage and here they go again amplifying that COO fragment from that podcast:

https://www.businessinsider.com/tokenmaxxing-debate-uber-exe...

> "That link is not there yet, right?" Macdonald said in comments that went viral, racking up over 2 million views on X. "I think maybe implicitly there is more that is getting shipped, but it's very hard to draw a line between one of those stats and, 'OK, now we're actually producing 25% more useful consumer features.'"

Yeah, something going "viral on X" is clearly a sign that it's quality information!

So no one got it right? Myself, tech outlets, non-tech outlets, everyone on twitter, etc?

That's why I wrote a whole section about it. I was deeply disappointed at how little thought people had appeared to put into this before amplifying the misleading headlines.

For someone who cares about media hype - https://hn.algolia.com/?query=author%3Ahansmayer%20hype&type... - you don't seem to be very discerning with regards to this particular story.

Reply on news.ycombinator.com