So why are investors so confident about Anthropic?

So y- you mentioned SpaceX, and that being the largest IPO of, of all time.

We're expecting Anthropic to, to surpass that, and its investors think that's gonna happen because it has shown just this prodigious growth over the last year, or really 18 months.

It has, in the course of that, leapfrogged its main rival, OpenAI.

That makes Anthropic the fastest-growing startup of all time.

So investors are looking at those numbers, and they also are looking at the opportunity that is associated with AI.

And Anthropic has really demonstrated this year that it can capitalize on that opportunity perhaps more effectively than some of its rivals, and it has really done that by focusing on selling AI to businesses, which has been a, a growing theme throughout this year.

Uh, tell me about some of what Anthropic is up against.

What happens if SpaceX can't do all this?

There is a very real business at the core of SpaceX.

They have a near monopoly on rocket launching, and Starlink has become a very, very successful part of the overall business as well.

But combined, those contribute the bulk of what was around eighteen, nineteen billion in revenues for the company last year.

On the basis of today's valuation, it's ninety-two times that revenue figure, which is just astronomically far above big tech peers and others in the public markets.

And if you're just looking at fundamentals there, analysts think that SpaceX is probably closer to a seven hundred and eighty billion dollar company.

That's a number from Morningstar.

Are investors just getting worried about all these stocks that are about to hit the market?

What, what are some of those things? What are those things, I suppose, that they will hope to do before this process goes ahead?

Well, may-- I mean, maybe just to step back a little bit on the kinda company that OpenAI is.

It's, it's unprecedented in all sorts of ways, not just its scale of it as a private company, but also, I mean, this is-- it was founded as an AI lab in twenty fifteen, and the idea is this very highfalutin mission to create, uh, artificial general intelligence for the benefit of all humanity.So it was never really established as a kind of, you know, this is gonna be a, a company with regular quarterly earnings reports and a, and a reasonable normal balance sheet that investors could read.

It's always been a fairly out there moonshot company, and it, and it continues to be that even though it has become more conventional in terms of its governance and its processes.

So I think the, the key thing that is helpful for them to do on the private markets is continue to lose a ton of money.

They just need the financial firepower for the things like the data center build out that Sunil just mentioned.

Yeah.

Yeah, I think there's maybe three things going on, and one of them I, I touched on a moment ago, which is the, you know, the, the fact they have all raised such huge sums from the investors who they have access to as private companies.

Well, what's the case for OpenAI's high valuation?

So as long as I've covered this company, which is about three and a half years, the story of OpenAI has been kind of, you know, turmoil and crisis and lurching from one disaster to the next, and all the time they have got bigger and better, their models have improved, their revenues have, have stepped up dramatically.

So I think the sympathetic read is that it is a startup that's 11 years old.

It's moving incredibly fast, and what that looks like sometimes is picking up projects and putting them down when they don't quite work and taking a new tack.

And I think you could regard what they're doing here in putting more emphasis on the enterprise business as a response to very obvious traction that they and their rivals are picking up, a very obvious effort to go and make more money by doing this, and in the fullness of time, investors might come to celebrate that decision.

What kind of impact do you think that could have on the wider AI industry?

So OpenAI are gearing up for a public listing.

So are Anthropic, and in a separate field, so is Elon Musk's SpaceX, and these will be, we anticipate, the three biggest IPOs of all time.

So let's back up first and talk about what happened in this dispute leading up to the lawsuit.

So Anthropic and the Pentagon have been negotiating terms for the last few months.

It's currently the only AI company whose model's used in the most sensitive operations, so classified operations.

And the center of this dispute is what Anthropic's AI models can be used for in military operations.

The Pentagon and Defense Secretary Pete Hegseth have pushed for the use of these models for all lawful purposes, which is a, a fairly wide definition.

And Anthropic have said, "We're comfortable with that in general, but in particular, we need assurances around two particular uses of AI, one of those being domestic mass surveillance, and the other being the use of AI models for lethal autonomous weapons." And the, the negotiations have really run aground on those two issues.

Now, the Pentagon has said it won't comment on the lawsuit, but George, what broader implications could it have on how AI is used in the future?

So it's worth touching on Anthropic's particular complaints or issues with how its technology might be used here.

Okay, George, just to start, can you help us understand OpenAI's somewhat complicated ownership structure?

I can, I can try.

I think first it's worth just saying what OpenAI are trying to do here.

So, they wanna become the AI superpower company, and that means spreading out from the 800 million or so users they have today to, to billions and, and basically becoming for AI what Google has been for search.

That requires massive amounts of computing power, which costs a huge amount of money, and in order to raise that cash, OpenAI has persuaded its investors of a couple of things.

One is that it will be worth trillions of dollars one day, but the second is that investors will get a piece of the action by owning equity.

There's a major wrinkle to that, which is that to date, investors haven't been able to own equity, and that's because OpenAI was founded as a nonprofit and it has a very convoluted corporate structure.

What are some of the hurdles to that? Can you break it down?

Why exactly are venture capitalists so eager to borrow this roll-up idea from private equity?

So let's maybe talk about what's made this effective for private equity firms and then we can come onto what VCs are doing differently in this space.

But for PE firms, one of the main benefits has been by consolidating these businesses, you get a whole lot of operational efficiencies, you can combine the, the backend of all of these businesses and, and save money that way.

You also get, when you sell the consolidated business on, you get a much better multiple on, on earnings typically if it's a larger business.

So that's how it's worked for PE firms.

For venture firms, some of the same logic applies, but they are taking a very different approach.

So rather than buying up these businesses, cutting costs and layering on debt, which is typically the private equity playbook, VCs will be buying up the business using money to go and acquire rivals, but then embedding technology into the underlying business.

But why have VCs only started to do this recently?

Get me up to speed on what's happened recently because there's been some big news there.

there.Yeah, so last week was a big one in terms of, of this restructuring plan, which has really been rumbling on for months and months.

The critical aspect of OpenAI's corporate governance is that the non-profit controls the rest of the company.

And OpenAI had been looking at undoing that structure and effectively turning the whole company into something called a public benefit corporation.

They rowed back on that plan last week.

Instead, what they are now planning to do is turn the for-profit subsidiary of the company, which will still remain under the non-profit, uh, board, they're planning to turn that into a public benefit corporation.

So they're hoping this kind of threads the needle of, on the one hand, pleasing investors who will now be able to have an equity stake in this subsidiary, and on the other, pleasing critics who've said giving up non-profit control will mean they're abandoning their mission.

What else could get in the way of OpenAI's restructuring?

All right, so can you break down for us what OpenAI's structure is, and what exactly Musk and this group of investors offered?

So OpenAI has a really complex structure, and that's a legacy of its foundation as a nonprofit research organization in 2015.

The company then launched a capped, for-profit subsidiary in 2019 so it could start taking in outside investment.

So it has this slightly Frankenstein structure of a nonprofit, which has a controlling stake in a for-profit entity.

And what OpenAI is trying to do is to streamline its structure and make itself into a more traditional, for-profit structure.

So Musk is offering $97.4 billion with a consortium of other investors to take control of those assets held by the nonprofit, which include control over the for-profit.

How much of a wrench could this throw into Altman's plan to convert OpenAI to a for-profit company? I have to imagine that there would be several hurdles there.

So Altman is trying to do something phenomenally complicated in turning this company, which was founded as a nonprofit, into a form of for-profit.