Comment by keeda | Hacker News Reader

Comment by keeda | Hacker News Reader

Parent Full thread

You're on the right track but looking at the numbers incorrectly, specifically, focusing on one example the OP gave. The better approach is to look macroeconomically. Here's the number to look at: Global knowledge worker salaries are at $50 - 70T annually. That is the number enterprises are already paying for knowledge workers.

If AI makes these workers even 1% more productive, that is $500 - 700 billion value annually. At an ongoing annual $0.5T return, a $2T investment (also over the next few years, note) doesn't seem too bad!

Then consider that actual studies from all the way back in 2024, i.e. the era of spicy autocomplete, before agents landed on the scene, put the productivity boosts much higher, like 30% or more. (Interestingly, this is corroborated by survey based data from the St. Lous Fed: https://www.genaiadoptiontracker.com/) Even assuming a conservative average boost of 10%, that is $5 - 7T value annually.

Add how many ever grains of salt you want to those numbers, the investment is nowhere near as out of whack to the potential revenues as people fear. This is why everybody from Big Tech to VCs to entire nation states are desperately scrambling to get in on the action.

I know why they're doing it.

My point is that their math is wrong.

1. There's no way China will let any of the Western frontier labs in, so that's probably 1/3 out of those $50-70tn that they'll never touch.

2. It turns out that LLMs are more of a commodity than expected because the basic tech is basically "Attention is all you need" plus a few things everyone has access to (mixture of experts, caching, batching, etc). So yes, it's a "winner take most" market, but there will likely be a healthy base of cheap models so the "collection" (price gouging) part of the cycle (or enshittification) will be hard to execute.

3. Either hardware remains expensive, in which case every N years entire DCs have to be rebuilt and then Capex needs to flood in - think highway systems being rebuilt, but instead of every 20-30 years for highways, here we'd be talking every 5-7 years.

4. Or hardware becomes cheap in which case cheap LLM hosters are competitive and problem #2 is even worse. Or the nightmare scenario for all of these investment scenarios, local LLMs become viable for most people.

Even if China accounts for a whopping 50% of the knowledge work market, the TAM is still $25 - 35 trillions. The US alone is $10 - 11T. Any fraction of that is still a huge number that can recoup $2T in a few years.

I agree LLMs are already a commodity market, definitely at the non-frontier model level, but I don't think it affects monetization prospects much. After all, server compute is a commodity and yet cloud businesses have been exploding even before AI.

And compute is exactly why it won't be a "winner takes most" market. It is clear now that compute capacity is and will likely remain the biggest moat. Looking at Claude Code is instructive; arguably it was the better product, but it kept going down so much that Codex and other competitors have gained on it. Similarly, China could have the best models, but its access to hardware is deliberately limited by geopolitics, so it's likely their threat will be manageable for a while yet.

A key part of the success of AI companies will be in securing hardware and operating that infra cost-effectively via economies of scale. Hardware will remain expensive for a long time yet because all the hyper-scalers and neo-clouds are severely crunched, and all the fabs (mostly TSMC) are already at capacity even as demand keeps exploding.

And for better or worse, most of that supply will still flow through Nvidia, despite attempts from competitors like TPUs and NPUs, for the simple reason that Nvidia has the monopoly profits to outbid everyone else on the real chokepoint, which is fab capacity.

Reply on news.ycombinator.com