OpenAI is losing money on its pricey ChatGPT Pro plan, CEO Sam Altman says | Hacker News Reader

OpenAI is losing money on its pricey ChatGPT Pro plan, CEO Sam Altman says

techcrunch.com

24 pointsmfiguiere11 comments

When the pricing first was announced for ChatGPT Pro I had the hot take of "it's possible from OpenAI's perspective that $200/mo is too low." I can't believe that turned out to be correct.

> “I personally chose the price,” Altman wrote in a series of posts on X, “and thought we would make some money.”

I can't believe the biggest companies apparently set prices due to vibes. That explanation is much worse than "we ran the spreadsheets and there's no way to price ChatGPT Pro such that we can both a) make money on it on average and b) not completely price people out of it."

> I can't believe the biggest companies apparently set prices due to vibes.

Spend some time in business and you'll learn that pricing is entirely vibes. If you start from first principles, you'd just take the cost to make the product, add a fixed percentage of that, and make your fortune, and that works in a limited amount of places. The real world is messier than that though, so one fixed formula won't cover all your bases.

Exactly, it's also a marketing thing. Even if ChatGPT should actually sell for $20/m (it shouldn't), the chance of Pro needing to sell at exactly a round number 10x for $200/m is near zero. The price is part of the UX, part of the marketing, part of product.

If your first primary goal with a product line is that it will turn a profit per sale from the start, then you kinda have to start with the cost of goods sold as your baseline. If a customer pays you less than the product or service costs to provide, then by definition you priced it wrong for that initial / first step goal.

You can make up whatever goal you want and succeed or fail at it, that doesn't have any bearing on the vibes-based pricing being how to price things. What you'd need to somehow show is that a) you can even sell the product at that price, and b) that setting the price be higher or lower wouldn't result in greater total profit.

Naturally, you do want to make profit from each sale, but first off, can you, and secondly, how much? If it's something that took years of R&D to develop, then you'll want to amortize that over a longer period of time, and not try and recoup that the instant you have something to sell. If you're dropshipping someone else's product, then you can't price it all that high because your competition is also going to be pricing it with fairly low margins. (Or they're not, and you can undercut them.)

Interestingly, in support of vibes-based pricing, and in direct contraction of Econ 101, sometimes raising the price on something increases it's percieved value, as being too cheap makes it seem like a worse value or lower quality.

You can't "run the spreadsheets" on a business with >0 fixed costs. You'd have to be able to predict future demand to know per-customer cost.

Pricing is always based on intuition.

I almost laughed. It's been years.

> I can't believe the biggest companies apparently set prices due to vibes.

I mean, in general they _very much do not_. What you will see sometimes is setting a price and then building to the price (for instance Apple has, for a long time, tended to go for $999 for its cheapest laptop), but "one guy just makes up a price for a thing that has an inherent cost disconnected from that price" is not ordinarily a thing.

Pricing is the alpha and the omega. OpenAI would be getting some very interesting questions today about its cost metrics and margins if it was a public company.

release all your sources of current (dark) and planned income and we might believe that's relevant.

the company knows how much (interactional) data it is collecting and how it will monetize on that in the future or rather, how fucking good their product will become once it's hardware needs match adequate enough capacities to be run privately.

I'd go pro again and pay 50 bucks per month if I could trust the company. BUT THERE IS NO WAY

Bonjour

Reply on news.ycombinator.com