Back in the 1970s, people began making personal computers. Everyone thought they were really cool but nobody really knew what to do with them, beyond playing primitive games. Then we got the internet and useful software like spreadsheets and word processors, and they became indispensable tools for work and life.
It took a while for people to figure out how to use the new technology.
Now we have AI, or LLMs to be precise. They are amazing, but what are they good for? Writing essays for lazy students? Generating shlock internet posts?
So for a while serious analysts were looking at the AI companies and thinking that they might end up being economic failures.
And then Anthropic created Claude Code, an "agent" that writes software. Suddenly the money started pouring in from tech companies who never thought they had enough programmers or could write code fast enough. Profits at Anthropic and OpenAI soared, and people started to say that AI had finally found the "killer ap" that would justify all the money invested in it.
But as Noah Smith explains, it isn't that simple:
Now AI had found its killer app — the equivalent of e-commerce and search for the internet, or spreadsheets and word processing for computers. Suddenly, everyone in the world was “tokenmaxxing” — trying to use coding agents as much as humanly possible.
An entrepreneur breathlessly told me that he ordered his employees to “spend their salary in tokens” — that is, to create so much code with Claude Code and Codex that it cost as much as their entire paycheck. I remember asking him: “What are they using all those tokens to create?” I don’t think I got a straight answer; I’m not sure he knew.
He wasn’t alone, though. Plenty of companies encouraged their employees to use AI coding agents as much as possible. Meta even briefly had a leaderboard for who could use the most tokens. One company reportedly spent half a billion dollars on Claude Code — equal to one percent of Claude’s annualized revenue!
Reading these reports, I just kept wondering: What are all these tokens actually producing? There never seemed to be a clear answer.
The stuff I’m hearing is just insane. People are spending hundreds of thousands of dollars a month on tokens? Guys, what are you shipping?…I am seeing people fully enraptured by illusions of productivity. They have swarms of agents coordinated by Byzantine Octopus harnesses. They’re munging thousands of tokens a second. They’re doing all this stuff, churning unfinished marginalia faster than ever before. Spinning their wheels and shipping absolutely jack shit for their customers…[W]e’re getting a lot of utility from AI for engineering at our company. I think we would really struggle to burn more than $5K per engineer per month.
Smith again:
Uber COO Andrew Macdonald said it wasn’t yet possible to draw a link between raw AI usage and useful products actually being shipped: “That link is not there yet, right?”Microsoft and several other tech firms have begun canceling Claude Code licenses and placing limits on how much can be spent.
What's going on?
As I see it, the price of coding just fell by a lot, but nobody knows yet how to productively use all that new coding power. Tech executives felt that they were being held back by a lack of coders, and maybe they were, but their overall process was adapted to the speed of what their people could actually produce, and they really had no idea what to do with this additional capacity. So much of it was just wasted.
And what use will we make of all this coding power and the new software it will create? I don't know, but I imagine people will find lots of uses. I curse every day about some stupid web site that doesn't work how it is supposed to, some process that is way more complex than it needs to be, games with too many glitches, and so on.
But it will likely to take years for all of this to really translate into better products for us, and more profits for tech firms.
No comments:
Post a Comment