Log in
Subscribe
May 1, 2026
SoftBank's $10 Billion Bet on ChatGPT Just Exposed the AI Bubble
SoftBank's $10 Billion Bet on ChatGPT Just Exposed the AI Bubble
00:00
11:43
Transcript
0:00
SoftBank is trying to borrow ten billion dollars against its OpenAI shares. This is kind of like if I come up to you and I ask you, and I say, "Look, will you just spot me fifty grand? I just need fifty grand right now.
0:14
I can pay it back, but me paying you back is contingent on the crackhead paying me back for the money that I lent him." That's exactly the same thing.
0:23
This is a signal that we've entered a very dangerous portion of the AI bubble pop.
0:29
We're gonna go through all of the facts and figures, and then we'll drill down into the deeper implications of what this means for AI financials going forward.
0:37
Before we hop into it, if you're not already subscribed to my mailing list, subscribe now and you'll be eligible for a steep discount on a PDF I'm gonna be releasing in the next couple of weeks that details how to make yourself layoff bulletproof.
0:51
And if you're already out of work, maybe you've been laid off, maybe you've been downsized, it's gonna give you the strategies, techniques, and insider views on how to get yourself back into a tech job quickly.
1:02
Again, that's drjoshccimmons.kit.com. Newsletter comes out every week. That's gonna make you eligible for a coupon discount code for the upcoming Layoff Resiliency Guide. Now back to it.
1:13
So they're seeking a ten billion dollar margin loan that is backed by those OpenAI shares. And ordinarily, you know, having a margin loan backed by some kind of underlying security, this is normal.
1:27
This is the way of doing business. Where it becomes very sketchy in this situation is nobody really knows the value of OpenAI. Now, of course it has a valuation, but the valuation is privately assessed.
1:39
We don't really know exactly what goes on in that process or how reflective that number is of reality. When the company goes public tomorrow, it could crater. It could crater after an IPO. It could shoot to the moon.
1:52
Nobody really knows or has 100% confidence, and if you think you do, look at WeWork, which was actually another one of SoftBank's investments, if that raises any red flags for you.
2:02
They had a sky-high valuation when they were private, and then as soon as they IPO'd, it cratered. The company went kaput.
2:08
SoftBank is borrowing money on a contingency that it can't really back up or even know if it's possible to back up at this point. And by the way, this isn't the first time.
2:18
They already secured a forty billion dollar bridge loan for OpenAI shares. So it seems like they're just borrowing this money, reinvesting it in OpenAI, and then loaning out on top of those new shares.
2:31
It's like an infinite money glitch, but it has to stop at some point. It's leverage on top of leverage on top of leverage. It actually reminds me of an activity I do with my daughter sometimes.
2:41
She's about four years old, and she's got these little magnet tiles. They're plastic tiles with magnets in them, sort of squares, and you can stack them together to build buildings.
2:50
And she likes to build towers out of these. She likes to see how high those towers can get. And right now, SoftBank's tower is, you know, fairly high. It's maybe two to three feet high.
2:59
And they just keep adding these magnet tiles, leverage on top of leverage on top of leverage, and what inevitably happens with a tall magnet tiles tower, it crushes under its own weight. It eventually must fall.
3:11
Or I guess you could sell it to somebody else before it falls, but that's beside the point.
3:15
If you're not super familiar with the finances of private companies, I'll walk you through it a little bit as somebody who's worked at a number of them.
3:21
And spoiler alert, just like anything else finance-wise, they throw a bunch of jargon at you to make you think it's really difficult to understand, but it's all quite straightforward.
3:31
So let me break this down for you in short order. When you're a public company, there's a price for your shares. If I want to go buy shares of Apple today, I need to pay the price that they're going for at the brokerage.
3:43
If you are a private company, there's not a massive marketplace of shares entering and leaving people's hands, so there's no nego- there's no price. There's no essential price to them.
3:53
Instead, if you need to value those shares privately, like you're gonna lend against them like SoftBank is, then you need a private valuation.
4:02
There's a formal process called a 409A to get to that private valuation, but before you think that it's all above board because it has a number, number, number, letter title to it, it's still very push and pull-able.
4:17
Because the people participating in that 409A all have a story about the AI company.
4:22
We all know the story that Sam Altman is telling, is that this is a generational company that will totally change the world and put most of us out of work. It's a bold claim.
4:31
It's especially a bold claim when you realize that it hasn't really put too many people out of work directly yet. The H-1B visa and offshoring has done that.
4:39
So there's a room full of people working on the 409A who want the number to go up. So can this be a little bit biased? Yes, obviously.
4:48
And I'm not saying for every company everywhere, but does this have the potential to be a bit of a dicey number? Yes, 100%.
4:55
But SoftBank is now trying to price that in as if this is a given, this is a fact, the valuation they're sitting at. This is factual.
5:02
When in reality, nobody can make that determination until the stock goes public, when you see what people are actually willing to buy and sell it for. And who is calling the shots on this?
5:12
None other than the CEO of SoftBank, Masayoshi Son himself. And this is a classic Masayoshi Son move. He doesn't make typical bets like most other traders.
5:23
He likes to identify a major shift in the technology that's occurring, find the choke point, and then own that choke point. Go all in on it. And he's done it with Alibaba, with Arm, and it's made him quite a rich man.
5:36
And I wanna be clear here, I am all for outrageous bets.
5:39
It's what makes investing interesting, and especially if you have someone like Masayoshi Son with a gut feeling for this instinct and a track record of making a crap ton of money off of this trading pattern, then I say more power to you, and you should be able to do whatever you want with your money.
5:55
The issue I have is that now he is taking out a loans with OpenAI as an underlying security, and he might not be able to pay back those checks.
6:03
So what was just a risky investment strategy that could potentially make him bankrupt or make him even richer is now something that's contagious and embedded into the financial system based on whoever takes that loan.
6:15
You know the slogan of this channel by this point. AI companies are real, but the people running them are not. AI companies are real, but the people running them are not.And by extension, the finances are not.
6:27
They're not real.
6:28
This has been getting less and less stable over time, but this seems to be a real signal that things are cranking up a notch when we're stacking multiple layers of leverage on top of one another and then loaning that money out to people who might not be into the AI game at all.
6:42
The internet was real, the housing market was real, but bubbles crop up when capital structures start to get wonky, and that's exactly what's happening here. You'd have to be blind to not see it.
6:54
We talked about the circular financing scheme before, how money throws-- flows through NVIDIA and back to the AI companies, but this happens in many ways to prop up this AI bubble, and this is another one in the works that is actually happening now.
7:08
I mean, here's what happens. As SoftBank invests in OpenAI, the valuation of OpenAI goes up. Those shares have some kind of value, even though it's undetermined.
7:18
SoftBank then seeks a loan backed by those shares, then it takes the extra money and invests more into OpenAI and other AI companies.
7:28
So it's sort of like stepping up, stepping up, stepping up until it reaches its breaking point. And when will that be? It's gonna be this summer at the rate we're going. So what exactly could go wrong here?
7:38
What could go off the rails for this deal? Well, a number of things, and we're already starting to see a few of them, which makes the timing of this even more confusing.
7:47
A number of data centers that have been planned by the hyperscalers before twenty thirty are behind schedule or totally canceled at this point.
7:54
We're just getting more reports on that just over the past few weeks, and this is due to a myriad of reasons. You can watch my episode on where all the electricity is gonna come from for one of these choke points.
8:05
It's everything from where are we gonna get the electricity at a good cost? Where are we gonna get the water? Those are obvious, but a lot of less obvious things than you think too.
8:13
The electrical transformers that you require to run these data centers, they take a very long time to build. I was just talking to my buddy who owns, uh, an electrician company. He's an electrician.
8:23
He goes out and does these big commercial jobs. Just had him over a couple of weekends ago, and we were talking.
8:28
I was asking a little bit more about transformers, and these are extremely, extremely complex bits of equipment that take a long time to manufacture, and they're ludicrously expensive.
8:39
You also have the high-tension electricity wire that goes cross-country, the ones you see on those very, very big poles out in the country. That is-- It takes a long time to get the permits to
8:50
actually put in rural areas, and it takes a long time and a lot of materials, and it's very expensive to haul the actual cable that flows across those.
8:58
You don't think about this stuff, but data center construction could be held up in any way. And OpenAI, uh, like I said, we're already seeing those hold-ups. OpenAI is deeply intertwined with that.
9:08
If the data center construction slows down and they can't serve their models, as we're already seeing with so many service disruptions with Anthropic over the past couple of weeks, what does that mean for their share price?
9:17
The, the thing that they sell, they can no longer sell at the same volume that they were selling at.
9:22
There are three or four other ways that this could go bad, but that's the one that has the most definitive proof right now, is that we don't know if the infrastructure is gonna be able to support Sam Altman's bets.
9:32
By all appearances, it looks like it's not going to. So let's zoom out and frame the whole cycle for you of AI products, where we're at now, and then I'll cap it off with an assessment of what to do now.
9:44
So in the early days, there is product hype. ChatGPT, Gemini, Claude, the new models are coming out. They're all significantly better than the last ones.
9:53
Everyone is talking about how this is gonna create a three-day work week. We're all going to be rich and prosperous and live in a utopian society. Everyone is very, very excited.
10:03
Then the first wave of reality sets in, and that's the data center construction. Who is gonna have access to the servers preferentially? Where is the money gonna come from?
10:13
How are you gonna build enough transformers in time? Even the government stepped in and financed some of this with Project Stargate.
10:19
Then, as if this was a mall map with a little arrow saying, "You are here," we are here right now, which is the financial engineering stage.
10:26
And this is the financial shell game that you see at the end of any bubble where you have highly leveraged positions from major banks, which means, as I mentioned, this is contagion.
10:36
This branches out to people with no AI exposure. They're ETFs, your private investors, your regular people with, uh, Vanguard ETF.
10:46
You're getting exposed to this one way or the other, even if you don't directly own tech stocks. So we're all floating in the same ocean. A rising tide lifts all boats, and a drought sinks all boats.
10:55
Phase three is especially concerning because this is where money people start to take over from the engineering people, so they don't actually know how to build the things.
11:02
They just know how to engineer sales and engineer money out of it, which can only go on for so long without a fundamentally functioning product.
11:10
At a first glance, the AI boom is about people, users, products, but what's being exposed is that it's not about that, and maybe it never was about that. It's about who can borrow against the future.
11:21
Right now, SoftBank is proving that they can borrow against the future, but the question that nobody is asking is what happens if that future doesn't arrive?
11:30
And so the AI boom that began with code and chips is now being built with debt, and that only leads to one place: financial ruin. But it looks like the clouds are gathering, and it looks like it's about to rain.
The Dr. J Show
Recent episodes
Speed Is Everything in Tech Right Now
May 4, 2026
The Flock Surveillance Problem Nobody's Talking About
May 2, 2026
