Welcome everyone to the Information's TITV. My name is Akash Pasicha. It is Friday, October 24th. We have got a great show planned for you today. First up, I am talking to the co-founder and CEO of Crusoe about their big funding announcement and their ambitions to take cloud computing into outer space. Then, we've got an all-star lineup of reporters coming on to talk about what OpenAI employees are calling the Facebookification of OpenAI. We're also talking about all things robotics. We'll discuss SoftBank's renewed robotic ambitions and how AI and robots can be used to treat cardiovascular diseases. And finally, we're talking to a VC who has just started a new fund about what consumer categories are attractive investments right now. We've got a lot to get to, so let's get right on into things.
Data centers have become one of the fastest growing elements of the AI boom, and Crusoe has become one of the most closely watched companies building a business in that arena. The company has today announced it has raised more than $1.3 billion at a valuation north of $10 billion. It has also positioned itself at the center of the government's Stargate initiative alongside big names like OpenAI and Oracle. And this week, the company said it was working to offer GPU capacity from outer space by early 2027. Joining me now is the co-founder and CEO of Crusoe, Chase Lochmiller. Chase, welcome to TITV. It's great to have you.
Amazing. Thanks for having me.
So, a big week of news and a big day today as well. I want to start with the space thing, though. So you've partnered with a company called StarCloud. You're sending GPUs into outer space. What exactly is the ambition here? Why are you guys doing this?
Yeah. You know, the StarCloud partnership is one we're certainly excited about. It's a reflection of, you know, number one, our energy-first approach to building compute infrastructure. You know, in outer space there are, you know, we have access to solar that's powered 24/7 at a very, very low cost. And you know, there are certain applications that, you know, need to be run in outer space. And our belief is that intelligent infrastructure will be everywhere and in everything. And you know, having access to local inference workloads in space will be a critical part to building out the space economy.
So it's literally, I mean, the ambition here is to have a data center floating in outer space. That's the idea, right?
Well, I think early days is really just focused on having a handful of nodes in outer space to serve local workloads. And I think that's kind of, you know, one of the key milestones. I don't see in the near term, you know, very large Stargate-sized facilities in outer space. You know, I think there's infrastructure that needs to be run locally, and having access to compute in space is an important part to adding intelligence to the space economy.
And some more silly questions here. I mean, you know, let's talk about space debris and cooling. I mean, I know it's cold in outer space, but you know, how do you dissipate the heat and how do you avoid it from knocking into the other satellites?
Yeah. The cooling is actually one of the interesting aspects to this because people think of space as cold, but it's actually a vacuum, which means that there's no medium in which to sort of thermally transfer the heat. So you actually have to radiate the heat. It's actually a more complex problem than building a data center here on Earth. But you know, the StarCloud team has built out a great radiative cooling solution that we feel like will work well for sort of these initial deployments.
So let's talk about building data centers on Earth then for a minute. You guys are key to the Stargate initiative. How's progress going on Stargate?
Things are going very well. You know, we recently announced the commissioning and going live with sort of the first few phases of the 1.2 gigawatt campus that we've been building in Abilene, Texas. You know, we're really proud of the team in terms of how we've been able to execute. We've had great partners with Oracle and OpenAI at that facility, and you know, today it's running workloads that, you know, are powering the AI economy.
One of the questions I wanted to ask you is, I look, things are moving so quickly in this space, and it's not just, I mean, the money is flowing in certainly to fund these initiatives. Are you getting any pressure from partners to move even faster at all for an initiative like Stargate?
Of course. When you move fast, you know, everybody wants you to move faster. You know, I think it's a reflection of the demand for compute infrastructure. There's a massive scarcity of compute today. You know, it's a very broad-based trend that we're seeing across all of our customers in terms of people don't have enough compute to serve the workloads, the ambitions, the, you know, whether it's training a new model or serving their own, you know, inference workloads for their customers. You know, there's a massive shortage of compute infrastructure to meet the current demands that we're seeing across the ecosystem.
What is it like to work with, you know, I'm thinking about Oracle in this equation because Oracle is of all the partners that are involved, they are the public company and they're the stock that the share movement is, you know, it's very up and down depending on any blip of information that comes out. What is it like to work with them as a partner and working with such a big public company like that?
Well, you know, from our standpoint, Oracle's been an incredible partner. They've been super thoughtful, working together, you know, arm-in-arm, you know, to make this infrastructure project a reality. You know, they've been incredibly supportive and thoughtful in terms of the design, the construction, the build, the financing of all of this to make this project a reality.
And what about your own cloud computing initiatives? It's something that you are working on at Crusoe as well. You obviously are involved in the construction of these data centers. Talk to me about what the ambition is here with your own cloud business and where you think that's going.
Yeah. So, you know, at Crusoe, you know, we really think of ourselves as an AI factory company, right? So we're building these factories that can produce intelligence for the world and for the economy. And that starts with, you know, the energy and the power and the infrastructure, the data centers, and it runs all the way up through, you know, the managed clusters of GPUs and managed inference workloads to produce intelligent results for end users and really produce value, this digital labor that's produced from artificial intelligence. And so, you know, we've really built a full-stack solution to really help enable the overall economy here. And you know, we've had great partners in sort of making this happen with, you know, Nvidia, with AMD, with, you know, a number of other partners throughout the ecosystem to bring this infrastructure to life. You know, in our cloud business we've seen tremendous growth, you know, tremendous demand from a lot of, you know, leading AI companies like Cursor and Fireworks and Windsurf/Cognition. And we've really helped a lot of these customers grow as their applications have grown in demand and usage.
And how do you see yourself standing out among all the other cloud providers that are popping up now? I mean, it's a really competitive space.
Yeah, the Neocloud space is absolutely an incredibly competitive space. I think from our perspective, you know, our goal is to provide scaled solutions reliably at a favorable price performance. You know,
You know, we've built an incredible team, ranging from, you know, the low-level infrastructure investments we've made in the engineering stack to create reliable solutions across not just the compute, but, you know, the storage and high-performance networking solutions that are required to meet the needs of a lot of the leading AI players. And then, you know, we have an incredible customer success team and a customer support team that's really helped people work through a lot of, you know, challenges that are associated with running this infrastructure at scale.
But on the price side, I'm just trying to understand the economics here. I mean, how do you get it to be such that the price is more efficient compared to all these other neoclouds? I mean, are you looking to own these chips ultimately? I mean, are you still leasing them? What do the economics look like?
We do own the chips, right? So we purchase, you know, large quantities of GPUs through some of our OEM partners. And you know, we install, we manage, and we operate those clusters of GPUs and integrate them with a lot of our shared services, like our managed storage products, our managed Kubernetes products that really enable customers to run the workloads that they need to serve their own specific workload. And so,
I'm sorry. Go ahead. Keep going. Keep going. From a pricing.
So we've been able to, you know, finance a lot of this through some asset-backed financing solutions that we've previously announced, you know, one with Upper90, one with Victory Park. But you know, we've been able to sort of scale the infrastructure with asset-backed financing. And you know, I think that's been a critical aspect to, you know, standing up the infrastructure is actually having the capital to make it happen.
And on that note, you know, I did just want to ask you about sort of how you think about the level of risk right now in the AI ecosystem. Certainly from your end, you know, if you're owning the chips, they're expensive, and like you said, you talked about ways to finance that. But I mean, how do you think about the level of risk in the AI ecosystem right now and the level of risk for your company given that it's so capital intensive to build anything in this space?
Yeah. No, it's a great question, especially when you have a business that has a tremendous amount of capex. The way we think about this is really, you know, we are constantly doing demand forecasts in terms of, you know, how much compute demand is coming from our customers in the cloud business and how much, you know, infrastructure demand we're seeing from customers on the data center side. And right now, I think the, you know, it's very broad-based, and you know, the demand, like, no one has enough compute, no one has enough data center space, no one has enough energy. You know, the capacity for infrastructure is largely saturated right now, and so we're really focused on building more to sort of meet the demands of customers. You know, when we think about GPU purchases, you know, there is kind of some payback math that we go through in terms of, you know, making the upfront investment in the capex and working with our customers to, you know, ensure that they're able to get a great service and high performance with high reliability, and we're able to earn a great return for our investors.
Chase, I want to ask you a quick question before I let you go. How many times have you climbed Everest? Is it once or twice?
So I've been on two Everest expeditions.
Since 2014, and then I summited in 2018. So,
Okay. So it strikes me you're not afraid of a lot of things, including heights and climbing mountains. What scares you as you build Crusoe? It's such a hard business. It's a crazy time for AI right now. What keeps you up at night?
I think probably the biggest risk to the business, or the biggest risk to the ecosystem, I think really is debt. So, you know, from our perspective, we try to be very thoughtful about how we take on debt and what our offtake looks like. And we have to have a lot of confidence that, you know, number one, we'll be able to pay down those debts through, you know, the customers that we have and the contracts that we have and the demand that we have and the services that we provide. And we'll be able to earn a very good return on top of that. You know, if you look at any of the major financial crises that have unfolded over the course of history, you know, many of them involved irresponsible use of leverage and of debt, taking on sort of an irresponsible amount of leverage against, you know, certain forecasts. And we just try to be conservative with that and really thoughtful about who our customers are and really having a portfolio of large investment-grade customers as well as, you know, some of the leading AI startups.
And you mentioned AI demand. I mean, this is something I think about a lot with these businesses, which is that if AI demand starts to plateau at all, or if it doesn't shake out the way that people think it does, what happens to Crusoe's business in that scenario?
Well, I think it's sort of a matter of what the contracts look like, and we have a range of contracts across our business. But when you look at, for instance, the biggest capital investments, you know, the biggest capex numbers that we have are really at the data center layer for building these gigawatt-scale campuses. And in those cases, we have long-term offtakes from investment-grade customers, right? So these are long-term, very committed revenue streams that we feel will be able to weather any sort of ups and downs that we see in the AI economy and the AI demand landscape. And that's something that we've gotten comfortable with as a business and in our underwriting of a lot of those investments.
Great. Well, Chase, I want to thank you for coming on. That is Chase Lochmiller, the CEO of Crusoe. OpenAI isn't yet one of the big tech giants, but in some ways, it's looking like it may be where the company is headed. My colleagues today published an in-depth look at how many of OpenAI's employees are split on what they call the Facebookification of OpenAI. It's not just about where employees are coming from. It's also where the company's focus is going. Joining me now is a superstar trio of reporters. Stephanie Palazol, Aaron Woo, and Cali Huang cover all things AI, Meta, and big tech at large. Welcome to the preview. It's great to have you here.
Friday energy. I get it. Okay, the weekend's coming. We're almost there, guys. Let's talk about this story that you wrote. Cali, I want to start with you. So let's just look at the numbers here. I mean, how many people who are working at OpenAI right now came from Meta, and how many senior executives also came from the company?
Yeah, so we found that of the people who list OpenAI as their current employer on LinkedIn, about 630 previously worked at Meta. So that's about a fifth of the company. And it's a lot of the company's leaders. Fidji Simo, the CEO of applications, was formerly head of Facebook. The CTO of applications also used to work at Facebook. And if you go through the C-suite, it's a lot of people who had leadership roles at Facebook in the late 2010s, early 2020s.
And but people are also going to Meta, right? Because we've written a lot about Meta poaching a lot of this talent.
Yeah, of course. Meta recruited very aggressively from OpenAI this summer.
And is there anything about this that is OpenAI trying to get back at Meta, or maybe, you know, is it something about talent and Meta that is really good, or anything like that?
I think, you know, I think a lot of this hiring has been happening since before this summer. But there are very few companies that have operated at the scale at which Meta operates at. So, you know, as OpenAI grows up, there are only so many companies to hire from.
So Erin, how has that sort of influenced the direction that OpenAI might be going now, with not just the company strategy but also the company's culture?
Yeah. So we wrote in the story about how there are some processes that employees feel like have been taken from OpenAI. We talked about Kevin Weil, who wrote this long product memo where he talked about wanting OpenAI to become like a tooth product that people use every day. Those are words that are borrowed from Larry Page, who's the CEO of Google. And so there are some people who feel like this borrowing from big tech is giving processes to a company that has been quite chaotic, has grown really, really quickly. And so maybe some of this isn't a bad thing. But there's also a focus on process and bureaucracy in terms like directly responsible individual and like SEV and like things that have been borrowed from big tech companies that for some make them nostalgic for the days of being like a scrappier, smaller startup.
There was some good detail in the story about this survey that went out. Tell us about that.
Yeah. So earlier this year, there was a survey that was sent out by kind of this employee-driven task force that was essentially asking people who had come from big tech backgrounds whether they felt like the culture at OpenAI was becoming too much like a big tech company and what they were meant to kind of do about that. So it's interesting because actually originally the survey was kind of framed as the Facebookification or the Metaification of OpenAI, and then later was kind of broadened to include just broader big tech companies. But it really shows the kind of influence that Meta and Facebook has been having on OpenAI and the mind share that that big tech company specifically has had kind of within this group of OpenAI employees. And so this is actually a survey that employees asked to have answered.
Yeah. So at OpenAI, they sometimes set up these kind of task forces which go after these big questions that they might be wondering about, like are we becoming too much like Meta, for instance. And so yes, these are kind of employee-driven surveys and task forces.
Well, Cali, this is kind of an interesting time for a survey like this to go out because then you look at the products and you look at Sora, for example, and OpenAI now moving into the social media realm. It's, I mean, it's not just culture. It's sort of taking after Meta a little bit in terms of business model too, right? I wonder how employees feel about that.
Yeah, I mean, some people, frankly, we found are not so happy about it. They've expressed doubts that Sora is the right direction for OpenAI. They're worried about OpenAI's ability to handle these problems that have long plagued social media companies like Meta. So content moderation, for example. And in the lead-up to Sora's release, Sam Altman tried to ease some of those anxieties. He told employees, you know, Sora doesn't necessarily further OpenAI's goal of superintelligence, but in a future with superintelligence, people are still going to want fun, light-hearted content. And Meta has actually had a very similar recruiting pitch to researchers.
Steph, I kind of want to come back to you because I sort of think about the strategy that Sam Altman has sort of built for the company, and it very much is a bit of the spaghetti strategy right now, where they're trying a lot of different things at once. Is this kind of inevitable as a company grows up in your mind? I mean, surely there must be people at the company who are actually saying, well, yeah, I mean, the goal is to be as big as Facebook. We have to go through this transformation.
Yeah. I mean, I think in some ways like OpenAI is obviously a company that wants to make a profit someday, and right now it's on track to burn $15 billion from 2025 to 2029. So it's obviously going to have to come out with products that are making money to make a dent in that cash burn. So I think in some ways people think that it is kind of just a part of growing up and a part of being a big tech company. But at the same time, I think it's just strange, I think, to see, you know, Sam Altman talk about AI that can cure cancer, that can fix the education system, and then at the same time you see products like Sora coming out, which are, you know, obviously just very similar to kind of like social media companies we've seen in the past, things like TikTok. And we've also seen at the same time even Sam change his viewpoints on things like ads, going from saying that he finds ads uniquely unsettling to saying that there are types of ads that he is interested in having at OpenAI.
Erin, I kind of want to take a step back here and just talk a little bit about Sam Altman as a leader. Mark Zuckerberg at one point was in his dorm room, and he's obviously grown the company into this big juggernaut. OpenAI right now is, they're at about 3,000 employees, I think is what we said in the story. And so they still have a ways to go to be as big as Meta. But I mean, help just riff with me here a little bit on the parallels between, you know, Sam Altman as a leader, you know, Mark Zuckerberg's trajectory. You know, do people sort of see him as the next Mark Zuckerberg? Do they sort of hope that he might avoid some of the challenges that Zuckerberg had to go through along the way? What's people's thinking around that?
Yeah. So I've been thinking about this for the past couple years. There's like a message from me in Slack in like 2023 where I'm like, Sam Altman is going to be the Mark Zuckerberg of this generation of technology. And I've been kind of wagering that Sam Altman has also been thinking very specifically about this. And I think a lot of what he's done has been trying to position himself, maybe like to try to position himself differently from Mark Zuckerberg. Like one of the things about Sam Altman is for a long time he was very accessible to the media. Like he's been around forever. Like people have, in the lead-up to this, I was rereading an essay that our founder Jessica Lessin wrote about like knowing him since he was a kid in the 2000s, like working on Loopt, his first basically failed startup. So I think he's been around forever and he's seen what's happened with Mark Zuckerberg. And I think for a long time he was trying to position himself as very accessible, very sober. He cares about like the harms this technology could have. Like a couple years ago he was on the Hill doing this whole big listening tour where he was talking about the kinds of
AI regulation. He wanted, like, meetings with Congress. So I think for a long time he was trying to position himself as almost, if anything, like the anti-Mark Zuckerberg. I think as the company has gotten bigger, as it's harder to do the like, 'ah shucks, boy wonder' thing, I think it works less well. But I think it's a really interesting parallel that will be worth continuing to pay attention to.
Great. Well, Steph, Cali, and Erin, I want to thank you for joining us. It was a great story. It's on our website right now. And with that, we're going to move to our next segment. SoftBank is renewing its robot ambitions even as prior efforts over the past decade have faced hurdles. This week, The Information reported that the company discussed a $900 million acquisition of Agility Robotics, a 10-year-old startup led by a former Microsoft executive. I want to play a conversation I had with our very own Rocket Drew about why that deal fell through. Here is that conversation.
Hey, Rocket. How you doing?
Hey, I'm doing good. How about you, Akash?
Doing good. Okay, so SoftBank is doing robots. I guess it's been doing it for a while. I haven't been paying attention to it. But tell us about this company that they were talking about buying.
Yeah, that's right. So, SoftBank has been in the robotics game for a long time, but what we're reporting is that SoftBank and Agility Robotics held talks about a possible acquisition late last year. And a price that was discussed in that meeting was at least $900 million for Agility. SoftBank has been interested in robotics for a long time, but AI is igniting a new wave of interest in robotics for many investors, and SoftBank seems to be no exception.
And why didn't the deal go through?
