SpaceX’s True Value Debate, Microsoft’s New Front in Agent Data Wars, Mercury CEO on $200M Raise

0:13Kevin McLaughlin:Welcome, everyone, to The Information's TI TV. My name is Akash Pasricha. It is Friday, May 22nd. Before we get to today's show, I want to highlight some exclusive reporting from The Information yesterday. My colleague Shreemupiti found out OpenAI's first quarter financials,$5.7 billion in revenue. That is$1 billion more than what Anthropic did in the same period. Codex was a big driver for OpenAI's revenue. There are more details in the story. I encourage you to check it out. Today on the show, we are talking about corporate data wars. Microsoft has entered the fray with its own contentious decisions around data access.

0:53Kevin McLaughlin:We are also unpacking SpaceX's valuation on our weekly segment of the Editor's Cut. We're going to take both a bottoms-up approach and also a market-based approach to our analysis. We're going to close out the show with the co-founder and CEO of Mercury. The company just raised a big funding round. We're going to talk to him about the current state of fintech and what is next. It's going to be a fun show, so let's get right on into it. Microsoft has waded deeper into the corporate data wars with a contentious decision that has ruffled some feathers of certain companies that it partners with, like Databricks.

1:30Kevin McLaughlin:I want to bring on Kevin McLaughlin, our enterprise software reporter, to walk us through the inside reporting that he published this morning. Kevin, welcome back to the show. It's great to have you here. Thanks, Akash. Okay, so what's going on with Microsoft and Databricks these days? So this is something that happened about a month ago, but the ramifications are now becoming clear. Basically, Microsoft, sorry, last month, Databricks launched a feature which allowed its customers to manage their data in Databricks and then do visualizations and charts and things in a Microsoft product called Power BI.

2:05And now what happened was Databricks launched the feature in early March, and then a month later, Microsoft abruptly cut off access to it. And so basically what's happening is this is a battle for a type of software known as the semantic layer, which is going to require some explanation. And it was a little bit of a journey doing this story, so I won't try to attempt that here. But basically, Microsoft wants customers to use its software to build their semantic layer. So does Databricks and so does Snowflake. So everyone wants to be the place where customers define their semantic layer. And so what's happening is this is an emerging battleground of what we've been writing about and what we've been calling the AI data wars.

2:47And I'm happy to go into sort of the deeper explanation of what semantic layer is.

2:52Kevin McLaughlin:Oh, we're going to do that. I mean, I think, I mean, this is core to the story. But I mean, so broadly speaking here, so, you know, we are going to dig into each of these terms. But I mean, this is an issue of basically Microsoft blocking access in certain ways to Databricks. And this being part of the broader corporate Data Wars, AI agent data access story that you've written about for quite some time now. Yeah, this has to do with like basically most of the Fortune 500 uses Databricks, Snowflake and Microsoft. And so what's happening is you can kind of do things as a customer using all three of those providers, or you can just do them all in one provider.

3:37And so Microsoft sort of has a history of saying, hey, you know, you can do everything within our products. And so why wouldn't you? And it's sort of what's happening is that everyone wants to be the place where customers define semantic models. So I wouldn't say it's like blocking per se, but it's sort of like, how do you make it easier for customers to kind of do everything within one ecosystem? And so I know that's kind of a complex answer to your question. Okay.

4:09Kevin McLaughlin:Well, maybe let's try to go a little deeper here then. So explain to us in the simplest possible terms, I mean, what is the semantic layer? and, you know, maybe let's think about it from a perspective of, you know, we use enterprise applications, you know, for our HR stuff. We have a good idea of that. So, I mean, I see what I see, you know, a dashboard, for example. What would be the semantic layer behind that? And then we can get to sort of what's the battleground here. Okay, let's say you're a company. You have raw data that's coming in from retail point-of-sale systems, from sensor devices, from all sorts of places.

4:49And so that raw data is inherently ambiguous. So for example, there's data about revenue, but that could mean net revenue, gross revenue, invoice revenue. And so the semantic layer allows companies to come to common definitions of what revenue means and what those categories mean so that you don't have finance creating their own definitions and marketing creating their own definitions. So semantic layers have been around for a long time. They're a core part of what's known as business intelligence software or data visualization, which is what Power BI is. And Power BI is one of the most widely used products for this purpose.

5:25And so semantic layers are now becoming very important, not only for that, but AI agents need to have these definitions as well.

5:33Kevin McLaughlin:So it's like a standardization of the data type of thing, like an agreement on the definition of revenue, of expenses, you know, dollars, you know, like stuff like that? Yeah, there's standard definitions of data from which different departments and companies can work so that, again, so everyone's not going off in their own directions. And the end result is more accuracy for the reports, for the visualizations. but it turns out that AI agents really need these definitions as well and they can actually be much more accurate and even cheaper to run if agents have these definitions. Now, and I guess maybe some semantic layer 101 here, so would these definitions be unique to each company sort of thing?

6:22Kevin McLaughlin:So, you know, how one company defines this segment of revenue, that would be unique uh yeah if you're a customer and you're using databricks snowflake microsoft you don't want to have to define those definitions in each product but essentially that is what is happening and so um when we say the battle um there there are companies uh out there that are trying to develop sort of standards for the semantic layer so you can do it once and you You can use it across a number of different products. There's an organization or coalition being led by Salesforce and Snowflake, which is attempting to define a standard so that customers don't have to reproduce this work across different products.

7:10Kevin McLaughlin:Got it. Okay, so now that we have a better understanding of what the semantic layer is, so then let's go back to it. So you said that it's not a block, so to speak, but what exactly is Microsoft doing then to make it more difficult to Databricks to access this? What's the issue here? So Databricks launched a beta feature that said, okay, if you're a Databricks customer, you can define your semantic layers inside of Databricks and then you can use them to create charts and visualizations in Power BI. And then a few weeks later, Microsoft, without explaining why, blocks stopped supporting that feature.

7:47So for Databricks customers who were using it, the reports just sort of vanished overnight. They stopped working. Subsequently, and for our story, Microsoft explained that they had concerns over reliability and accuracy with the feature. So they were concerned that customers of Databricks would use this feature in Power BI and they wouldn't get accurate results. And then at the end of the day, customers might blame Microsoft for that. And so it's a valid explanation. And later, a Microsoft executive explained that, look, we're not going to support third-party semantic models in Power BI because it's complex.

8:25There's not really much reason to do that for customers. Again, a valid explanation on technical terms, but perhaps one that also has a business motivation as well. So it's kind of this thorny issue that's happening. And these are all partners that work closely with each other. So these are not like rival companies like putting up toll booths around their data like we've been talking about. This is a little more subtle than that. But at the end of the day, the semantic layer is the new battleground.

8:57Kevin McLaughlin:So, Kevin, I want to go in that. So if this is not the story that we've been talking about, the rival companies sort of putting up the toll gates, like you say, and it's more partners that are sort of developing nuances to the relationship. I mean, why is this something that we ultimately should be paying attention to? What are the stakes here in terms of what this could mean for end customers? Everyone wants AI agents to be more accurate and even cheaper to run. Semantic layer is part of this type of data known as context. So you're hearing vendors talk a lot about context, and some products even have the name context in them.

9:32Context is the rich food that agents makes them operate more accurately. And so if you have the context and you want to give that context to an agent, it even means that the agent doesn't have to reason as much before making decisions, which means they're consuming less tokens and they're cheaper to run as well. And so there are lots of benefits to this type of data. And that's why semantic layer, you're going to be hearing a lot more about it in the coming months.

10:02Kevin McLaughlin:Because this is really, I mean, as long as these customers or as long as these companies are developing more and more guardrails, I guess, around what data is accessible or not, that will then have an implication on how good the models need to be, how much compute is consumed. And so this is actually one of the starting points here of what could very much be these companies ending up turning a profit in the end, if I'm understanding you correctly. Sure, sure. And like I said before, there's this sort of industry coalition that's working to sort of make this a non-issue. And I think that you're going to be hearing more about these open efforts as well, because at the end of the day, the proprietary approach is going to slow things down.

10:48And we've seen this many times before in enterprise software. That's why open source software exists. And so, you know, the proprietary versus open debate in regards to semantic layer, it's going to be a hot topic. It might not be, you know, headline news for everyone who reads tech news, but trust me, it is very important.

11:09Kevin McLaughlin:But we like to talk about it here on the show, so that's why we like to have you here helping us make sense of it. Kevin, I want to thank you for coming on. That is Kevin McLaughlin, our enterprise software reporter here at The Information. SpaceX's S1 gave the world some new data to crunch as we all seek to answer the question what should the company really be worth outside of that this week there were also a number of other big IPO related headlines to unpack all of that in this week's edition of the editor's cut I want to bring on Martin Pierce and Meredith Mazzilli welcome to you both it's great to have you here hey Akash hey Akash how's it going I'm doing well so I don't know if we'll get an answer here but we'll get some versions of an answer what is SpaceX worth is the topic of discussion Martin, you wrote about this.

11:59Kevin McLaughlin:You took a bit of a bottoms-up approach, I would say, to valuing the company in last night's briefing. And I want you to walk you through how you were thinking about coming to your valuation. Sure. Now, I will just say, as Meredith, I'm sure, is going to say, so let me preempt her. Oh, no. I really annoy her. That SpaceX is worth whatever people are willing to pay for it. So let's just get that aside. But really what I wanted to do was to try to clear away all the murk and the sort of drama and mythos around flying to Mars, flying to the moon, all this sort of stuff that Elon loves to talk about.

12:39and just look at what actually is in SpaceX. And what it really is, is a telecom firm. It rents or sells satellite broadband terminals to people around the world and they can get access to the internet from them. That means it's basically like a cable operator. So, you know, that's a fairly plain vanilla business. it's also moving into cloud computing it's rented out some of its you know some of its AI computing capacity to various companies like Antropic I guess that's the only one although it said it wants to do more of that and it's got a small advertising business in X and so then of course it also launches the Rockets And that is maybe one of the businesses that people are most focused on, but it's actually one of the smaller parts of the business.

13:44So what I did was I just looked at each of these three and I looked at the closest comparison in the public market. And I looked at where those companies are trading and I applied the multiples of which they are trading to the revenue figures that we got from SpaceX. And I tried to also calculate, based on the first quarter numbers, what those businesses might do this year. And I came up with a number that suggested that the most generous, and I'm being very generous, valuation for SpaceX would be around 700 and I think it was 50. No, sorry. It was around 700.

14:28Kevin McLaughlin:That's 700 billion. Yeah. Yeah. I think you got to 600 and... 78. $678 billion. The reason I'm a bit confused is I went through various different versions and had to keep adjusting some things. Okay, so$678 billion, whatever,$700 billion. It's not$1.7 trillion. Some people, and you had someone on TIGR TV yesterday, Ross Gerber, who's much smarter than I am, who said it's worth$500. And, you know, I wouldn't argue with that. But it really depends on how you do the calculations and what you compare it to. But let's say it's worth somewhere of that order, 500, 700, something like that. The idea that it's worth 1.75 trillion is crazy.

15:16But, you know, that extra is just... Okay, okay, okay, okay, okay, okay. So we got it.

15:21Kevin McLaughlin:I follow your logic. And I will say to folks that it is a great analysis and a good piece. So I encourage everyone to read it. But, Meredith, I want to come to you now. I guess I'm going to ask you now to justify the$1.7 trillion valuation or at least provide another perspective as to how you think investors might get there. Yeah, look, so I'm not going to sit here and argue that SpaceX should be worth$2 gajillion or whatever. But I will point out that I do think the sum of the parts is just one way to model valuations, and it doesn't capture the whole picture. So I don't think we should put too much stock in that as the golden number of what something should be worth.

16:05You know, as Martin explained, you're kind of adding together what all these individual businesses should be worth on their own and kind of squishing them together. But with SpaceX, it's actually very different. You know, unlike a lot of other satellite companies or mobile operators, they're a lot more vertically integrated. You know, their launches put their satellites into orbit. And there's very few other companies that can do that. So just on that basis, I think there is a bit of an issue, sorry, Martin, with the sum of the parts analysis. And that's before we get to even Starship data centers in space, all of that.

16:44And once you start looking at that, that really just starts becoming more of a scenario analysis. So if this happens, what weight do we assign to this happening? What value did we assign to that? Again, I'm not saying that the big number that is out there for the IPO is necessarily the right number, but there are different ways to think about it.

17:05Kevin McLaughlin:So the other topic I wanted to talk to you about, Martin, anything to add? You know, some of the parts is one, you know, only one thing and it's not doing that. That's true. It's definitely true. But, you know, companies that have, you know, many different businesses usually trade at a discount to that analysis. So you could say that SpaceX is worth$500 or even less. I mean, anyway. Even less? How low are we going to go here? I don't know. But, I mean, look, the issue, the big problem is that we don't know when any of these things will happen. And do you really want to buy something when... It's almost as if you were buying something in a store and the person says to you, this will work, but I don't know when.

17:56And so you're paying an awful lot for, like buying a car. You pay a huge amount for the car and it won't work immediately. But at some point in the future. I mean, it's venture capital. It's venture capital. That's what a venture capital investment is. Let me point this out. You're buying a car. say you're paying three times the price of the car and on the promise that when it does eventually work it'll drive really fast but here's the problem we have no idea if it will ever work and when it will that's my you know that's my analogy for this ipr i mean it's kind of meredith

18:35Kevin McLaughlin:it's kind of what it's like a it's like a venture capital like it's just offering stock market investors a way to be VCs in some ways. Yeah. I mean, maybe the more accurate sum of the parts is you have this functional real business. I think that you can't argue with that. And then there is this venture-like part of it that is on top of it and thinking of the two. And how to combine those together. Well, Martin, and let's just go back to, I mean, so$678 billion, I should say just walk us through uh roughly speaking how did that break down in terms of the uh what you got for the space business the uh connectivity business and then the ai business what was the biggest in the end for well the biggest was the space business and that's um a bit of a strange thing it's because even though the space business actually revenue from that is fairly small um i I was using as a comp a company called Rocket Lab, I think it's called, which is trading at an extraordinarily high valuation.

19:42It's the closest comp to SpaceX, which I thought was the right one. And I also thought, well, given the promise of what Elon is hoping to do, maybe it's fair to apply a very high valuation. That's why when I come up with this overall number, it's very generous. And it's higher than what other people have actually put on it.

20:05Kevin McLaughlin:And I found it interesting. And, you know, Meredith, maybe I'd love your reaction to this. So the AI part of the business, XAI that Martin valued here, I mean, the number that he got to was$165 billion. And that was based on the revenue that it had generated. uh well no that's based on the number that they will generate from and tropic um right but my question for you meredith i mean xai here i mean it it would be the the smallest of the three segments here and it's the one that uh well yeah i mean the um the uh starlink is worth about 150 Oh, smaller. You're also forgetting X. It's worth three.

20:51Let's just think. How much did Elon pay for that? Oh, yeah, 44. Water under the bridge, yeah. 44 billion. He's destroyed$41 billion worth of value. This is the guy that investors want to give all this money to for SpaceX. Elon is very good at destroying value as well as sometimes creating it. Let's not forget that.

21:18Kevin McLaughlin:um okay last topic meredith i i do want to get your take here quickly on the ipo market at large because this was just one ipo related headline that we saw this week we also had the news that open ai is fast tracking things we saw aura has filed confidentially blockchain.com is back in the mix is a company that you and i tracked very closely on our ipo tracker once upon a time It's been sitting around there for a while, yeah. What do you make of this? Is this a window that's going to last a long time? So it will be interesting to see how this plays out. And if this really turns into a have and have not situation, we do have potentially three mega IPOs coming this year.

21:59And then we have these smaller IPOs from consumer companies, crypto, though. We'll see what happens there. But do the big IPOs really just take all the attention, money out of the picture for these smaller IPOs? I think that with consumer companies, there is an argument right now that they're also well-positioned because they're kind of like the AI-proof play, right? You know, like AI is not going to get rid of the fact that you need to wear an aura ring. It might impact maybe some of the software, other add-ons. But there is an argument that some of these consumer brands are pretty AI proof and we're not going to see a SaaS apocalypse there.

22:41Blockchain, I honestly do not know what they're thinking. We have a couple other crypto companies that actually filed confidentially a few months ago, I think even late last year, that are still just sitting there. The crypto market is down. They could just be positioning for optionality, but that one was a bit out of left field. Can I just make the counter argument on Ura? And I do agree that if you look at the comp on that one, I think it's Garmin. Its stock is up 18 % this year. So you might say that Ura has a pretty good chance. But Google has just come out with its own version of this sort of health tracking product.

23:29I've just ordered it, actually. It cost$100. Ura costs how much? $400? And you have to pay some fee? I think I would avoid Ura.

23:40Kevin McLaughlin:Well, but it's a ring. I mean, it's a ring. It's not the wristband. It's a ring. It's a bulky ring. Who puts a Google on there? I mean, well, okay, this is more of a Google taking on a whoop than it is on there. But Samsung, I think if I'm not mistaken, I mean, Samsung was going after the ring category as well. The ring's so over. Okay. All right. Well, we will see. You heard it from me first. Ura is over. Google. Whatever the time. Okay. I've forgotten what it is now. All right. Well, I look forward to seeing how that plays out. I think it's a rich thing. I want to thank you both for coming on.

24:19Kevin McLaughlin:That is Martin Pearson, Meredith Mazzelli, two of our editors here at The Information. Fintech company Mercury raised$200 million in Series D funding at a$5.2 billion valuation. The funding round was led by TCV and also included Andresen Horowitz, Sequoia, and Cotu, and a whole host of other investors. Joining me now to talk about the current state of play in fintech is co-founder and CEO Imad Akoun. Imad, welcome to the show. It's great to have you here. Yeah, thanks for having me.