Ask HN: If AI wins, don't AI companies lose?
10 comments
No. First of all, AI companies make money per token, not per user. The unlimited plans are all dead or dying.
Second, they can increase prices if they're really killing jobs. The price for a model that can kill 1 job is much lower than the price for a model that can kill 100.
Second, they can increase prices if they're really killing jobs. The price for a model that can kill 1 job is much lower than the price for a model that can kill 100.
That's a fair point. Model prices can increase. Let's say Anthropics/OpenAI/Gemini figure this out.
Doesn't this mean that any company that depends on headcount growth (every SaaS), loses?
100 SWE -> 10 SWE, 100 slack/gmail/notion/zoom/etc. subscriptions become 10.
Doesn't this mean that any company that depends on headcount growth (every SaaS), loses?
100 SWE -> 10 SWE, 100 slack/gmail/notion/zoom/etc. subscriptions become 10.
> Doesn't this mean that any company that depends on headcount growth (every SaaS), loses?
Yes, assuming they aren't also scaling their costs down with AI.
But this is mostly a moot point because we don't yet have any evidence that AI is killing lots of jobs. Companies are doing necessary/planned layoffs after over-hiring for years, and some of them are making it look better to investors by saying it's because they're smart (for using AI) instead of the truth, which is that they stupidly over-hired.
You also have to remember that AI is way, way under-priced right now. That $200/mo. Claude bill should probably be double or triple what it is. All of the AI companies are plowing money into keeping prices artificially low.
The economics will change a lot once they can't do that anymore. Google will likely dominate because they can burn their own cash from other businesses instead of cash from VCs or retail investors.
But if prices go up and these companies charge enough to be profitable, people will start to question whether it's cheaper to just have people doing the work instead.
Yes, assuming they aren't also scaling their costs down with AI.
But this is mostly a moot point because we don't yet have any evidence that AI is killing lots of jobs. Companies are doing necessary/planned layoffs after over-hiring for years, and some of them are making it look better to investors by saying it's because they're smart (for using AI) instead of the truth, which is that they stupidly over-hired.
You also have to remember that AI is way, way under-priced right now. That $200/mo. Claude bill should probably be double or triple what it is. All of the AI companies are plowing money into keeping prices artificially low.
The economics will change a lot once they can't do that anymore. Google will likely dominate because they can burn their own cash from other businesses instead of cash from VCs or retail investors.
But if prices go up and these companies charge enough to be profitable, people will start to question whether it's cheaper to just have people doing the work instead.
Well, if the AI really is that good, what's stopping the AI Company from charging just slightly less then the 90 saved engineers cost?
That's a fair point. Model prices can increase. This raises the next question.
Doesn't this mean that any company that depends on headcount growth (every SaaS), loses?
100 SWE -> 10 SWE, 100 slack/gmail/notion/zoom/etc. subscriptions become 10.
And now let's recurse.
These SaaS companies that use AI and dropped 90% of engineering headcount lose revenue because their customers also drop headcount.
Let's say AI costs 89% more than it did before, so the SaaS companies still get 10x productivity from the 10 engineers, but now all their customers headcount is 90% smaller too. So what now? Does every company make a pact to grow headcount ;)
Doesn't this mean that any company that depends on headcount growth (every SaaS), loses?
100 SWE -> 10 SWE, 100 slack/gmail/notion/zoom/etc. subscriptions become 10.
And now let's recurse.
These SaaS companies that use AI and dropped 90% of engineering headcount lose revenue because their customers also drop headcount.
Let's say AI costs 89% more than it did before, so the SaaS companies still get 10x productivity from the 10 engineers, but now all their customers headcount is 90% smaller too. So what now? Does every company make a pact to grow headcount ;)
Everyone will adjust pricing to cover losses ad infinitum
If you get into a situation where you are dependent on any tool for 90% of your engineering output, the maker of that tool has a lot of leverage when it comes to setting prices. So my guess is that if we get into that situation, the AI companies will be able to set prices to make sure they don’t lose money.
100 SWEs running Claude Code generating 400mn tokens/mo = 400mn * $25/mn = $10,000/mo of revenue for Anthropic
10 SWEs running Claude Code generating 400mn tokens/mo = 400mn * $25/mn = $10,000/mo of revenue for Anthropic
If AI can make 10 engineers a productive as 100, then AI companies bank at least the same revenue.
10 SWEs running Claude Code generating 400mn tokens/mo = 400mn * $25/mn = $10,000/mo of revenue for Anthropic
If AI can make 10 engineers a productive as 100, then AI companies bank at least the same revenue.
But AI companies would want more engineers generating more tokens.
In this case, Anthropic would want 100 SWEs generating 100,000/mo of revenue. Replacing the very headcount that is responsible for token usage would hurt company growth.
Not to mention what happens when companies start doing this recursively. 100 SWE -> 10 SWE, 100 slack/gmail/notion/etc. subscriptions become 10.
In this case, Anthropic would want 100 SWEs generating 100,000/mo of revenue. Replacing the very headcount that is responsible for token usage would hurt company growth.
Not to mention what happens when companies start doing this recursively. 100 SWE -> 10 SWE, 100 slack/gmail/notion/etc. subscriptions become 10.
Wins what?
AI companies don't even know what they want.
AI companies don't even know what they want.
100 SWEs all running Claude Code. Okay, company is 10x more productive than before.
Now, company cuts 90% of SWEs because AI makes 10 engineers as productive as 100 engineers.
Yay, company saves 90% of engineering costs.
But now, 10 engineers run Claude Code. And now Claude revenue drops by 10x multiple?
So if AI gets good, it will reduce the very engineering headcount that AI companies depend on, right?