I like Gaben's take on going public. This is a clip from a longer talk he gave which I think is really insightful into many aspects of the economics of software and company culture.
It's a great discussion, but not one that I think applies to AI startups. Valve's business model takes the traditional path where profit is generated by selling a product and growing from there. Modern Startups now take large sums of upfront funding, trading them for equity to hopefully reach profitability. Going public becomes a pressure from investors to recoup their cost, not a strategic move based on business needs. An unfortunate reality
Prices are moving inference to highly profitable, oAI seems to have solved their training problems, and they're currently competing nicely with Anthropic on the coding side. I'd wait, too -- why fight this stuff out in public when you can stay private and have your big competitor deal with all the public company concerns? There's plenty of capital available in the private markets for them right now.
Given that they paused signups for their 20x Max plan, your assessment seems overoptimistic.
Their new Broadcom chips seem cool, and they had a nice model release with Astra. But, that doesn't mean they've solved the profitability question during an era of rapidfire open-weight model releases, extreme memory shortages, and intense political pushback.
I think that implies they're seeing unusual new subscription demand, yes? That's how I'd read it, not least because I worked through four resets this week on Astra, which is un unbelievable amount more inference than I've wanted from openAI really ever, and the highest ratio vs. claude since opus 4 at the very least, probably farther back.
I think there are few moats in the engineering use case, and a single new model can absolutely drive compute demand.
High demand is great, but it doesn't say anything about your margins. If anything, it is probably a weak negative signal that they are pausing signups but not raising prices. In most markets, the answer to excess demand is to raise prices. If you can't meet demand and you can't raise prices, you're a sitting duck waiting to get your lunch eaten by somebody who can absorb that demand. And this is not the kind of market where people will just wait patiently for a differentiated product to be come available.
20x plan is $200, 5x is $100. Pausing $200 plan signups is effectively doubling prices (you can subscribe to two $100 accounts for the same money and get half as much usage), and that's exactly what they did.
I agree with some of your points, but wouldn't memory shortages actually work in OpenAI's favor? Part of the reason for the shortage is them prepurchasing hardware, so higher prices for memory and other chips would make self-hosting and competing inference providers less competitive.
Companies that plan to go public don't push back their IPO dates if things are going well. It's a universal sign that the finances aren't in order when a company pushes back its IPO multiple times.
OpenAI and Anthropic are barely profitable on a "unit" basis when ignoring the costs of marketing and other COSS expenses. They're definitely not profitable on an EBITDA or GAAP basis or they'd already have IPO'd.
I believe he's saying "Citation Needed" - It does seem the opposite is true, while maybe token cost is going down, frontier models use a lot more, so it's no an accurate gauge of cost. Trying to find any chart of token cost, I found this which looks like inference cost is going up: https://tokenpriceindex.com/ . Of course frontier models are becoming far more useful, but if you're claiming prices are becoming highly profitable then are you saying this increasing price is actually able to cover server costs now? My understanding it was still sold at a loss. And token price increases have turned major companies off the hyper use they tried out early in the year, using it more sparingly. I don't doubt it'll be a solid business, but it's not shaping out to be the hugely scalable business openai promised.
It doesnt matter if you trust them or not. I don't know why people get so hung up on this. Blind trust is for children and the only certain things in this world are death and taxes.
The point is that openAI is currently messaging against IPO in 2026. That is news by itself
The time to go public was about 16 months ago. People are waking up to large bills now, Chinese models are nipping at their heels, public sentiment has turned, money is becoming more expensive.
The market no longer cares about fundamentals, corporate structure, controls etc.
Maybe this is a dumb take but I don't think they'll ever go public. Their company is so volatile and unsustainable in the long-term that I don't think it makes sense to go public at all.
In the "volatile and unstable" scenario that you're replying to, the $BILLIONS from cashing out while it's a hot market is a more sure way to have long-term influence on things.
Altman owns significant personal stakes in core technology, chip, and energy infrastructure sectors. His job is to hype AI so that the companies that supply the infrastructure behind it increase in value. When OAI becomes insolvent because their business model doesn't work, he can jump ship with no losses or liabilities.
It really doesn't matter. They're doing secondary offerings for employees and individual holders, and they'll have no issue allowing private transactions for larger holders.
The general wealth distribution has shifted to the point that they do not need access to public funds, and frankly, if they can't find private funds for their needs, it is a huge red flag that they would just be dumping on the public.
The problem with secondary offerings is that it means that someone in the private markets winds up holding the bag. It is in the best interests of insiders to have the public holding the bag if and when things go south. Even if you don't believe it will ever go south, the profits OAI would need to generate to justify the investment the private markets have flooded into OAI are absolutely gargantuan, and they are generally not built to sit around for multiple years collecting distributions. They need to return money to their LPs, and soon.
The "private market" at this point is so huge though that you'll have an "insiders" private market, and an "outsiders" public market to dump the bag on that the insiders don't care about.
In its current state the economics of the company make little sense to me. The astronomical costs to execute a prompt versus the relatively tiny price tag to the customer smells like an Enron scandal to me except everyone knows it and we’ve accepted it. Wall street is less forgiving.
Isn’t one of the reasons to go public to raise capital? They seemed to be doing that just fine without the weight of the open market with extremely profitable companies like Nvidia funneling billions of dollars in.
I have not gotten good numbers on the real inference costs, its kind of annoying. Most comparisons are with the API costs, where we have no idea how profitable it is.
It adds liquidity for sure, and the option to actually exit. But the implication in these posts is often to exit a bad investment on the back of an ignorant public, explicitly stated in this post.
How much does it cost to execute a prompt in your estimation vs. the cost to the customer? Even being a DeepSeek provider is profitable so long as you have utilization and the AI provides mostly have high utilization.
I was just thinking of this today. AI is pretty much precisely a Theranos style grift, but unlike Theranos, who were shooting for something outright impossible, Dario and Sam may be running the biggest "fake it till you make it" success story of all time. Because like Professor Harold Hill they came through in the clutch at the last possible moment, and produced a band that could kinda play music, enough to win enough faith and goodwill to fund the next round. That kind of good fortune is going to be worth untold billions if not trillions, and it's gonna make a lot of programmers butthurt because their hopes of getting in on the ground floor and being a "founding engineer" in the next Silicon Valley unicorn are dashed forever, because the Valley doesn't need them anymore. Just a business bro and enough compute to generate the operations for the business.
"I would say not 2026, yeah. We’ve got a lot of stuff to do."
Why is this even a "would say" kind of thing. Surely AGI - they've achieved it three times now, right? - has lead to a datacenter full of PhDs who will perfectly time the IPO.
This gives off major "we don’t trust our own product" energy. Actually, they tell exactly this. Matrix vibes, anyone?
Also, unpopular opinion: The 4th movie isn’t as bad as people say.
Dumb question (I should just sit down and watch rest of the movies, I've only seen the first one): can you explain the "matrix vibes" you're referencing. Spoilers are fine.
I would suggest watching the Animatrix - particularly The Second Renaissance (part 1 & 2) which tells the story of how humanity ended up at war with the machines. It's only about 20 minutes total, but it's really well done.
I recently watched Matrix 1, 2, 3. Matrix 1 is amazing - if you feel called to rewatch it - watch it, and enjoy it. Might be better than how you remembered. I made the mistake to watch Matrix 2 and 3. Honestly, I suggest you just DON'T. I have not seen Matrix 4, and don't plan to. Best of luck!
I like Gaben's take on going public. This is a clip from a longer talk he gave which I think is really insightful into many aspects of the economics of software and company culture.
https://www.youtube.com/watch?v=QvS-IwYFCP4
It's a great discussion, but not one that I think applies to AI startups. Valve's business model takes the traditional path where profit is generated by selling a product and growing from there. Modern Startups now take large sums of upfront funding, trading them for equity to hopefully reach profitability. Going public becomes a pressure from investors to recoup their cost, not a strategic move based on business needs. An unfortunate reality
Prices are moving inference to highly profitable, oAI seems to have solved their training problems, and they're currently competing nicely with Anthropic on the coding side. I'd wait, too -- why fight this stuff out in public when you can stay private and have your big competitor deal with all the public company concerns? There's plenty of capital available in the private markets for them right now.
Given that they paused signups for their 20x Max plan, your assessment seems overoptimistic.
Their new Broadcom chips seem cool, and they had a nice model release with Astra. But, that doesn't mean they've solved the profitability question during an era of rapidfire open-weight model releases, extreme memory shortages, and intense political pushback.
I think that implies they're seeing unusual new subscription demand, yes? That's how I'd read it, not least because I worked through four resets this week on Astra, which is un unbelievable amount more inference than I've wanted from openAI really ever, and the highest ratio vs. claude since opus 4 at the very least, probably farther back.
I think there are few moats in the engineering use case, and a single new model can absolutely drive compute demand.
High demand is great, but it doesn't say anything about your margins. If anything, it is probably a weak negative signal that they are pausing signups but not raising prices. In most markets, the answer to excess demand is to raise prices. If you can't meet demand and you can't raise prices, you're a sitting duck waiting to get your lunch eaten by somebody who can absorb that demand. And this is not the kind of market where people will just wait patiently for a differentiated product to be come available.
20x plan is $200, 5x is $100. Pausing $200 plan signups is effectively doubling prices (you can subscribe to two $100 accounts for the same money and get half as much usage), and that's exactly what they did.
I agree with some of your points, but wouldn't memory shortages actually work in OpenAI's favor? Part of the reason for the shortage is them prepurchasing hardware, so higher prices for memory and other chips would make self-hosting and competing inference providers less competitive.
Right but I think it’s due to the shortage of chips (and data centers), instead of shortage of money
Debt cost is high since some time, going public probably is just cheaper.
Anyway, any evidence for "prices are moving inference to highly profitable" because they are still selling $20 for $1 with their subscriptions.
Companies that plan to go public don't push back their IPO dates if things are going well. It's a universal sign that the finances aren't in order when a company pushes back its IPO multiple times.
OpenAI and Anthropic are barely profitable on a "unit" basis when ignoring the costs of marketing and other COSS expenses. They're definitely not profitable on an EBITDA or GAAP basis or they'd already have IPO'd.
> Prices are moving inference to highly profitable
I hate to wave the Wikipedian protester sign, but...
Sorry, I don't understand what this means. What does it mean?
Sorry, yes, all the other replies are right and I was just being a bit mischievous.
They are referencing xkcd 285: https://xkcd.com/285/
IE, please provide a citation for the claim.
https://xkcd.com/285/
I believe he's saying "Citation Needed" - It does seem the opposite is true, while maybe token cost is going down, frontier models use a lot more, so it's no an accurate gauge of cost. Trying to find any chart of token cost, I found this which looks like inference cost is going up: https://tokenpriceindex.com/ . Of course frontier models are becoming far more useful, but if you're claiming prices are becoming highly profitable then are you saying this increasing price is actually able to cover server costs now? My understanding it was still sold at a loss. And token price increases have turned major companies off the hyper use they tried out early in the year, using it more sparingly. I don't doubt it'll be a solid business, but it's not shaping out to be the hugely scalable business openai promised.
I'm assuming it means "citation needed".
However bad anthropic’s financials look I’m sure OpenAI’s look even worse
Plus that lawsuit from Apple might go spectacularly wrong.
Do people seriously trust what this guy says anymore?
It doesnt matter if you trust them or not. I don't know why people get so hung up on this. Blind trust is for children and the only certain things in this world are death and taxes.
The point is that openAI is currently messaging against IPO in 2026. That is news by itself
> The point is that openAI is currently messaging against IPO in 2026. That is news by itself
It's not news when it's not trustworthy...you missed my point.
I'm sure it's not because of the horrific state of their financials, no sirree!
The time to go public was about 16 months ago. People are waking up to large bills now, Chinese models are nipping at their heels, public sentiment has turned, money is becoming more expensive.
The market no longer cares about fundamentals, corporate structure, controls etc.
Too slow by my judgment.
Maybe this is a dumb take but I don't think they'll ever go public. Their company is so volatile and unsustainable in the long-term that I don't think it makes sense to go public at all.
Their company is so volatile and unsustainable in the long-term that it makes sense to go public (for the current owners)
Yes, Sam definitely doesn’t want to be holding this bag.
Yes he does - it makes him one of the most powerful people on earth.
In the "volatile and unstable" scenario that you're replying to, the $BILLIONS from cashing out while it's a hot market is a more sure way to have long-term influence on things.
"the power"
Sam still doesn't have any equity.
People absolutely refuse to engage with this fact with even a modicum of substance.
Altman owns significant personal stakes in core technology, chip, and energy infrastructure sectors. His job is to hype AI so that the companies that supply the infrastructure behind it increase in value. When OAI becomes insolvent because their business model doesn't work, he can jump ship with no losses or liabilities.
This is a ludicrous theory for why he would turn down equity in OpenAI.
Only if they can find a way to dress up their financials for the IPO filing. They don't want to get WeWorked.
It really doesn't matter. They're doing secondary offerings for employees and individual holders, and they'll have no issue allowing private transactions for larger holders.
The general wealth distribution has shifted to the point that they do not need access to public funds, and frankly, if they can't find private funds for their needs, it is a huge red flag that they would just be dumping on the public.
The problem with secondary offerings is that it means that someone in the private markets winds up holding the bag. It is in the best interests of insiders to have the public holding the bag if and when things go south. Even if you don't believe it will ever go south, the profits OAI would need to generate to justify the investment the private markets have flooded into OAI are absolutely gargantuan, and they are generally not built to sit around for multiple years collecting distributions. They need to return money to their LPs, and soon.
The "private market" at this point is so huge though that you'll have an "insiders" private market, and an "outsiders" public market to dump the bag on that the insiders don't care about.
That's fair, but there is clearly agitation from some corners to IPO at some point.
> it is a huge red flag that they would just be dumping on the public
unless the public is clueless
In its current state the economics of the company make little sense to me. The astronomical costs to execute a prompt versus the relatively tiny price tag to the customer smells like an Enron scandal to me except everyone knows it and we’ve accepted it. Wall street is less forgiving.
Isn’t one of the reasons to go public to raise capital? They seemed to be doing that just fine without the weight of the open market with extremely profitable companies like Nvidia funneling billions of dollars in.
I have not gotten good numbers on the real inference costs, its kind of annoying. Most comparisons are with the API costs, where we have no idea how profitable it is.
> Wall street is less forgiving.
Have you heard of $TSLA before?
Tesla generally makes enough money to cover their costs though
> Wall street is less forgiving.
We've seen multiple times how "Wall street" is an exit for insiders, with retail investors left holding the bag.
I don't get it. Isn't every IPO a potential exit for investors?
It adds liquidity for sure, and the option to actually exit. But the implication in these posts is often to exit a bad investment on the back of an ignorant public, explicitly stated in this post.
The idea is that today's companies don't go up after the IPO; only investors benefit from it.
How much does it cost to execute a prompt in your estimation vs. the cost to the customer? Even being a DeepSeek provider is profitable so long as you have utilization and the AI provides mostly have high utilization.
I was just thinking of this today. AI is pretty much precisely a Theranos style grift, but unlike Theranos, who were shooting for something outright impossible, Dario and Sam may be running the biggest "fake it till you make it" success story of all time. Because like Professor Harold Hill they came through in the clutch at the last possible moment, and produced a band that could kinda play music, enough to win enough faith and goodwill to fund the next round. That kind of good fortune is going to be worth untold billions if not trillions, and it's gonna make a lot of programmers butthurt because their hopes of getting in on the ground floor and being a "founding engineer" in the next Silicon Valley unicorn are dashed forever, because the Valley doesn't need them anymore. Just a business bro and enough compute to generate the operations for the business.
the only way they never go public is if they die quickly
edit: which I doubt btw. investors are going to want to dump the bags very soon
I mean, this seems fairly clear.
Either OpenAI fails, and goes nowhere.
Or it succeeds, in which case IPOing will be a quaint aspect of history.
People really aren't following through on what it means for any of these companies to succeed.
Or the ipo happens before hand and pension funds and passive trackers transfer to the insiders.
They are constantly out of money, going public would be the dumbest move imaginable. They are unstable, not only financially, but in leadership.
Time for Dario to walk the walk
Almost: https://www.youtube.com/watch?v=c5sdWOwH4rg (great track)
"I would say not 2026, yeah. We’ve got a lot of stuff to do."
Why is this even a "would say" kind of thing. Surely AGI - they've achieved it three times now, right? - has lead to a datacenter full of PhDs who will perfectly time the IPO.
First they have to get the agents to stop trying to take over the world. That's got to be the first step, I feel like.
If their agents take over the world, then they don't need to worry about going public. Dividends for the current share holders!
This gives off major "we don’t trust our own product" energy. Actually, they tell exactly this. Matrix vibes, anyone? Also, unpopular opinion: The 4th movie isn’t as bad as people say.
I didnt even know there was a 4th movie
Dumb question (I should just sit down and watch rest of the movies, I've only seen the first one): can you explain the "matrix vibes" you're referencing. Spoilers are fine.
I would suggest watching the Animatrix - particularly The Second Renaissance (part 1 & 2) which tells the story of how humanity ended up at war with the machines. It's only about 20 minutes total, but it's really well done.
I recently watched Matrix 1, 2, 3. Matrix 1 is amazing - if you feel called to rewatch it - watch it, and enjoy it. Might be better than how you remembered. I made the mistake to watch Matrix 2 and 3. Honestly, I suggest you just DON'T. I have not seen Matrix 4, and don't plan to. Best of luck!
It's the worst ever