Hacker News Reader

Comment by ymhr | Hacker News Reader

Not OP but I think the most obvious change will be rapid price increases as investors begin to demand profits - no more subsidised $20p/m plans for example.

View on HN

So far I know, service providers hosting open-weight models are not in such a financial conundrum and can continue serving at the current price levels.

Also, even if the 20 dollar subs disappear, that just means regular folks can't afford AI. API pricing is still a great bargain for businesses replacing generic office workers.

Replacement of generic office workers has been tried many times now and every time it has ended up in business catastrophe.

Flight had been tried many times, ending in crash and death each time until the tech got worked out. That it's been tried so many times shows the demand is enormous.

> That it's been tried so many times shows the demand is enormous

Capitalists want to pay their workers as little as possible or even not have workers and pocket close to 100% of the profits since forever. Whether they'll be able to do it this time is yet to be seen. Putting the tech itself aside, which is far from good enough yet imo, the only way to avoid a massive social backslash would be to use some of the additional profit they're able to keep because of AI for social spending. And I just don't think they realize yet that it's probably in their interest to do so if they want to replace 'human resources' with AI.

Agreed but the AI lab investors and CEOs aren’t ideologically aligned with concepts like UBI. Some of them have even commented on social media about the unsustainability of social security and need for cuts. Also why would you they do trillions in capex investments just to give their returns away.

They are clearly smart/dumb enough to think pushing labor replacement to 20%+++ unemployment levels would have no consequences for them personally. Maybe with their private jets, private islands, doomsday shelters, etc they are right.

Seems like a bit of a gamble to me, or probably anyone who has read a bit of history. But I’m not a billionaire.

Well, it is good enough for a huge percentage of what general administrative office staff are doing all day. Essentially copying information between textboxes, slightly rephrasing, clicking buttons, formulating emails to people to send in various documents, scheduling meetings, making reports that are fairly mechanical summaries of other information etc. Not all of it is like this, but probably in the range of 30-70%? Some of it may involve human judgment though, which is harder to automate, like whether some expenses are justified enough etc. But you can just replace those kinds of judgments with more clearly defined rules and cost limits etc.

Have you tried using "computer use" models? They just started to get good in the last few months. They can comfortably use GUIs, click the right buttons, enter the data faster etc. This is very significant and my programmer friends often don't appreciate the earth shattering consequences. The vast majority of software in such applications is legacy stuff or proprietary and does not and likely will never have a programmatic API. The only way to interact with it is via GUI. And an AI working that GUI can be deployed unilaterally by the customer of the software company who made the GUI program. This makes adoption and automation much easier. My prediction is that within 1 year (maybe 2) GUI-using AI will be much more powerful and it will be everywhere (of course it doesn't need to be displayed on a screen, it will do its virtual clicks based on a virtual screen buffer rendering in an isolated virtual machine, along with countless other agents in parallel).

There will be huge controversy around it, because this will be the first instance that regular people will see AI do work that resembles their day job all over social media. There will be panic. There will be new license terms saying only humans are allowed to operate the GUI, and AI use is either forbidden or costs extra. But this will not be checkable properly. Even when forbidden, some employees will pay from their own pockets to have cheap AI do the job instead of them.

And if the bubble does pop chaotically, the open weight inference providers will become even more viable as they'll be able to buy racks full of compute for pennies on the dollar.

Something very similar happened with the late 90s telecoms bubble, which collapsed a little before the dot com bust. There was a very nice living to be made by many small/mid-sized providers who were able to feed on the remains for about a decade afterwards.

Probably not. They aren't paying sustainable prices for their supply chain.

1. Datacenter builders are all up to their eyeballs in debt. The companies serving open weight models are getting their GPUs and power from somewhere, and we can infer that some of that is coming from the new build datacenters. But are those financially sustainable?

2. They are relying on subsidized giveaways of the underlying models by Chinese firms. But it's hard to see how that can be sustainable: it's being done for a reason and that reason is probably not commercial. If and when China stops doing this the open weight serving companies will get steadily less competitive with every passing month as the knowledge in the weights becomes obsolete.

Opinions are my own, etc.

Labs are innovating to bring token prices down. You can already see the focus shifting from frontier intelligence to frontier efficiency. Not to say models aren't improving, but improvement is slowing while token efficiency is increasing rapidly.

Commodified products competing on price does not bode well for valuations.

Huh? Part of the reason there is a tremendous boom in data-center construction is that there's currently quite high demand for inference. High demand means high prices. Even if the investors funding this construction all go bankrupt, the data centers don't just vanish: someone will buy them from the bankruptcy court and operate them. The effect of the current construction boom will be a huge increase the supply of inference, and of course increasing the supply will lower the price.

If one is in the data-center construction business or if one is a supplier of GPUs or other "inference hardware", then yeah, the current boom conditions are not "sustainable". But if one is a purchaser of inference, the tremendous investment our society is making in "inference capacity" ensures the the supply of inference is quite sustainable (barring a nuclear war or some other reason the data centers or the electrical infrastructure they depend on might be destroyed). And if the expectations for a rapid increase in the demand for inference turn out to be false, then that just means that the prices for inference will be nice and low for those who remain interested in purchasing it.

Open weight models don’t train themselves, someone needs to do it, and that costs a lot of money. I guess they’ll want to see an ROI.

Today's frontier models are strong enough to be useful even if no further progress is made. The knowledge cutoff is a bit inconvenient, but with web search and in context learning, it can pick up new library versions or programming languages just fine.

I’m sure we’re going to get open-weight equivalent of Astra in like two years at Luna price point. With that, I’ll be living in science fiction world.

Two years? Pessimist.

Sure, but if the companies start competing on price they become commodities and their valuations collapse, leading to knock-on effects everywhere (eg all the circular financing also collapses). I'm not saying this is a bad thing.

I agree that by what we are seeing so far, there is no magic secret sauce on the horizon. The tech is not super hard to understand and develop, such that hiring a few capable ex-employees of the big labs can help you build a new one. Any big step in capabilities has been quickly matched by the other labs so far, including agentic AI but also image and video generation.

Also, "circular financing" is a meme. A private company cannot create money, there is no circularity. NVIDIA may buy services from OpenAI and also sell them GPUs. If a lumberjack buys eggs from his neighbor and also sells him wood to improve his chicken coop, that's not circular financing, just business happening in both directions.

So why is mistral so far behind?

It's more like the lumberjack is lending money to the farmer so the farmer can buy his wood. (NVIDIA doesn't consume tokens). If the tokens (eggs) suddenly become much unprofitable then the AI labs aren't going to need as many GPUs (lumber for hen houses) and the borrowed money won't be paid back. Or perhaps more accurately, has even less chance of being paid back than it does now, which is not a high chance.

The numbers are staggering eg: https://www.cnbc.com/2026/08/17/nvidia-financing-open-ai-dat...

> Also, "circular financing" is a meme. A private company cannot create money, there is no circularity.

Tell me you know nothing about economics without telling me you know nothing about it.

Dear god. The pace of AI progress may slow down just enough to extract real value from these models.

That was not my point. I agree with you that producers of these models should be able to extract value from their models, I'm just surprised that everyone expects free models. There's a real cost in producing those.

> I'm just surprised that everyone expects free models. There's a real cost in producing those.

There's a lot of free stuff incurring "real costs" in its production, one more or less doesn't make a shred of difference.

> I agree with you that producers of these models should be able to extract value from their models.

Nobody is stopping them from extracting value but they fail despite being placed in a very advantageous position vis-a-vis others, which tells me, they should call it quits or go open like everybody sane.

Open is the way to go, nothing else works in the field of AI, that's the bitter lesson closed labs are yet to learn.

"Everyone sane goes open?" What a delusional take, do you really think open companies would remain open if they were in a lead?

A little bit like electricity plants. Countries will nationalise for sovereignty reasons, as China already does probably. And that’s not bad for people given current trajectory.

As I see it, price increases will lead to a huge loss of users, the two big companies will go bust, hardware manufacturers will have to slash prices, cancel increased capacity projects, etc.

And users will make do with what their newly bought cheap hardware will run, which I think is going to be more than enough for everyone that doesn't need to solve another millenium problem.

Right the most likely outcome to me is a pop and slow fizzle. Mega capex labs become uneconomic at the prices they need to charge to do more training runs. Non frontier cheap models become good enough at far less cost that people use that.

The use cases LLMs are useful for without lots of RF/fine tuning proves more limited than the expansive views currently held.

We get a generalized productivity boost the way the internet gave us, without any one sector or company being a winner take all of that productivity gain.

Reply on news.ycombinator.com