Microsoft needs to open up more about its OpenAI dealings
wsj.com147 points by zerosizedweasle 3 hours ago
147 points by zerosizedweasle 3 hours ago
Companies have a lot of tools at their disposal to hide things on their balance sheet for a while. However when that happens it typically means the numbers are bad. Really bad. If they weren’t, they’d do everything they can to highlight how great the investment is going.
Same reason why seemingly every CEO on the planet is making hand wavy statements about how their company is leading with AI and it will revolutionize their industry, and yet almost nobody is willing to break out this amazing stuff in their P&L. Funny how that works.
Even if this isn't Enrony, this sounds so Enrony (if you know anything about the Enron accounting scandal)
"How Microsoft has managed to avoid disclosing such basic details is baffling. The company in its financial reports identifies OpenAl as an equity-method investment. That means OpenAl, by definition, is a related party of Microsoft under the accounting rules. Microsoft, however, doesn't identify OpenAl in its financial reports as a related party, and doesn't say anything about its transactions with OpenAl in its related-party disclosures."
If its equity accounted it won't be considered a related party as far as I understand. Related party in IFRS isnt what you think it is. Its the equivalent of "extended family".
https://viewpoint.pwc.com/dt/us/en/pwc/accounting_guides/fin...
26.5 Common related party transactions
In order to comply with the related party disclosure requirements, a reporting entity must identify all of its transactions with related parties.[…]
26.5.1 Disclosure of related party equity method investments
Equity method investees are, by definition, related parties of the equity holder.
AFAIK Microsoft didn't put any kind of liquid money into OpenAI, it's something like "you can use up to nB USD of our resources for free", not sure how that should go into accounting. It could even be a liability without much juggling.
Yeah, Enron didn't really put any money into its related parties, it just used to them to move things around. I don't think it is Enron, but related parties shenanigans give me the chills.
I think you’re calling out two different phenomena: 1) the gold rush mentality leads to bad investments (at least in the short term), and 2) in a hype bubble companies are incentivized to attach everything to the hype, even if it’s not real (many companies talk AI but aren’t seriously investing).
Both are true in many cases. But to the extent companies are making major investments that are strategically correct but won’t make money for years, it’s still the right move to hide stuff in financial statements.
Markets don’t reward long term investments. Everything has to be short term, and if it’s not paying off instantly, short term investors get no value and want it stopped.
Net result: lots of PR about AI, but almost every company is incentivized to downplay it financially.
IF said investments are strategically correct (considering their amounts). A few years bck companies were making strategic investments in the VR/AR/XR goggles, on a bet that tech would become cheaper in a few years while quality would improve dramatically. And they were correct in the price and quality aspects. But they were fundamentally wrong strategically, regarding millions of people wanting a display semi-permanently strapped in their visual area. Apparently not many people want it and even less need it, which was drastically different from the mode of work of CEOs, which is to always move and issue commands while walking/driving etc. Same story with multiple failed voice assistants.
You’re not wrong, but the point of the article is that for a publicly traded company there’s an expectation of more transparency. The size of the losses is getting to a point where it can’t just be kept hidden inside a handwavy “other” line item.
> Markets don’t reward long term investments. Everything has to be short term, and if it’s not paying off instantly, short term investors get no value and want it stopped.
The AI investment bubble is almost entirely about making long-term, extremely expensive investments. That's what the gigantic datacenter build-out is about, not short-term investments and short-term returns. They're telling everybody, persistently, that they're making huge long-term bets, and the market is rewarding them like crazy. See: Oracle's run due to long-term bets on AI (it's certainly not short-term results causing the stock to do that, their short-term growth has been mediocre).
Amazon for two decades repeatedly told investors they were making extremely expensive, long-term investments in build-out (eg their fulfillment build-out era), where the primary payoff would be far into the future. The market bought into the long-term on the basis that it was attached to Bezos at the center (that he'd be there to deliver that long-term result). The same is true about Elon Musk with Tesla: they have endlessly made outlandish long-term proclamation to drive their stock. Tesla: robot super business, self-driving taxi business, et al - these are 10-20-30 year long-term claims by Tesla and the market has aggressively rewarded it. That's because they think Musk will/might be there to guide it to actuality. In most cases investors don't buy in because they know the CEO & team won't be around even seven years from now.
Markets (investors) reward long-term if they can be made to believe in the long-term. The issue is that most companies are not believable on long-term statements, they don't have a leadership that will be around for any long-term delivery. Buffett, Bezos, Musk, Zuckerberg were/are long-term attachments so the market has been willing to buy in on various long-term bets.
Agree on the long term concept, but there’s little comparison between Amazon’s early years and now. Amazon spent money on tangible capital infrastructure that was highly differentiated and long lasting (fulfillment centers, logistics networks). That costs a ton up front but can be used for decades. Folks struggle to compete with Amazon now because that infrastructure is a giant physical logistics moat that’s not easily replicated.
The AI bubble is far from that. Companies are spending tons on GPUs that have limited lifespan, building models that have limited lifespan, using algorithms that are all basically the same, in a space where someone can dump a “good enough” open source model on the market and blow up your business overnight. There’s very little lasting value in what’s being built and the “we’re investing for the long term” arguments don’t hold much water. It’s like saying you’re investing in real estate but then you keep tearing down the building and rebuilding it every 18 months. That just doesn’t work.
There might be some longer term fungible value from some of the baseline infrastructure investments (data centers, electrical upgrades) but those are undifferentiated and highly fungible.
> Companies have a lot of tools at their disposal to hide things on their balance sheet for a while.
That's why some analysts ignore most company-provided metrics and just focus on cash-flow. You need inside and outside of the house fudgers to mess with that metric.
Tim Berners Lee: during .com bubble, companies were high on marketing and low on profits.
Microsoft won't even publicly say how much they paid to help demolish the East Wing of the White House.
https://www.cnbc.com/2025/10/23/trump-white-house-east-wing-...
This is a silly article. Since MSFT took a ~49% stake in OpenAI, it records its share of OpenAI's net losses in the other income line under the equity method of accounting. MSFT is offsetting its taxable income based on a prior investment
The complaint is not that MSFT is reporting the loss. It's that, in terms of both the % of OpenAI and the dollar value of the stake, it's large enough that there should also be a related party disclosure.
IANAA so I don't know how true that is. Just wanting to point out that I don't think you're responding to the key point of the article.
Can you elaborate on this a bit more? Does that mean OpenAI had a ~$9.4 billion loss so MSFT needs to put 49% of that loss in their books?
Does Microsoft even have an OpenAI stake? Their original more public deal was revenue sharing up until they reached 100x $1 billion. That doesn't sound like a stake.
They also had tech sharing valid until the OpenAI board declared 'AGI'.
That seems like a really bad deal. And that was probably at the time when Microsoft had the most leverage to make a deal. Their subsequent deals would make sense to be worse.
How is it already losing that much money on its OpenAI investment?
OpenAI is a textbook example of having fun by burning cash. Nobody doubts it’s “cool.” Lots of people questioning if there’s an actual business there.
I think it makes Microsoft feel bigger though.
If Microsoft were just Windows, Teams, Azure, Bing and whatever it is, Microsoft would actually feel like a competitor for firms like Canonical or Red Hat or SUSE which happens to be big but nothing special relative to the others, whereas it now, with with this very public service feels like a behemoth.
I think you might be right but the other side of it is that Microsofts business outside of AI is mature other than their lock on the windows desktop you listed the others :-
Windows competitors are OSX (and the very good Apple hardware), Linux (which thanks to Valve is gaining users at an increased rate).
Teams competes against Slack/Discord
Azure competes against AWS/GCP.
Bing "competes" against Google Search
While they do have a share of each (and a big share of the desktop) they don't really have anywhere they can grow, they've filled their existing niches and are competing with other equally sized companies in all of them.
So spaffing some cash on AI on the off chance it pays off down the line might look smart.
Hell if AI does pay off then they look good and if it doesn't, it'll look bad for everyone who invested and they can at least shrug off the cash hit.
Huh, that’s a funky statement because having “Azure”, one of the largest Cloud Providers outside of China is definitely a different camp than Canonical, Red Hat and SUSE.
Although I don’t particularly like their cloud services they are undeniably an important part of Microsoft’s business. (And they also own a large chunk of the gaming industry nowadays).
Yes, of course those things are huge, but I was focused on perceptions.
Suddenly Microsoft has gone from being some software to being everywhere. I know that Azure is huge, but you don't see Azure.
>(And they also own a large chunk of the gaming industry nowadays).
They’re shuttering half their studios, cancelling half their games, and firing game devs by the thousands as they hand halo over to PlayStation lol. You’re technically right but they clearly aren’t taking that part of their business seriously anymore. IIRC Gamepass has plateaued on subscribers for years now even prior to their very aggressive price hikes over the last 18 months.
I saw an article the other day that said Microsoft is telling developers they have to have a 30% return on their games, which is almost double the industry standard. That’s just absurd.
Edit: worth mentioning that you have people openly speculating at this point that they might not even make another Xbox. I’m not quite in that camp, but I also think it is a distinct possibility given the back slide they are clearly in right own when it comes to gaming. Fun fact: It’s been 4 console cycles, almost 25 years, since we saw a major player drop out.
> firing game devs by the thousands as they hand halo over to PlayStation
They kind of took it from Apple to begin with. We almost had Halo on the iMac before Microsoft acquired Bungie [1].
Yeah, you're right on all those fronts.. but they do own some large IP like Call of Duty which will continue to generate money for them for the foreseeable future.
I did hear the speculation about Xbox as well and I hope it's not true. I quite like the Xbox as a console (Series X was the first one I bought, used to be on PlayStation before that). Competition is good for the console space, and Nintendo and PlayStation aren't really competitors IMO. The audience for Switch and PS/Xbox isn't the same.
You may not be following OA for a while! They’ve never turned a profit. So, someone on the other side has to lose for them to stay afloat!
> They’ve never turned a profit
Now that OpenAI is starting to talk about ads and allowing "erotic" content, I feel more comfortable in my prediction that not only have OpenAI never turned a profit, they never will. They will be consumed by Microsoft or crash the market so hard it's not even funny. The technology will survive, and it will be useful, but OpenAI as a company is done.
It’s amusing how “ads” is seen as an obvious way to make profit for OAI as if Google’s (especially) and Meta’s ads businesses aren’t some of the most sophisticated machines on the planet.
Three generations of Twitter leadership couldn’t make ads on that platform profitable and that exposes far more useful user specific information than ChatGPT.
The hubris is incredible.
There's an absolutely massive disconnect between the technology Sam Altman is presenting in interviews and what is available. Like they're going to create an AI that will design fusion power plants, but right now they can't turn a profit on a technology that millions of people actually use in their day to day work? Can you sell enough ads to carry you through to the fusion capable AI?
More and more OpenAI is drawing parallels to the Danish scandal of IT Factory. Self-proclaimed world leading innovation and technology in the front, financial sorcery in the back.
If they really believe their AI is going to be so great, I guess they can just ask it for a business model when it gets there. So their lack of business model is at least self-consistent.
Right now they may be a bit scarce on business plans and revenue, but I hear they're ushering in an era of "post-scarcity" so that should fix it.
That is more or less their actual plan. They ignore or want us to ignore that the technology is commoditising so fast that even if it is great, they won't have enough of an advantage for this to provide an edge for more than a matter of months. Just as Microsoft and anyone betting on AI data centre rollouts want us to ignore that the equipment they are rolling out will be functionally inadequate to support new models in far less time than they can make money to offset the cost; the only part of this capital expenditure that will provide lasting value is the building/power/cooling infrastructure, and probably not all of that.
It's a giant money pit, funding a bunch of people who are not long off the crypto grift train if they are at all.
"Could it be that, once again", Sam Altman is really not that far removed from a grifter?
https://www.bloomberg.com/opinion/newsletters/2025-10-15/ope...
> There’s a famous Sam Altman interview from 2019 in which he explained OpenAI’s revenue model [1] :
>> The honest answer is we have no idea. We have never made any revenue. We have no current plans to make revenue. We have no idea how we may one day generate revenue. We have made a soft promise to investors that once we’ve built this sort of generally intelligent system, basically, we will ask it to figure out a way to generate an investment return for you. [audience laughter] It sounds like an episode of Silicon Valley, it really does, I get it. You can laugh, it’s all right. But it is what I actually believe is going to happen.
> It really is the greatest business plan in the history of capitalism: “We will create God and then ask it for money.” Perfect in its simplicity. As a connoisseur of financial shenanigans, I of course have my own hopes for what the artificial superintelligence will come up with. “I know what every stock price will be tomorrow, so let’s get to day-trading,” would be a good one. “I can tell people what stocks to buy, so let’s get to pump-and-dumping.” “I can destroy any company, so let’s get to short selling.” “I know what every corporate executive is thinking about, so let’s get to insider trading.” That sort of thing. As a matter of science fiction it seems pretty trivial for an omniscient superintelligence to find cool ways make money. “Charge retail customers $20 per month to access the superintelligence,” what, no, obviously that’s not the answer.
> On a pure science-fiction suspension-of-disbelief basis, this business plan is perfect and should not need any updating until they finish building the superintelligent AI. Paying one billion dollars for a 0.2% stake in whatever God comes up with is a good trade. But in the six years since announcing this perfect business plan, Sam Altman has learned [2] that it will cost at least a few trillion dollars to build the super-AI, and it turns out that the supply of science-fiction-suspension-of-disbelief capital is really quite large but not trillions of dollars.
> [1] At about 31:49 in the video. A bit later he approvingly cites the South Park “underpants gnome” meme.
> [2] Perhaps a better word is “decided.” I wrote the other day about Altman’s above-consensus capital spending plans: “'The deals have surprised some competitors who have far more modest projections of their computing costs,’ because he is better at this than they are. If you go around saying ‘I am going to build transformative AI efficiently,’ how transformative can it be? If you go around saying ‘I am going to need 1,000 new nuclear plants to build my product,’ everyone knows that it will be a big deal.”
The switch from "Ai might kill us" to "you'll goon to Ai" was kinda funny, not gonna lie
Just based on the number of ads I get for thinly veiled erotic chatbots, and the success sites like character.ai have with pretty bad LLMs, there has to be a lot of money in erotic LLM content. OpenAI turning to that market is a sign they are running out of easy investor money, but if they can survive the associated controversy without lobotomizing the models this sounds like a method to turn the entire company profitable over night. They might have to raise prices or abandon the flatrate model to deal with heavy users, but locking adult content behind separate plans might even increase acceptance
Not sure if increased availability of LLM porn or the gradual erosion of LLMs with ads and sponsored content would be the greater evil on a societal level. Neither is particularly great. But they will certainly drive shareholder value
This. If you really think you're a couple of years away from building Digital God, and today have virtually unlimited access to capital, you are not going to spend time shipping a sexy mode.
Why not do both if you have unlimited capital today?
Do they also have infinite labor?
Isn’t the whole point of AI to replace labor? If they really want to put their money where their mouth is, they’d be casting off staff at a prodigious rate.
Do you predict the same for Anthropic? Hopefully they will stick around.
If we go purely by economics, then Anthropic belongs to the same category of LLM corporations - ones which have only LLM as a product. As opposed to the likes of Google, Microsoft, even Facebook. Sure, these LLM-first corporations have a very tiny lead in both technology and (LLM) brand recognition, but it is shrinking fast. I suspect that only companies which will bundle LLMs with other big products (and do it cheaply) will survive in the long run.
The problem, I think, is that IF OpenAI fail, they'll take with them a lot of other AI companies, simply because funding will be redirected away from the field entirely. If you're profitable, then you're probably going to be fine. If anything your operating costs will go down as there is less competition for staff and compute.
Porn is enormously profitable. This might just be the saving grace for AI. Historically, porn has been a pioneer in new tech industries (home video, online commerce, video and streaming). This time they aren't first to the game but don't underestimate the industry.
It's not as profitable as the original promise of AI to cut labour costs on a massive scale. Stockholders won't be happy.
The saving grace here might be that you can hide your porn subscription in your monthly OpenAI subscription. Only question then: Will VISA and Mastercard cut off OpenAI for peddling porn?
Assuming it deals somehow with the folks who use it for revenge porn, deep fakes of non-consenting parties, and depictions of minors…has it said anything about doing that? Has anyone? Does anyone know the legal implications of generating mass quantities of sexual exploitation, or is this just another thing with AI that society will have to simply tolerate in the name of “progress”?
They're in a tough place with respect to pricing. Qwen3 and DeepSeek's latest local models are too good and are practically free -- so if they try and jack up pricing to a level that ensures profitability, it won't work, as they're simply going to lose too many customers.
There's a mechanism here similar to a Laffer Curve: Charge too much, they lose; charge too little, they lose. OAI needs to strike a delicate balance vs. surging low-cost competition.
To be fair, MSFT is likely making a ton of money (or more likely, preventing churn) with their GPT-powered products for the enterprise.
So the math is probably harder than it seems.
Nobody really knows, they hide Copilot inside each of the business units and then claim it's too hard to split out.
Surely nothing bad could happen by basing our entire economy and stock market health on these companies!
The other commenters in here oblivious to the history of companies like Amazon and Google.
Profit is what you have when you have no confidence in how to reinvest what you earn already.
Because OpenAI loses far, far more. It may be one of the least profitable companies in history.
They're doing hundreds of billions of revenue a year, a one-off 4.7B to OAI honestly sounds like nothing on that balance sheet.
Reuters says $12B: https://www.reuters.com/business/openai-hits-12-billion-annu...
They can open up all day long, I still don't want them forcing "AI" down my throat.
The real story will be when companies report valuation losses from their investments in AI companies after the bubble bursts, or even deflates a bit.
Expect lots of hand wavy “non-GAAP” numbers pushed by leadership trying to gloss over their failed AI investments.
That’s earnings call speak for “If you ignore the pile of your money we lost with bad AI investment decisions, we’ve had a good quarter. Moving on…”
Most of these data centers are built with special purpose vehicles to make the balance sheet look better. Imagine the gnashing of teeth when those get written down
Yes… a lot of this is pretty hidden on the balance sheets but eventually GAAP catches up one way or another.
Microsoft and open... I know kids these days don't bother learning the story of things but... that's a hard sell.
But hey they are betting on optimism!
Oh, so they did lay people off because of AI.
Maybe I just don’t know this writer very well but this is a surprising take to see on the WSJ
"How is it a write-off?"
"They just write it off."
"Write it off what?"
"Jerry, all these big companies, they write off everything."
This is exactly why I am telling people the AI bubble is different. It is not poppable, even if it's size is bigger than other bubbles.
Megacorps control so much of them that the financial side has so little volatility.
No seriously this time line really will go up forever guys.
I don’t know about forever, but there’s a reason all these companies gave so much money to Trump. I think it can keep going up a while.
I figure it's at least a couple of years until the Great Disappointment sets in. When you've been investing half a trillion a year with no realistic hope of a descent ROI, you'll do anything you can to not admit your error.
Would you say we are in a New Paradigm?
https://transportgeography.org/contents/chapter3/transportat...
Hahahaha, that's the spirit. We're invincible. Unsinkable. Nothing could ever stop the music : )
Fair point that there’s less direct retail investor exposure than there was with the dot-com implosion, but I don’t think many folks believe this bubble is non-poppable. Even megacorps can’t just keep inflating it forever if no viable business model for AI materializes
Knowing Microsoft I assume this is due to a sharepoint bug.