AI tools have dramatically increased engineer output, with Anthropic engineers shipping 8x more code and most commits now AI-assisted.

The steady-state AI spend will be $100k of AI tokens per engineer.

“I think it'll land at 100 grand per engineer equivalent. I think that's what we'll I think we'll give each of our best engineers $100,000 of tokens and in return we'll cut the size of our dev teams 30 40% effectively.”

Very narrow deep specialists and new grads without enough experience are the roles waning in demand because products like Cursor can now do some of that work.

“I think the people who are maybe like very narrow, very deep, and like overly specialized in a way are waning like in like in the same way along that vein might also be like the new grad who just hasn't had enough like experience to be able to do that where a product like Cursor or others can do some of that work right now.”

AI tools now let a single contributor do data analysis work in a Hex thread that would have taken an Amazon L7 a week or two.

“You can call data now on a hex thread that is basically what it would take a week, two weeks with an Amazon L7.”

Software development has compressed so dramatically that a single person can now launch a product in a few weeks, a task that previously required tens of engineers and multiple quarters.

“Today, one person a few weeks can possibly launch their ideas into a product and scale quickly.”

Software engineering is shifting to a model where one senior engineer manages multiple unsupervised downstream agents that implement features, with engineers reviewing outputs rather than writing code.

“You have one senior engineer managing a bunch of downstream agents that are actually doing the work. And then we're now getting to the point we have like mostly unsupervised agents taking feedback from the team on things, implementing features, and then the engineers are coming in and actually checking that what it built makes sense.”

AI coding tools already represent close to 20% of total US software engineering wage spend.

“If most of this open AI and entropic enterprise web if a large percentage of that is software they really are close to 20% already.”

Some tech workers report being forced to use AI under threat of job loss, yet still observe continued layoffs even after adopting AI.

“I've been forced to use AI or lose my job. And even when I use AI, I'm still seeing people lose their jobs. I just hate it.”

By 2027, enterprises may be unable to prove any productivity gains from AI investments.

“We may reach a situation in 2027 where we just cannot prove any of these productivity gains.”