Report estimates generative AI has generated 110 billion dollars in sales over the past year

AI companies are making a lot of money. Apparently the revenues from this wave of technology are growing "roughly three times more rapidly than the mobile or Internet waves."

There is an interesting implicit claim here: that mobile and internet revenues grew at the same rate. It would be interesting to know if that is in fact true.

The comparison misses one thing: the internet had to build distribution first. This is exactly what Bresnahan and Trajtenberg’s General Purpose Technology theory explains.

AI didn’t start from zero. It arrived on top of cloud, SaaS, APIs, payments, mobile, app stores and enterprise software.

That’s why the revenue curve is steeper: AI is not just a new product category, it is a force multiplier on every existing one.

Great that folks are attempting to build a bottom-up model of total token revenue… and put it against the AI build-out

… but these are wildly moving targets

The thing that makes me a bit nervous/inspired is that many of these tokenmaxxing customers are experimenting with heavily discounted tokens

Real work and value is happening, but a lot of experimentation — who can even guess the percentage?

We’ve never seen a technology embraced this rapidly or at this scale… feels like it’s manifest destiny, with a small chance of a premature head fake

@paulg When we see normal people dropping $200 in subscriptions wishing it was more, I’m surprised it’s not more than 3x.

@paulg thanks @paulg - btw happy to catch up in person in london or oxford or wherever.

@quantam_lock If you were a college student you wouldn't feel that.

@paulg So the bubble is not gonna burst

@paulg But why I feel there’s no compelling consumer AI product so far which is life-changer. If you compare it it mobile revolution in first 5 years we got Instagram, Instant Messaging, Video Calling, Uber, etc. In AI era I don’t see any equivalent of that.

@ashirwadsingh_ @paulg AI-ass slop comment.

The 3x faster than mobile framing is striking when you put the cost side next to it: OpenAI and Anthropic alone are spending more than $175B annually on compute and infrastructure. The revenue run rate matches what the top two labs spend — before salaries, before research. This is the fastest-growing revenue wave in history that’s also, at the moment, structurally unprofitable at the frontier.

@paulg AI 的变现速度快，背后是企业直接把效率账算进预算里了。接下来要看毛利和留存，收入冲得快不代表所有玩家都能活得舒服。

@paulg builders keep making tomorrow look obvious five minutes after it ships

@paulg hi paul,

i am an ai company and i have lost a lot of money i am not sure this is neseccerily true. you shuold check your facts i hope this helps

all the best, cheers

@paulg you're not locked in enough

AI is the new meta now

@paulg $110B and the buyers still can't tell the models apart. Imagine the number once they can lol

@paulg Ai is in fashion so it has to

@paulg The time you have to get comfortable with AI before it's just expected of you is much shorter than it was for Email, or Smartphones, or the Internet.

In simple terms in future, people who don't use AI will be slowly moved away

@paulg Will it decrease the chances of forming a bubble?

@paulg I bet @benedictevans has a chart.