Heat. Heat. Heat. Heat.
Welcome everyone to The Information. My name is Akash Basicha. It is Tuesday, October 27th. Before we get into today's stories, I've got a quick announcement for you. We want to know how you are thinking about the news cycle right now. And that's why we have just launched our brand new app, The Information Polls. Every day, The Information Polls gives you five quick questions on the biggest stories in tech. You can cast your vote and then immediately see how people across the industry answered. At the end of the day, you'll get a summary of the results and key insights into what the industry really thinks about these stories. It only takes a couple of minutes, so make it part of your morning routine. Download The Information Polls and vote today.
Today on the show, we are going to dig into SpaceX and Broadcom's big multi-billion dollar chip financing deals. We'll also unpack our reporting on DoorDash's talks to acquire Instacart for $17 billion. We'll then talk to the CEO of Anaplan about his enterprise AI playbook. We'll also look at Microsoft's voluntary buyouts and we're going to close out the show with our partner General Compute to talk about their $100 million Series A funding round. It's going to be a great show. So, let's get right on into it.
Broadcom and SpaceX have announced two major financing deals this week. Broadcom has lined up $63 billion from major banks and Blackstone. SpaceX is leasing $40 billion in chips from Apollo, or at least a corporate entity affiliated with Apollo. I want to bring on Gil Luria, head of technology research at D.A. Davidson, and Dan Campbell, our finance reporter, for their analysis. Gentlemen, welcome to the show. It's great to have you here. Gil, I want to start with you. So, these deals feel like they have similar flavors. The way that I understand it, and you're going to correct me here if I get this wrong at all, we've got SPVs, which are separate from the company, separate from Broadcom, separate from SpaceX. The SPVs are the ones that are raising the debt. And then we've got the AI companies, SpaceX, and in Broadcom's case, Anthropic, that are leasing chips from the SPVs that the SPVs have bought. Do I have this correct, Gil?
I don't like it any more than you do. Right. Remember where we learned that term SPV? We learned it from Enron. We don't like that at all. Now, the big difference is Enron hid these, right? That was illegal. SPVs are legal. They're just a way to take debt off your own balance sheet and put it elsewhere. It doesn't mean that there's not a line in your balance sheet for operating leases and financial commitments, etc. It is on the balance sheet, but it doesn't go on the debt line, which has implications both in terms of financial analysis as well as underwriting. This is, as long as it's allowed, as long as investors put up with it, it's going to continue. We don't have to like it, but it's going to continue. There's a lot of capital, debt capital, looking to invest in AI. That's a mandate many fixed income managers have, is to invest in AI, get high yields in what is a promising area, and they would like to partner, they would like to lend that money out to the biggest, most creditworthy customers, and that's Microsoft, Amazon, Google, SpaceX, Anthropic, Nvidia. And if they can't lend to them directly, at least they're happy to take the guarantee from those companies through a special purpose vehicle, a third-party entity that would operate these data centers. So again, don't have to like it, but that's why they're doing it.
Okay. But Dan, I mean, look, Elon's putting out these tweets. I mean, you know, debt markets are closed off. I forget his exact wording, but we've certainly written about how debt is becoming more expensive. So help me reconcile this with the mandate that Gil is talking about.
Yeah. I think what we've seen here and this week is the fact that the largest companies in the world and the most creditworthy companies in the world can still get deals done. You know, Broadcom, Nvidia is sort of a partner to the SpaceX deal in terms of selling their chips. You know, these are some of the strongest balance sheets in the world, and so they are getting deals done in these credit markets. It's worth noting, you know, the Broadcom debt is at a higher interest rate than maybe, you know, Anthropic might be wanting to pay on the lease. So they are having to sort of pay up a little bit to get deals done, but the deals are still getting done. Contrast that to smaller neoclouds that Elon Musk was talking about, many of those deals are not getting done or are taking different forms. So it's a sort of a story of haves and have-nots, maybe.
So, Gil, I mean, the whole reason for using this structure is to sort of separate the debt from the company's own balance sheets. And like you just said, I mean, it does show up in some way, but I guess it's meant to reduce risk for the companies that are leasing these chips. My question for you is, there is still some risk involved here, right? Isn't there reason to be systemically concerned about these deals?
Yeah. We shouldn't finance speculative investments with debt, right? That's how we get ourselves in trouble, and that's what is happening now. And again, we have cash on hand for the big hyperscalers in Meta and SpaceX. We have very strong cash flow, at least for the first four, that it makes sense to make speculative investments in, which data centers are. They're a speculative investment. The rate of growth of AI, we can be optimistic about, but it's highly unknown. Using debt to finance speculative investments, never a good idea. That's how you get in trouble. That's why everybody's warning about a bubble. Now, to be fair, most of the data centers to date have been financed with cash on hand, with equity, etc. But now we are starting to load more and more debt onto this, and per our discussion, that may be running out. If Broadcom has to pay up and the neoclouds are mostly out of the game, we may not be able to continue on this trajectory in terms of finding more and more debt to finance this buildout. And I'd argue that's very healthy. We should build data centers with the cash we have. That's the right way to do it, and see how it goes. And if we have slightly less data centers than we need, that's okay. The data centers will be worth more, the chips will be worth more. We'll still be able to grow the use of AI. But we are at a moment where we're going to have to see if capital markets still continue to allow this, or they may seize up completely.
Well, Dan, one question I had for you. You and I were talking this morning in the newsroom about this. I mean, in the case of the Broadcom deal, you've got these banks that are going to need to syndicate the debt out. And so, I guess my question for you is, is there another sort of layer of demand that we have to see play out? Are these guaranteed to find buyers as the banks seek to unload this debt, or is that still a variable here?
No, that's definitely still a variable here. Citi and Bank of America were the two big syndicating banks in the Broadcom deal. They have some concerns, we believe, and some reporting suggests this, that they are going to be able to syndicate the debt in a way and get it off their balance sheets in a way that they are happy with. So, we really need to see a long tail of banks wanting to buy into these deals still. And the announcements this week, you know, they contemplate that. They wouldn't make the announcement if they didn't think they could syndicate the deal as it's currently structured, but that hasn't happened yet. And so, it still remains to be seen if they can, and if there are other buyers down the line who want this debt at these levels.
So, Gil, despite all this, do you think we're going to see more or less of these types of financing deals?
I think it looks like we're going to still see the largest companies able to raise that capital, and they have the tolerance to pay slightly higher rates than they're used to. If they're used to paying 5%, they may be able to justify paying 6% or 7% if they believe that the return they're getting on these data centers is in the double digits, which they do. And frankly, look at their financials, they look like they get pretty decent returns. I know that's been a source of controversy, but if you look at Microsoft, Amazon, and Google, they have accelerating growth at stable to increasing margins. That means they're getting a very good return on these investments. So, they're likely to continue as much as the capital markets allow them, but they will have some price tolerance. At some point, they'll slow down the buildout if the rates get too high.
And Dan, help me square this with the big $500 billion commitment that Nvidia got from all these Wall Street giants. This was the big roundtable that Jensen did with all these private equity firms, private credit lenders, etc. on CNBC. Is that the $63 billion here? Is it part of that? Is it affiliated? Are these separate pools of capital?
It's largely separate, but I think it will take much the same form as what we're seeing here with an off-balance sheet SPV doing a lot of the activity. You know, when that announcement was made this summer, I talked to a lot of people in and around the rooms, and what they told me was there is still a lot of work to be done on understanding and structuring the deals. It was basically a press release without a whole lot of substance behind it. I talked to some people this week, and there is some hope that the first deals will be done before the end of the year on that Nvidia partnership. I can't say what firms think they're going to be the first movers on this, but there is some idea or some desire certainly to get the first deals on that consortium out this year. One thing I just wanted to mention, you know, we've talked about how these are the biggest firms in the world. There is some element of this of them wanting to get access to capital before it all dries up. To Gil's point, we may see a point where even the biggest and the safest and the most financially secure firms can't get capital. And so it feels like, and this is, you know, my interpretation, it could very well be wrong, but it feels like there is a rush to secure capital now in, you know, mid to late October, thinking that, you know, come November or December or something, there may be a lot less appetite from investors to buy into these kind of deals. So a lot remains to be seen, but that's an element of this as well.
Right. And Gil, I do want to ask you, I mean, this all bodes pretty well for Broadcom. I mean, you upgraded your rating on the company, and as long as they can get the financing, this is all great news for the chip companies.
That's right. For Broadcom to live up to its expectations over the next couple of years, they need Anthropic and OpenAI to be healthy. And for those companies to be healthy, they need to be able to get the compute they want and ramp up demand and ramp up the ability to serve demand. And the S-1 reads really well for Anthropic. I think there's been a lot of scare about Anthropic is never going to be profitable. It's inherently unprofitable. Destroys value with every AI query. This one tells us a very different story. It tells us about a company that 85% gross margins, 45% after you include all the compute including training, which is really R&D, and was GAAP profitable in the second quarter, cash flow positive for the year. That's a company that can raise capital. That's a company that could get people excited for an IPO next month if they can raise capital. By the way, you still have $134 billion of the money they raised. They've only burnt through $7 billion to get to this sale. They're exiting this quarter at more than $70 billion.
But they got all these, what, $700 billion and $800 billion in commitments coming. I mean, the money's not going to last forever here.
Well, it depends on them getting the revenue. It looks like they're on track to getting a lot more revenue. So again, they've been able to ramp revenue faster than compute. If that continues to be the case, they may be unintentionally profitable, unintentionally free cash flow positive, in which case the cash could last a very, very long time. And all it needs to really right now is last for a couple of years for them to be able to buy those chips from Anthropic. And if that happens, Anthropic's valuation will be far higher than it — I'm sorry, Broadcom's chips. If that happens, Broadcom will be worth far more than it is right now. Broadcom is trading at 16 times. That's basically the market saying, you know what, Anthropic and OpenAI will never have capital to buy chips again. And that appears not to be the case. It looks like Anthropic will be able to raise capital, which means OpenAI will be able to raise capital. They're both sitting on more than a hundred billion dollars. So, they're going to be able to buy a lot of chips from Broadcom, which means Broadcom should be worth a lot more than 16 times.
Great. Well, I want to thank you both for coming on. That is Gil Luria from D.A. Davidson and Dan Campbell, our finance reporter here at The Information.
DoorDash is in advanced talks to buy Instacart for around $17 billion, according to new reporting from The Information. Our deals reporter, Valita Pow, was first to break that price and first to report on Friday that Instacart had been in talks to sell itself. Valita joins us now with the details. Valita, welcome back to the show. It is great to have you here. So, we reported back in 2021, loyal subscribers of The Information will remember, we reported that Instacart was in talks with DoorDash. The valuation at that time would have been somewhere around 40 to 50 billion. We're now at 17 billion, which I think is certainly a sign of the times. What happened here to Instacart's business?
Well, Instacart's business, like back in the days in 2021, it was like the height of the pandemic. So it has been growing very fast. Like, you know, average Americans, you know, who cannot go out, they need to get groceries from Instacart. And, you know, its valuation soared to like 39 billion in 2021. And at that time, The Information reported that, you know, DoorDash is in talks to buy it for like 40 to 50 billion, which would rank as, you know, one of the largest internet deals at that time. But, you know, the talks fell apart because of antitrust concerns, because, you know, there's some great overlaps in, like, you know, the businesses, and, like, you know, they don't think it will kind of get through the regulators and will be like a costly acquisition to take on and swallow. But ever since then, Instacart has been trying to focus on going public, and, you know, they didn't get the valuation they wanted when they went public, and they only went public in 2023 around, like, you know, 10 billion, and it's been growing slowly, the market cap to like, you know, 12 billion before we reported on the talks, you know, back last Friday. So, like, you know, now that Instacart is, like, you feel like there's a chance for them to, like, you know, get back into the game again to try a deal with DoorDash.
Are there not antitrust concerns that could come up here this time around?
Of course, there will be antitrust concerns, but I feel like, you know, companies in this regime, like with the Trump administration, like people can see like the willingness to kind of make compromises or, you know, make negotiations that, like, you know, maybe I can sell some part of the business that really overlaps. So, like, you know, it can ease some of the regulators' concerns about, like, you know, competition. So, like, you know, maybe they think like this is a chance for, like, you know, get a deal done before, like, the Trump administration's, like, you know, their term ends in 2028. So, like...
So this is just like people trying their hands at big deals. Can you talk about how inquisitive DoorDash has been in the past?
So DoorDash and Uber have been like very busy kind of consolidating the food delivery business. We've seen like earlier this year that, like, you know, Uber has launched a very public takeover bid for Germany's Delivery Hero, and like, you know, same as DoorDash, they have been like kind of buying companies like Deliveroo and like, you know, restaurant software, Olo. Like, you know, everyone is kind of like coming together among a few big players. So like this is something like they have been really active in doing M&A, even though their stock is a little bit down like year to date. But I feel like it seems like something both strategies for, like, you know, Uber and DoorDash is doing like all these years.
Right. And I guess the main rationale here is just economies of scale essentially between these two businesses, or what do you see?
Instacart is kind of one of the leading kind of grocery delivery services, and, you know, like DoorDash also offers grocery deliveries and like among other kind of types of, like, you know, food deliveries and other things. So like it could mean a lot to like bring the two businesses like closely together. There could be lots of synergies, and like, you know, there's of course economies of scale. Like, you know, it would be one of the best-known kind of grocery delivery platforms they can like kind of own. And of course, like the companies have lots of history together, like to try for a deal, and I'm sure like, you know, they have lots of relationships ongoing before already.
Right. Last question for you. I mean, were you expecting this deal to come through? I mean, when this came across your desk, or when your sources told you about this and you initially broke the news, did it surprise you, or was this one that people around you were sort of expecting for a while?
I feel like the Uber-Instacart, the DoorDash-Instacart rumors have always been going around. It seems like it will come up every year, but it seems like, you know, the fact that it's finally happening is like very exciting. And like, you know, the fact that they think it will get done was also kind of like surprising to me. But, well, I have on my phone, I have Instacart right next to DoorDash and also next to Uber. So like it's fun to see like, you know, how that's going to change in the future.
I don't, and I will not do even if it's part of DoorDash. I'm not ordering my groceries. Sorry to say. I like going to the store.
Oh, but we could find out if it's going to get announced like, you know, as soon as tomorrow. So it will be exciting to see how it plays out and how like, you know, investors and shareholders react to the acquisition.
Great. Okay. Well, it was a great scoop indeed. I want to thank you for coming on. That is Valita Pow, our deals reporter here at The Information.
Public software companies have been very loud in their earnings calls about their pivot to AI. Private software companies like Thoma Bravo-owned Anaplan have also been adapting to the AI era. I want to bring on CEO Charlie Goddiner for a conversation. Charlie, welcome to the show. It's great to have you here.
Well, gosh, thanks. It's a pleasure to be here. Looking forward to it.
So, it's been a while since we've talked about Anaplan on the show. Been a while since investors have paid attention to Anaplan. Can you just remind us what the company is all about these days?
Sure. Thoma Bravo took us private in 2022. So, that's why it's been a while. You know, we're a scenario planning and analysis platform. We think about ourselves as serving as decision infrastructure for the world's largest companies. Half of the Fortune 50 are our customers, and they're doing financial planning, financial forecasting with us. They're running their supply chains and planning their supply chains on Anaplan. So those are the two biggest areas that we participate in.
Okay. And so tell me a little bit about the AI strategy here. Do you guys have your own agent that is doing analysis for enterprise users looking, you know, further into the future with data? What does this look like?
Sure. So we've got our own agents. Our own agents are leveraging our core technology to come up with the deterministic answers that our calculation engine provides. And so for what we do for financial planning, financial forecasting, running supply chain, our answers have to be 100% accurate, and that's what our agents ensure that they are. And so if a CFO is not going to issue a forecast that's 95% accurate, they wouldn't have their job for very long.
Okay. Okay. So, Anaplan is the — so all the guidance that all these public company CFOs are using, that they're using tools like Anaplan to basically check whether or not they can issue that forecast and reach it.
Yeah. They're generating those forecasts from tools like ours, in many cases ours, because they're never going to use an LLM to generate a forecast and deliver it 100% accurately. That's the real issue. LLMs can't do it. They're not architected to do it. They're language models. They're really good at predicting the next word, not doing arithmetic at scale.
Well, so talk about that a little bit more then, because, you know, we see these models solving these, you know, thousand-year-old math problems, math competitions, etc. And I mean, I find it confusing then to hear you say that they can't do basic arithmetic when they're doing these math problems. What's the disconnect here?
