Transcription
[Music] In October of 2025, OpenAI conducted a share sale which valued the company at $500 million. This makes it the most valuable privately held startup in history.
The company's reportedly planning to conduct an IPO in either late 2026 or early 2027. They're planning to price that IPO at a $1 trillion valuation and hope to raise more than $60 billion from the listing. They will most definitely need this money because the company is a cash incinerator. In previous videos, we've talked about how the company is losing a lot of money. Up until now, we've been relying on sketchy estimates and media reports. But now, we finally have some real data revealing how much OpenAI is losing.
Open AAI's first large corporate backer was Microsoft, which made an inaugural $1 billion investment in 2019. To date, Microsoft has invested over 11 billion into OpenAI, mostly in the form of credits to its Azure cloud computing service. OpenAI was originally conceived as a nonprofit entity, but they transitioned to a for-profit company. This simplifies the corporate structure, and we now know that Microsoft owns 27% of OpenAI's outstanding shares. OpenAI is now classified as an equity affiliate in Microsoft's financial statements. That means that Microsoft has to report its share of OpenAI's profit or losses each quarter. For the 3 months ended September 30th, 2025, Microsoft recognized $4.1 billion of losses from its OpenAI investment. That $4.1 billion is Microsoft's 27% share of the loss. So in the third quarter alone, OpenAI lost 15.2 billion.
We don't know for sure how much revenue OpenAI is generating, but according to media reports, they generated $4.3 billion of revenue in the first half of 2025. Their revenue is growing quickly and is projected to be 12.7 billion for the full year. So they expect to make $8.4 billion in the second half of the year. If we assume a somewhat linear rate of revenue growth, maybe they'll make $3.5 billion in the third quarter and $4.9 billion in the fourth quarter. So in the third quarter, OpenAI probably incurred a net loss more than four times greater than its revenue. Open AAI spends an obscene amount of money on computing power to develop and run its AI models. To date, we've seen no clear path to profitability.
The company's most recent product is an AI video generator called Sora 2. This product has raised even more questions about the long-term viability of the company's business model. In conjunction with the launch of Sora 2, they also released a mobile app that lets users create and share AI generated videos. You can think of the Sora app as basically a worse version of Tik Tok full of AI slop videos. Generating these videos is extremely computensive. Open AAI is likely incurring billions of dollars of computing costs to operate Sora while generating almost zero revenue. In this video, we'll try to understand OpenAI's increasingly bizarre business strategy and what we can expect going into the company's expected IPO next year.
This video is brought to you by Mumu, a one-stop trading platform trusted by over 26 million users across the US and globally. Mumu offers zero commissions on stocks, ETFs, and options trades, free level two data, and real-time option chain data. What differentiates Mumu from other brokerages is it provides a powerful all-in-one investment platform. For example, let's look at Apple. If we go over to the fundamentals tab, we can see that Apple's price to earnings ratio is high compared to its historical range, indicating that the stock might be overvalued. We can see their smartphone market share, their revenue broken down by segment, their reported financial results versus analyst expectations, and a bunch of financial data. It's super useful to be able to kickstart your investment research all within the Mumu app.
Currently, Mumu is running an amazing new user promotion. They offer a 3.6% yield on your idol cash, but for the first three months, they offer an additional 4.5%, bringing your total yield up to 8.1%. Additionally, if you open a new account and deposit $100, they'll give you a $20 fractional share of Nvidia for free. The more you deposit, the more Nvidia shares they'll give you, up to $1,000. This promotion only lasts until the end of the year, so make sure to sign up today with the link in the description below.
OpenAI released this video generation service called Sora in December of 2024. At first, it was only available to people who already subscribed to ChatGpt Plus or Pro. It was available on the desktop and allows you to make videos based on text prompts. At first, Sora didn't get that much attention. You can make short clips of woolly mammoths walking across a desert and other gimmicky things like that. But there weren't very many obvious use cases. Not that many people used it because it was only available to paying ChatGpt subscribers.
On September 30th, 2025, OpenAI launched Sora 2. Sora 2's technical capabilities are much greater than Sora 1. Sora 2 can generate longer and higher fidelity videos with synchronized audio, realistic motion, and persistent characters. But perhaps the most important new development was a Sora app. Concurrently, with the release of Sora 2, OpenAI launched the Sora app. The Sora app has a Tik Tok-like scrolling feed. The Sora app is free to download. Anyone can download it, start generating AI videos, and post them on the app for other people to view. The app quickly went viral, hitting 1 million downloads in less than 5 days. People went crazy making videos of Spongebob getting arrested, Michael Jackson hanging out with a bear, and things of that nature.
When the Sora app first launched, Open AAI implemented almost zero content controls. This soon caused controversy and risked putting OpenAI in legal jeopardy. At first, some of the most viral videos on Sora were very offensive. We won't share them here, but things like Stephen Hawking getting beat up in a boxing match. Within two weeks, OpenAI had to block users from depicting Martin Luther King Jr. because so many people were making offensive and racist videos with his likeness. Another issue is copyright. People made videos with Spongebob, Mario, Pikachu, etc. These characters are copyrighted and Sora should not be allowed to make them without permission. A few weeks into the launch, OpenAI made an update blocking users from depicting copyrighted characters. We can see here that OpenAI is following its long-standing strategy of breaking the rules first and asking for forgiveness later. It was predictable and inevitable that people would depict copyrighted characters and make offensive depictions of both living and deceased celebrities. It was predictable that these types of videos would gain the most virality. Open AAI made the decision to initially launch the Sora app with zero safeguards in place. This allowed Sora to gain huge amounts of attention and virality in the first couple of weeks. Open AAI later says, "My bad. We didn't think people would abuse the platform." And they eventually put in place the safeguards. But of course, they knew people would abuse the platform. They wanted this to happen.
Openai CEO Sam Alman has a website called sam alman.com where he writes blog posts. On October 3rd, a few days after the launch of Sora, he made the following post. quote, "We are going to have to somehow make money for video generation. People are generating much more than we expected per user, and a lot of videos are being generated for very small audiences. We're going to try sharing some of this revenue with rights holders who want their characters generated by users. The exact model will take some trial and error to figure out, but we plan to start very soon." Our hope is that the new kind of engagement is even more valuable than the revenue share, but of course, we want both to be valuable."
Alman's post is delusional for multiple reasons. Firstly, he says engagement on Sora is going to be valuable for rights holders. The type of engagement brands can hope to receive on Sora is in no way valuable. Do you think Nintendo appreciates the thousands of videos that are made of Pikachu getting arrested? Does this help support their family-friendly brand image? Then he says they'll share revenue with the copyright owners if they allow their characters to appear in Sora videos. If you're going to share revenue, you need to have some revenue to share. Currently, Sora generates almost zero revenue and there's no obvious path to monetization. Sora allows any user to generate 30 videos per day for free. There are no ads on Sora. It's unlikely that advertising will ever be a viable revenue model for Sora. No self-respecting brand would want to advertise next to AI slop.
Just to see what's what, I downloaded the Sora app and tried to make some videos. I messed around with it for about an hour and used up the 30 free generations it gave me for the day. I was able to make the following video. >> These Sora videos cost us 40 million last month and we got what 12 views. The server bill alone is bigger than the revenue line. This is insane. Why did we think people wanted a 2-hour generative skating dragon movie? Nobody's watching it. I >> I tried making this video more than a dozen times, each with a slightly modified prompt. This is the best one it made, and it wasn't even that good. I didn't post it, but even if I did, it probably would have gotten almost zero views.
Open AAI incurs massive computational costs to operate Sora. So, how much money did OpenAI lose from me trying to make my crappy Sam Alman video? OpenAI partners with Microsoft's cloud service Azure to sell Sora capabilities to external customers. If you have your own software or website, you can pay to integrate Sora on the back end. By looking at the external prices on Azure, we can estimate how much computational cost it takes to make each Sora video. Even if you use the worst low res Sora model, it costs 10 cents per second. Sora generates 10-second videos, so that's $1 of computing costs. I maxed out my daily limit, so I incurred $30 of cost to OpenAI. At least based on Azure's pricing to third party customers. I didn't even post my Sora video, so nobody watched it. But let's say I had posted it and somehow it got 1 billion views within the Sora app. OpenAI would still generate zero revenue because there are no ads. Currently, OpenAI is probably losing hundreds of millions of dollars per month just by operating the Zora app.
So, what is the plan? How does OpenAI ever intend to make money from Zora? The OpenAI executive in charge of Sora is a man named Bill Peebles. He made a Twitter post trying to explain, quote, "We are launching the ability to buy extra generations on Sora. We've been quite amazed by how much our power users want to use Sora, and the economics are currently completely unsustainable. We thought 30 free generations per day would be more than enough, but clearly we were wrong. This will let our procreators get as much usage as they want to pay for. We imagine a world where rights holders have the option to charge extra for cameos of beloved characters and people. We will soon pilot monetization, prioritizing people and companies who got onto the platform early. This will also be a way for a new generation of Sora creators to make money."
First off, this guy's a senior executive at a massive tech company, probably getting paid millions of dollars, but he doesn't know how to capitalize the first word of each sentence he writes. Poor grammar aside, there are a few aspects of a statement that make no sense. He's amazed by how much the power users want to use Sora. How does he know that? I maxed out the daily limit of 30 Generations to make my crappy Sam Alman video. Does that make me a power user? A lot of people did make Sora videos just to mess around like I did. Almost nobody will pay real money for this. The plan seems to be that the so-called power users want to generate Sora videos so much that they'll be willing to pay real money to generate more than 30 videos per day. And brands like Nintendo are going to be willing to license their valuable intellectual property. The power users are going to pay extra money so that they can make AI slop videos of Pikachu being arrested. This whole plan is like a bad joke.
People says they will soon start monetization. Presumably, he means they'll show ads on Sora. It is inconceivable that Sora can generate enough ad revenue to cover its massive computing costs. Short form video ads are notoriously cheap. For example, Tik Tok pays its creators between 40 cents and $1 per 10,000 views they receive. This is a pittance. Sora will have a much more difficult time monetizing than Tik Tok. No self-respecting advertiser will want to associate their brand with AI slop. There are many Tik Tok influencers who make a lot of money. They build up a large following which allows them to land lucrative affiliate marketing and brand sponsorship deals. This is not possible on Sora. You're not going to become an AI sloth influencer. There's no incentive for anyone to spend real money to generate Sora videos.
Sam Alman has long made grandiose claims about the economic potential of his company's AI. For example, he says AI will one day cure cancer. Open AAI wants to raise trillions of dollars to build massive AI data centers. Alman justifies this investment by saying they will eventually achieve a vague concept called artificial general intelligence or AGI. AGI will be so revolutionary that it will be worth a quadrillion dollars, thereby making up for the massive losses the company has incurred to date.
This past August, Sam Alman went on CNBC where he answered questions about OpenAI's losses. >> The company's still losing money and obviously you're building building Stargate among them. What is the path to profitability? When do you see the lines cross if you will? Um, I mean, I think it could happen sooner than I originally thought if we wanted it to, but it seems to me like the right thing to do is to just keep investing in compute and and continuing to grow. >> But is that >> continue to like grow our training compute budget? >> Is that a couple years from now then, you think, or you're saying if you stopped investing? No, I think we I think well I think we should be willing to keep growing to to keep investing in in training compute for a long time like as long as we're on this very steep curve of the models getting better and better. Uh I think the rational thing to do is to just be willing to run at a loss >> uh for quite a while and continue to do that which is >>
Alman tries to justify OpenAI's losses by saying they're investing in training compute. The models are getting better and better at a very fast rate. No doubt OpenAI spent a lot of money to develop the Sora 2 model. And maybe this research somehow gets them closer to AGI. But allowing millions of people to make Sora videos for free, all at OpenAI's expense. This does not serve any useful purpose. Why do they create the Sora app?
The Sora launch was widely clowned across social media. For example, a Twitter user called Runner to Charard made the following post. Quote, Sam Alman two weeks ago, we need $7 trillion and 10 gawatt to cure cancer. Sam Alman today we are launching AI slop videos marketed as personalized ads unquote. Sam Alman is very active on Twitter and he replied to this post quote. I get the vibe here but we do mostly need the capital for build AI that can do science and for sure we are focused on AGI with almost all of our research effort. It is also nice to show people cool new tech products along the way make them smile and hopefully make some money given all that compute need. When we launched chat GBT there were a lot of who needs this and where is AGI. reality is nuanced when it comes to optimal trajectories for a company." It seems like it's against company policy to use proper grammar in a tweet.
Supposedly, Sora is intended to make profits and they'll reinvest those profits to achieve AGI and cure cancer. But does Sam Alman really expect Sora to be profitable? On October 8th, Sam Alman appeared as a guest on the A16Z podcast. A16Z is a venture capital firm that is a shareholder in OpenAI. During the podcast, Alvin may have inadvertently told the truth. >> Which bets would you say are enablers of AGI versus which are sort of hedges against uncertainty? I >> think you could say that on the surface, Sora, for example, does not look like it's AGI relevant. But I would bet that if we can build really great world models, that'll be much more important to AGI than people think. There were a lot of people who thought CHBT was not a very AGI relevant thing. It has been very helpful to us not only in building better models and understanding how society wants to use this but also in like bringing society along to actually figure out man we got to contend with this thing now. We for a long time before touchp we would talk about AGI and people were like this is not happening or we don't care and >> then all of a sudden they really cared and and I I think that so research benefits aside I'm a big believer that society and technology have to co-evolve. It's you can't just drop the thing at the end. It doesn't work that way. It is it is a sort of >>
Sam Alman is a big believer that society and technology have to co-evolve. What does that mean? OpenAI is losing a lot of money. As we explained at the beginning of this video, in the third quarter alone, the company lost more than $15 billion. These losses can be partially attributed to research and development as they build new AI models. But they also lose money operating their existing models. The ongoing computational cost of running ChatGBT is so great that they lose money even on the pro subscribers who pay $200 per month. From a pure business perspective, ChatGBT has been a commercial failure. It cost billions of dollars to develop. They're never going to recoup this investment because they lose even more money just to operate it. But ChatGpt did kick off the AI investment craze, allowing OpenAI to raise $60 billion from venture capitalists. From this perspective, it was a huge success.
Now, OpenAI's cash burn is so huge that they will soon exhaust the pool of available venture capital. They need to keep retail investors excited for their upcoming IPO, where they're looking to raise at least $60 billion. Open AAI needs to maintain the appearance that they're growing and advancing regardless of the unit economics and unit economics don't matter because they're eventually going to develop the mythical AGI. But is it really true that OpenAI's AI models are increasing at a very steep rate?
In March of 2023, they launched ChatGBT4. ChatGBT5 was launched in August of 2025, more than 2 years later. Sam Alman hyped up ChatGpt 5 to a grandiose degree. In January of 2025, Alman wrote the following on his website. Quote, "We are now confident that we know how to build AGI as we have traditionally understood it. We believe that in 2025, we may see the first AI agents join the workforce and materially change the output of companies. We continue to believe that iteratively putting great tools in the hands of people leads to great, broadly distributed outcomes. We are beginning to turn our aim beyond that to super intelligence in the true sense of the word. We love our current products, but we are here for the glorious future. With super intelligence, we can do anything else. Super intelligent tools could massively accelerate scientific discovery and innovation well beyond what we are capable of doing on our own and in turn massively increase abundance and prosperity."
In this blog post, he is presumably referring to Chat GPT5, which would be released 8 months later. They already knew how to build artificial general intelligence. Chat GBT5 would presumably be a big step towards AGI and according to Altman will materially change the output of companies. In the launch video, Sam Alman said the original Chat GBT3 was like talking to a high school student. GPT4 was like talking to a college student. GBT5 is like talking to a PhD level expert on whatever topic you're asking about. And it is a major step towards AGI. But upon its release, it appeared to be at best a marginal improvement over GPT4. And for many users, their experience actually got worse. For example, this Reddit user who pays $200 per month for ChatGBT Pro complains of it giving him blatantly false information. He shared his chat history. He asked ChatGpt to give the GDP of EU countries. It told him that Poland had a GDP of $2.2 trillion. It claimed to have gotten this data from the IMF website. In reality, Poland's GDP is $900 billion. So, Chad GBT made up a number that was off by a factor of more than two. It will give wrong ingredients for basic recipes. And when you ask it about world leaders and presidents, it gives out outof-date answers. In this user's experience, it provides false answers more than half the time. Chad GPT's propensity to confidently make up fake information has always been its major drawback. And this has not been fixed by GPT5. If anything, it's gotten worse. If you can't even tell me the GDP of Poland, how are you going to create a super intelligent AGI?
It's questionable whether or not AI is yielding the massive productivity gains Alman and other tech bros have promised. For example, in June of 2025, Salesforce CEO Mark Beni off said 30 to 50% of their internal work is now being done by AI. Salesforce has been aggressively implementing AI tools internally since shortly after Chat GBT was launched in 2022. If AI is now doing 30 to 50% of the work, you'd expect them to lay off 30 to 50% of their employees. As of the end of Salesforce's fiscal year on January 31st, 2025, they had 76,000 full-time employees. This is an increase from the 73,000 employees they had at the end of 2024. They only report their employee count once per year in their annual report. In September, Benoff revealed that he had laid off 4,000 customer support workers out of a total of 9,000 because a lot of customer service inquiries are now being handled by AI. So, it is indeed true that almost half of customer support workers have been laid off, but this represents only 5% of the company's total headcount. It seems like AI is pretty much only useful for customer service. If AI was really doing 30 to 50% of all of Salesforce's internal workloads, as Ben off claimed, they should have laid off at least 22,000 employees, not 4,000. Not only that, but it looks like the company's net headcount is actually increasing. On September 12th, 2 weeks after Beni off bragged about laying off 4,000 customer service workers, he gave an interview to CNBC where he said they now have 80,000 employees. So, while they laid off 4,000 people, they hired another 8,000. Why would you need to hire so many people when AI is supposedly doing half the company's total work?
It's unclear if AI is even effective at customer service. If you've ever called a customer service hotline and were answered by an AI voice, you probably didn't have a great experience. For most jobs, AI's propensity to hallucinate makes it unusable. Maybe this problem is unsolvable. For this core problem, maybe there is no solution. Of course, Sam Alman can't admit that. His company is burning so much cash that they need to raise at least $60 billion in their upcoming IPO. They need to distract away from the lack of progress at ChatGBT and the fact that AGI seems to be slipping further and further out of reach. This whole sora thing is a distraction. It got OpenAI back in the headlines for a few weeks thanks to the crazy videos it can generate. But this is completely useless. It just serves as yet another money sync burning through their venture capital funds.
All right, guys. That wraps it up for this video. What do you think about Sora? Let us know in the comments section below. As always, thank you so much for watching and we'll see you in the next one.