Hacker News Reader

Comment by andy99 | Hacker News Reader

IPO is to leave someone holding the bag, right? The narrative has really changed from AGI, the most consequential technology, blabla. If I had that I would keep it private. If I wanted to cash out, I’d do an IPO. Is there a narrative that doesn’t make it seem like pump and dump?

View on HN

I was watching the OpenAI livestream yesterday.

Sam was saying they have a $1.4 tn financial obligation to build 30 GW worth of data centers https://youtu.be/ngDCxlZcecw?t=1179

That's a lot of money to raise given they have no earnings.

He also says they are talking about being able to build 1 GW a week at a cost of $20bn/GW

I find it all a bit worrying. Weren't we supposed to be careful with energy use to reduce CO2 emissions? Could the billions maybe better deployed to schools hospitals, housing and the like rather than chatbots?

Rather than a fictional paperclip maximising AI eating the world we have a chatbot maximising Altman trying to do so?

>Weren't we supposed to be careful with energy use to reduce CO2 emissions?

The Programming has changed get with the times. AI earnings have outweigh ESG scores.

You didn't notice as soon as energy was needed for AI, all the tech companies gave up on green power and reducing emissions?

Nuclear doesn't reduce emissions?

It does, but its partly just the hot thing because you can increase energy usage without going back on CO2 promises.

They certainly arent reducing energy usage rn which could theoretically be a puzzle piece of reducing CO2.

They have 800M weekly active users that have yet to be monetized but enormous capital costs. It makes sense they'd be looking to raise large amounts of money in an IPO.

How many of these users will be able to be monitized is an unanswered question as of yet however.

API usage on the other hand has a clear path to monetization and as of right now Anthropic is looking a lot better than OpenAI there.

Everyone. People are becoming dependent on chatgpt. They literally cannot function professionally or even socially without it. They will pay their last 20-30 dollars if needed. It's literally like a drug especially when it's asking you if you want to followup/continue.

Everyone also uses Google, YouTube or Instagram. No one would ever pay for any of it though and it is financed through ads. So far it is unclear, if this is also a viable option for chatgpt.

YouTube Premium has over 125 million subscribers.

Out of about 2.7 billion users. So about 5% of all users are subscribed to YouTube.

If the same were true then Open AI at 1 billion weekly users they should have about 50 million subscribers. Right now that numbers sits at 20 million though and growth is slowing down. [1]

So people are more willing to pay for YouTube than ChatGPT and that is ignoring that YouTube is still largely relying on ad revenue and can only allow itself to have more and more ads as there are no alternatives. Open AI has plenty of competitors that would love to offer users free access if ChatGPT were to start showing you ads.

[1] https://fortune.com/2025/10/14/openai-subscriptions-flatline...

> Everyone also uses Google, YouTube or Instagram. No one would ever pay for any of it though and it is financed through ads.

A lot of people already pay for YouTube since the introduction of Premium. Google/Facebook don't push for paid versions of these products because the data from billions of "free" users is more valuable to them than payments from millions of paid users.

If Google search were paywalled (pre-AI) the most likely outcome would be a separation of consumers into "premium" customers paying for Google, some people paying for cheaper but not quite as good alternatives, and everyone else getting by with the free alternatives. There would also likely be some kind of enterprise tier for indexing your corporate resources or some such.

There's a reason Adobe is still extracting billions from its ~~victims~~ users despite many great free (or reasonably priced) alternatives existing.

At least one of us is inside a bubble. Nobody I interact with regularly uses chatgpt for anything more than novelty. Even people who used it as glorified google for looking up things reduced their use.

There are definitely big?bubbles where everyone has outsourced their thinking to AI. I’d like to think it’s mostly at the lower end of “knowledge work” - think Deloitte, but it seems that even people / orgs that you would expect more critical thinking of are using it uncritically.

Of course this all occurs in a very small segment of society, I think the majority of people don’t really use it, and certainly haven’t moved any of their day-to-day thinking over to it.

How do you explain that not even 5% of users are on paid subscriptions?

This has nothing to do with what i said. I said they are addicted. The free limits are designed this way. If openai suddenly removed the free plan, i guarantee you a lot of people would buy. They dont have an alternative they cannot think independently anymore

monetization is straightforward - the only question is who gets the money and which business models succeed.

OpenAI burns <5B$/year in inference. They're raising that money to hyperscale larger models.

A capital investment is predicated on there being future profits. I don't understand why they can't be profitable already. If 800M AU is not a critical mass for flipping on the revenue switch, I can't imagine what is.

So what stops them from monetizing those users today? Why would it be any different in the future if they can't?

I say further limit the free tier and more aggressively push those free users to a paid plan. Raise the prices for Business users and API access. An IPO isn't going to raise the trillion they need to keep running off capital.

Once OpenAI reaches the eventual pricing needed to break even, I suspect we'll see that it no longer makes sense for many of their customers to replace humans with AI after all. As it stands now, their investors are essentially paying OpenAI to put employees of other businesses out of jobs by masking the true costs. The sooner they can reach the sustainable pricing phase the better.

800m users, 25m paying subscribers

LLMs have been commodified... Why would I pay for chatgpt when I can rotate free accounts on all providers and essentially never run out of tokens?

In the before times, companies went public to raise capital in order to grow their business.

Does that seem like what’s happening here?

Maybe? They want to grow the company, and they've run out of VC and circular financing money. Are they going to use the capital raised to make actual profits? That's the question.

If they’re on the path to AGI, there’s probably a predictable cost structure to get us to the point where these models are self-reinforcing and we get fast takeoff. This is the dumbest they’ll ever be remember.

It hasn't been like that for quite some time. The corpse is sold off to mutual funds and pension plans

IPO day would be a great time for someone else to release a revolution of an open model to the public...

Yeah, it's a pump and dump. You wouldn't sell AGI for any amount of money. You'd literally be a god king. You could build an army of sentient robots and just seize entire countries.

Reply on news.ycombinator.com