OpenAI Just Admitted Something Utterly Damning
Where's the money, Altman?
It’s Sunday morning; the burnt orange dawn pierces through the clouds, and the smell of fresh coffee fills the air. I settle down, the coffee gently reviving me as my dog wraps himself around my feet. Pure bliss. Stupidly, I decide this would be a great time to read the news, and my mini-nirvana is crushed. Amidst the background noise of the usual devastating Trump and Musk idiocracy that would make anyone with even a shred of empathy weep, one story stood out to me as being incredulously deceptive and damaging. OpenAI’s CFO gleefully announced that the company has a chance of reaching $11 billion in revenue this year. This implies that their catastrophic $3.7 billion revenue and $5 billion loss of 2024 and subsequent $6 billion investor bail-out won’t repeat itself again this year (read more here), right? Well, no! This all but proves that OpenAI is a fundamentally broken company that is doomed to fail. Allow me to explain.
This prediction seems to be based on OpenAI’s ballooning number of users, given that they now have 400 million weekly users, up from 300 million in December 2024. However, let’s not forget that DeepSeek’s market-breaking free-to-use R1 model forced OpenAI to unleash their latest o3-mini model for free to compete. Many users have switched from their paid plans to this, and while there have been no official figures reported yet, it seems the o3-mini is wildly popular. So, this significant leap in users does not mean a leap in revenue, and because OpenAI has a bit of a track record of exaggerating and keeping its cards close to its chest, I can’t help but feel this $11 billion figure isn’t accurate.
But let’s just assume it is, because even then, this prediction is a giant red flag.
OpenAI loses money on even its most expensive $200 per month plans; that is how expensive it is to run the models. It’s also why OpenAI would have posted a loss last year, even if they didn’t spend a penny on developing new models.
Because the AI race is still very hot, and the latest AI models have some significant flaws, OpenAI is having to develop more extensive and better models. Sadly, as AI is hitting a point of diminishing returns known as the efficient compute frontier (read more here), this also means that AI development costs are set to increase exponentially! OpenAI has said it is on track to spend $320 billion on AI development between 2025 and 2030, meaning that its development costs this year could easily quadruple last year’s to around $20 billion.
So, the fact that OpenAI has the possibility to reach $11 billion in revenue is nothing to celebrate. It’s a damning indictment that, even with users flooding to use their questionable services, they are getting further and further away from profit every year.
And it’s not just me who has noticed this. The Information has also analysed OpenAI’s predictions and found that OpenAI implies they will post losses of $14 billion in 2026! To give some context, some multinational banks posted smaller losses in 2008, resulting in their closure or bail-out.
I can’t stress enough how horrific it is that OpenAI is reaching for just $11 billion in revenue! This announcement should have utterly tanked the tech stock market. The fact that it hasn’t is profoundly worrying.
Some of you might say that this is just in the short term. OpenAI itself expects to reach profitability by 2029, when it plans to reach $100 billion in revenue. Well, I’m sorry, but even that argument looks more than a little bit ropey.
Take Project Stargate, a $500 billion joint venture to build AI data centres in the US, with OpenAI being the only customer. Data centres cost 3–5 times their build cost in operational costs over their 15-year lifespan, averaging out to an annual operational cost of 26% of their build cost. This means Stargate will cost $130 billion to run each year. And even this cost might be too optimistic, as Stargate is looking to build its own nuclear reactors to power itself, meaning its energy cost could be far higher than expected. Now, because OpenAI is the sole customer and isn’t putting any money towards building Stargate, it would be fair to assume that they will have to pick up this monumental tab.
Furthermore, data centre costs aren’t the only expense AI companies must worry about. In fact, they only make up around 40% of their expenses, with staffing, maintenance, and other infrastructure costs making up the rest. As such, by 2029, OpenAI’s annual expenditure could be well over $325 billion, even without any AI development costs.
This means that, even by their own optimistic predictions, OpenAI is set to reach record-breaking annual losses by the decade’s end.
Even worse, the AIs OpenAI could make in this period likely won’t be revolutionary at all. The efficient compute frontier and its diminishing returns (read more here) mean that even after this expenditure, these futuristic AIs will likely only be marginally better than those we have today.
So, in admitting that it only has a possibility of making $11 billion in revenue this year, OpenAI has exposed its dark secret. It is a shit show of market share before profit and hype. It makes a mockery of business fundamentals and the notion that our economy even vaguely represents the real world.
Make no mistake, the fact that the largest banks and investment groups in the world are willing to pour billions upon billions of dollars into OpenAI, despite its apparent inability to turn a profit, means that there is far less money being invested into the actually productive parts of the economy where you and I exist, starving us of opportunity and prosperity. This is billionaires making other billionaires wealthier at the expense of everyone else.
Thanks for reading! Content like this doesn’t happen without your support. So, if you want to see more like this, don’t forget to Subscribe and help get the word out by hitting the share button below.
Sources: pymnts, The Information, Tech Radar, Will Lockett, The Guardian, Will Lockett, Site Ltd, Optrium, Infiniti, DCD, OpenAI, IDC