AI bubble won't last forever when a lot of compute is burned at a loss just so people can generate AI videos of sharks driving cars for social media shorts. It will burst at some point, at which HW manufacturing will have to lower prices if they still want to have enough sales to stay in business, since most of their current sales boom comes from HW they haven't even made yet.

OpenAI can't keep losing investor money forever with nothing to show for, at some point the first domino will fall, then the rest of the industry will go too from investor panic.

▲

arisAlexis a day ago | parent | next [-]

Nothing has ever burst when production can't meet demand

▲

joe_mamba a day ago | parent | next [-]

It always bursts when demand stops. Current demand is artificially pumped up and financially unsustainable aka a bubble,

▲

hiuioejfjkf a day ago | parent [-]

[dead]

▲

mikkupikku a day ago | parent | next [-]

All exponential growth curves are actually S curves, before the inflection point.

▲

re-thc a day ago | parent | prev [-]

The free demand is increasing. With price increases hitting the cloud, will there still be demand?

▲

hiuioejfjkf a day ago | parent [-]

[dead]

	▲	alt227 a day ago \| parent \| next [-]
		At this point all the programmers will stop using ai, hence bubble bursting. What is the point of paying $5000/month to keep a job which pays $10000/month?
	▲	Spivak a day ago \| parent \| prev [-]
		I mean it will be the employers of those programmers and at $5000/dev/month I expect a businesses will start demanding very tangible returns from this spend. And as much as I love the tools I don't think it's generating that much direct business value. It's very obviously not turning $140k devs into $200k devs.

▲

lm28469 a day ago | parent | prev [-]

The demand being: "trust me bro we will pay you when we're profitable"

All it'll take is one company to go bust, oracle for example, for the whole thing to deflate

Plus you're factually wrong, it happened for fiber optics and railroads

▲

joe_mamba a day ago | parent [-]

>All it'll take is one company to go bust, oracle for example, for the whole thing to deflate

Provided that of course, the US administration will be incorruptible enough to not bail out these tech companies with taxpayer money when they do eventually fail.

But when you see the connection between Larry Ellison and Trump, you realize the whole "free market competition" is a scam for suckers. Always has been, just that now they don't even bother to hide it via some complex facades and shell games to garner a veneer of legitimacy, it's straight up banana republic style of corruption.

▲

lm28469 a day ago | parent [-]

> Provided that of course, the US administration will be incorruptible enough to not bail out these tech companies with taxpayer money when they do eventually fail.

I'd love them to try that because virtually no one on any part of the political spectrum would get behind that besides the most corrupted and soulless ghouls masquerading as politicians

	▲	GCUMstlyHarmls a day ago \| parent \| next [-]
		I dunno between following the party king and "we must bail them out to avoid total economic collapse" (real or imagined), I wouldn't be betting against bailouts.
	▲	lotsofpulp a day ago \| parent \| prev [-]
		I'd love them to try getting caught on audio asking a governor to find votes, and campaign on pardoning people convicted of treason because virtually no one on any part of the political spectrum would get behind that besides the most corrupted and soulless ghouls masquerading as politicians. I could have substituted many other things up there. I was very naive when I thought getting caught on audio talking about grabbing women by the pussy and being able to do whatever you want to them because you’re a celebrity was one of those things too.

▲

echelon a day ago | parent | prev [-]

I really don't think so. I feel like we're at a takeoff.

Senior engineers using AI coding are 10x more productive. My output has jumped dramatically. I'm a senior engineer and built six nines, active-active systems that moved billions of dollars a day. I am absolutely a beast with these models. I can replace an entire team just by myself. I'm literally shipping an entire week of features in half a day. I'm reviewing the code and planning the architecture - I am not dialing this in.

Video editors using video models can replace entire studio production departments. Writer-directors who know how to direct are essentially now Hollywood studios in their own right. I know a lot about this in particular because I've been making films as a hobby for 15+ years and work with a lot of industry professionals.

You'll see a lot of slop, but that's the same thing we got when we gave the masses cell phones with cameras attached to them. We still have plenty of amazing photographers in the world, and the means of creation are only getting cheaper/easier and the scope of creation for any individual is growing and growing and growing.

This is the next industrial revolution.

▲

sz4kerto a day ago | parent | next [-]

So prices will need to increase -- if it makes a senior engineer 10x more productive then coding assistants could easily cost 20x-100x more then what they cost today. Same for video generation.

▲

echelon a day ago | parent [-]

Given that 10x engineers cost in the millions and that movies cost in the hundreds of millions - this is okay!

Edit: HN rate limit won't let me reply, so here -

I'm saying that hiring ten senior engineer costs millions. (Not a single 10xer - that's such a debated thing anyway, Fabrice Bellard or not.)

AI companies will make bank when they've hooked us all on the tools and raise prices.

Companies would likely rather pay $500k/yr to Anthropic and $750k/yr to engineers than $2M/yr to an uneven team of humans with HR, taxes, and other expenses, attrition, etc.

Anthropic is going to make bank.

▲

joe_mamba a day ago | parent | next [-]

How many 10x engineers paid millions are out there? How can you stay in business as an AI company by only charging those 10x engineers 200/month?

Edit: Fabrice Bellard is a 10x engineers because he invents cool and innovative tools that didn't exist, not because he can bang out code 10x faster. AI can't replace fabrice Ballard.

▲

fauigerzigerk a day ago | parent | prev | next [-]

The price of tools isn't determined by how much money they make or save the user. That's just the price cap. The price floor (in the long run) is the cost of making the tool. The actual price will be somewhere in between depending on competition.

▲

alt227 a day ago | parent | prev | next [-]

> Edit: HN rate limit won't let me reply, so here -

There is a reason for that...

▲

Bombthecat a day ago | parent | prev [-]

Yap, that's what people don't want to hear.

Right now we are in the cheap phase.

Price can easily triple

	▲	s1mplicissimus a day ago \| parent [-]
		> Price can easily triple They can just as easily plummet to a fraction. Really depends on wether there's value for the entities that are paying

▲

ndr42 a day ago | parent | prev | next [-]

If you are able now to create 10 products instead of 1 in the same time frame you will have to plan, review and maintain 10 things instead of 1. How can this work? I mean to double your productivity is a huge jump but 10x sounds unsustainable.

	▲	hypeatei a day ago \| parent [-]
		Well, AI fanatics aren't about longevity or maintaining things. The fact that the LLM spit out a bunch of code is good enough for them. Drive-by PRs and vaporware are their bread & butter.

▲

iamtheworstdev a day ago | parent | prev | next [-]

Yea but are you paying a profitable amount of money to your service provider for you to do it? I find it hard to believe that Anthropic is profiting off of my $100/mo subscription based on how active I keep my machines running.

	▲	mrweasel a day ago \| parent [-]
		The numbers mentioned by Ed Zitron in his podcast Better Offline recently suggested that a $200/mo Claude subscription allows you to spend $2300 - $2700 worth of Anthropic tokens. That's pretty bad, but better than I expected. I don't see it being unreasonable that models and infrastructure could improve enough to bridge the cost gap within five to ten years. It's just that the AI companies already spend so much money that it might not matter.

▲

lm28469 a day ago | parent | prev | next [-]

Do you get paid 10x? Does your company make 10x?

Nope, because the only companies making money on this bs are companies selling pickaxes and shovels

▲

silver_silver a day ago | parent | prev | next [-]

The video models aren’t that good yet but for coding the utility is clear, yes. To be fair Darren Aronofsky also overestimates their quality.

Correct me if I’m wrong, but generating video is also much more resource intensive than equivalently productive text-only model use. It seems the industry could save itself a lot of hassle and infamy by simply avoiding artistic fields.

▲

mirsadm a day ago | parent | prev | next [-]

This is yet to be seen. Certainly feels like I'm more productive but I'm not seeing any faster results. It would be nice for this to be studied more.

	▲	jacquesm a day ago \| parent [-]
		What people mostly see is the illusion of productivity. But the measure should be outcomes, not the amount of stuff made. If a factory produces 10x the product but it is only 1/3rd the quality of what it was before that is long term unsustainable and leaves the door open for a competitor to attack them on quality. This is the key driver behind all those 'enshittification' problems that we see. Quantity over quality is almost always a balance and not a binary, if you start treating it as if one should always trump at the expense of the other then sooner or later it will catch up with you.

▲

joe_mamba a day ago | parent | prev | next [-]

>Senior engineers using AI coding are 10x more productive.

Are the subscriptions of those engineers enough to make their use-case profitable and on top to also be subsidizing the cost of AI video slop generation and keep the company profitable?

>Video editors using video models can replace entire studio production departments.

Then why is OpenAI losing more an more money?

>This is the next industrial revolution.

I'm not saying it isn't, but we did have the .com bubble burst even though that was also revolution. Something can be a bubble and a revolution simultaneously. The internet didn't go away after the .com bubble burst, just the crazy speculations did, which is what I was saying will happen with the AI bauble. The bubble will burst and only the profit generating parts of AI will remain.

▲

re-thc a day ago | parent | prev | next [-]

> I can replace an entire team just by myself. I'm literally shipping an entire week of features in half a day. I'm reviewing the code and planning the architecture - I am not dialing this in.

So you can review so much code so fast? Are you sure?

In many companies code reviews (properly) are the bottleneck. This was the case without AI. Now you're saying AI is giving you 10x more code reviews and you're even faster.

What am I missing?

p.s. I agree AI can make you and things faster just not suddenly god mode.

	▲	oohbkkb a day ago \| parent [-]
		10x AI speed up only happens when you stop reading the code (or start skimming it, etc). This is pretty obvious to anyone that uses the tools and many vibe coding proponents have said as much. Sacrificing quality for quantity makes these tools much less impressive. I say this as I tab over to my bug ridden memory hog CC tmux tab.

▲

GlacierFox a day ago | parent | prev | next [-]

This is soul destroying. Literally made my day worse thinking about this.

	▲	joe_mamba a day ago \| parent \| next [-]
		>This is soul destroying. Why?
	▲	echelon a day ago \| parent \| prev [-]
		10,000 students go to film school every year. A handful of them will have the autonomy and scope they want. The rest of their dreams die on the vine. This is my friends' lives. That should make your day worse.

▲

throawayonthe a day ago | parent | prev [-]

the dotcom bubble bursting didn't mean the internet wasn't an extremely valuable technology - still a bubble