Thank you everyone for joining us. My name is Keith Weiss. I run the U.S. Software Equity Research effort here at Morgan Stanley. Super excited to have COO Shyam Sankar from Palantir joining us this afternoon to talk about very big and what I think is a pretty exciting story going on at Palantir. Quickly with a brief research disclosure. For important disclosures, please see the Morgan Stanley research disclosure website at www.morganstanley.com/researchdisclosures. If you have any questions, please reach out to your Morgan Stanley sales representative. With that out of the way, thank you so much for joining us.
Great to be here. Thanks.
I think this is an exciting time to be talking about Palantir and talking with Palantir, both from multiple directions. One, in terms of sort of the product, as you mentioned, I think there's a lot of really exciting stuff you guys have been doing with the product, that make it a lot more interesting on a go-forward basis. It was interesting before, even more interesting on a go-forward basis. There's interesting stuff you guys are doing on the go-to-market side of the equation, and just kind of where we are in terms of the geopolitical world right now, all kind come together to make it a really opportune time to be able to sit down and talk to you. I really appreciate you coming in.
Maybe to start out, just from a kind of high level perspective, Palantir is known as solving, like, the world's hardest data problems. Unpacking that, I think people underestimate how much is under the hood, if you will, how many components of sort of the data equation that you guys have put together into your platforms and your solution portfolio to get it done. Can you give us a bit of an overview on sort of the strategy, vision, and sort of the product behind Palantir?
Yeah. Over the last 15 years, we've had a maniacal focus on solving these problems in an end-to-end way, and we've developed a lot of tech as a consequence of that. I think the most cynical way to think about Palantir is that it took something as sexy as James Bond to motivate engineers to work on a problem as boring as data integration.
Right.
Right? It started with this idea that we were deeply committed to helping a handful of institutions in the world solve problems around counterterrorism.
Right.
When you started engaging with these problems, you realized that the immediately precedent layer of the stack didn't really work. As soon as you solved that, you realized that the one before that didn't really work, and that if you were going to be the company we wanted to be, and if you were gonna have the impact in the world that we aspired to have, you had to take on end-to-end ownership of the problem.
Right.
Having that sort of focus over 15 years has really meant that we've developed three capabilities that I think we are still world-class at, and there's really no competition for. The first is that our platforms make the marginal cost of data integration effectively zero. How much time and money is it gonna take me to get to the starting line? How long will it take you as the federal government to integrate with each of the 6,000 hospitals in the U.S. to create the first national level of visibility into ICU bed utilization and PPE consumption rates? How long will it take you as an automotive OEM to integrate with your tier one suppliers to have visibility into your cost of quality and cost of warranty?
How long will it take you to bring together 400 source systems from the Army to have the first opportunity to see yourself as an institution?
Right.
By creating software-defined data integration, where essentially the software is writing the data pipelines, not people, we're able to radically transform that part of the equation. What I think that means in practice is that our customers can go on offense with data integration. Instead of thinking about it as this albatross around their neck, this kind of cost to manage, you know, what is the least amount of data I can use to survive, it becomes, you know, it's so cheap and easy to bring it all together, how can I use the fact that I have more disparate data than anyone else to create alpha in the market, whether that's having a better quality program as an airplane manufacturer, or that's having an ecosystem that integrates your information with your customer's information in a way that creates stickiness and transformation.
The second major pillar of our product is that gets you to the starting line. I've integrated all this data. That's really useful. Only if I do something with it to make a decision, right? It's not inherently valuable unto itself. The second pillar of what we do is we make the marginal cost of application development effectively zero. Importantly, it's application, it's not dashboards. This is not read-only insight into some data asset. It's actually about making decisions, making decisions as a blue collar worker. You know, if you go to any Chrysler factory in North America, you will find Foundry on the factory floor used by every single team leader at the assembly line to make quality decisions in real time.
Right.
Having that sort of software that creates a connection between the data scientist and the front line and translating that to action. The third major pillar of what we do is simulation and modeling. The buzzword around this is digital twin. As a consequence of simply integrating your data into Foundry, you now have a digital twin of your entire enterprise. That proved to be extraordinarily valuable in the context of the multiple shocks that were happening throughout COVID and the pandemic. My supplier gave me 50% of what they told me they were gonna give me. What do I do? By the way, what do I want to do? Am I optimizing on margin or revenue or customer satisfaction? How do I ask counterfactual questions of my entire operation?
That creates an enormous amount of resiliency and sort of your responsiveness, your ability to seize opportunities in the market. We bring these things together in a capability that we can deploy in hours. The real focus that we've had over the last 18 months is how do we take this monolithic capability that, you know, is uniquely. We built Operation Warp Speed. We got every vaccine distributed in the U.S. and the U.K. You know, we did the Afghan evacuation with 22 hours notice. Monolithically, it clearly creates value. How do we chop it up and make it highly modular and make it something that helps IT accomplish the reference architecture as opposed to something that's coming up and blowing up their whole world? How do we build a sales force around those product innovations and the modularization?
I think that's the exciting thing that we saw in 2021 is the quintupling of the U.S. commercial customer count and the incredible growth in the new revenue from new 2021 customers, $53 million, which is pretty substantial relative to our past performance.
Right. That's a great transition to the next kind of part of the equation I wanna talk to, which is the evolution of the go-to-market strategy. That monolithic platform solving the solution end to end, that's how the federal government pictures technology, right? They have a problem, they want you to come in and solve it. What we found when we're talking to, like, commercial entities is, hey, listen, if I have a $200 million problem, Palantir makes a lot of sense. If I have a $20 million problem, it's hard to fit it into my architecture. One of the things that I heard from Palantir, particularly on the most recent conference call, and I upgraded the stock on the back of it, was, you're changing that dynamic. Is the...
You've been seeing it in terms of the product. You've been creating a more modular product on the go-forward basis, but also the go-to-market is-
Yeah.
... less adversarial at the IT department. We're not trying to replace, we're trying to augment and help you out with what you're trying to do.
That's right. I mean, I think, look, we've kind of screwed this up historically. I'd say we were really focused on creating extraordinary business value, selling to business users, avoiding IT, and what-
Right.
We've really pivoted the sales force to sell to IT first. I don't even want you to have a business conversation until you understand what IT is trying to accomplish, what the architecture is that they have in place, what the modules are that we have that we even apply there, and then let's take that knowledge and then go find how we can create business value and bring it together in a way that both sides are really gonna like.
Yeah.
We're seeing that acceleration in the sales cycle where the monolithic sale could happen really fast in the context of a crisis, but absent that, you might be talking about, you know, a year or longer. With a modular approach, the product that we have around HyperAuto, that's the software-defined data integration for ERP products, it's a three-month sales cycle, it's a one-day pilot, it's a price point IT can make a unilateral local decision on.
Right.
It creates an upsell motion for us. It is the land that allows us to then expand. As we go to the business, we solve more problems. We're already part of the blessed architecture, and it's a win for everyone.
Right. I think there's some also synergy between sort of product development and sort of the trying to help out IT more with some of the projects like Apollo, right? You're trying to make the IT department's life easier and more efficient and more cost-effective. Can you talk to us about Apollo and why that's important?
Yeah. We talked about Apollo around the listing as really this kind of third platform that we had built for ourselves. It's our continuous delivery infrastructure and so much more really our platform engineering and production infrastructure capability. What it's allowed us to do is to deploy our product as SaaS environments that you typically don't have market access to SaaS for. So on-premise Swiss banks, where there, you know, it just needs to be on-premise in Switzerland or into exotic edge environments on the factory floor where there is no cloud given the data scale and the cost of dealing with the cloud. Or more exquisite environments like at the edge of a car in an automotive OEM context.
Right.
How are you gonna manage your software in all these contexts where you don't just have a small number of public multi-tenant cloud instances without giving up all your margin?
Right.
Apollo allows us to automate that. Our engineers write code once. That code is securely deployed. It has automated service level obligation adjudication to know that the software you're delivering is actually at least a performance improvement or not a degradation, and it automatically adjudicates that, those factors. It has built-in security, so when you have a vulnerability in a code package, it automatically blacklists that package and rolls back to a prior good known version or patches forward with new releases. It's integrated. It has a software supply chain capability. It's integrated with how you write code. Each of your code commits are actually signed with a hardware token. I know that Keith committed this code-
Right.
-not a Russian hacker spoofing Keith trying to get a backdoor into our software. I think we're less than a year away from a world where the U.S. government requires all vendors who sell software to the government to have an audited software supply chain-
Right.
-given the amount of risk we've seen between Log4j and SolarWinds around that.
Right.
We would love to bring, you know, this tooling has been transformative for us. We also think, to your earlier point about geopolitics, the world is getting more fractured here. You know, the U.K. is no longer part of the EU. If you wanna really go after market access, you're gonna have to have many environments specific to the markets you're in that are running your latest and greatest code.
Right.
... that don't overly burden you as a company. Making our customers' lives easier as they're writing first-party software is a big part of the Apollo vision.
Got it. As we talked a little bit about sort of the product and modularization, can you talk to us about where we are in ramping up the sales force? 'Cause you need people on the ground to go out and be able to sell this. What were you able to accomplish in 2021? Sort of what's the goal into 2022 in terms of getting that commercial full to sales force?
Our real focus over the last 18 months has been, can we get it to work really well in one place, the U.S.?
Yeah. Right.
We exited the year with about 80 folks. We started the year with 12 folks in the U.S. commercial business. We exited with about 80 folks, of which 25 have been here 9 months or longer. What we're excited about, the work over those 18 months was coming up with the modules, ensuring the modules had the right sort of product market fit.
Right.
The ability to figure out the buyer personas, to drive that into the sales force, where we have the sales force organized both by accounts, and the accounts they're covering, their job is to be an expert in the customer, and by these horizontal overlays, the modules essentially. How do they go and meet in the market between these two things? I think the result, what we've really been able to show is that we can drive down the cycle time on these relationships. We can nail a land-and-expand motion that's driving to expand.
Right.
We can meet the customer where they need to be on price and still have a way of capturing the value over time. We hit the ground running very hard this year, building out the same sales force, same approach in Europe, and continuing to hire at the same pace in the U.S.
Got it. I wanna touch on one of the more controversial parts of the commercial business, which is your strategic investment portfolio. I think you've done nearly $400 million in strategic investments this year, and a lot through SPACs and those strategic investments contributing nearly $50 million in revenues. Trying to understand sort of the vision and the philosophy behind that. From a high-level view, it doesn't seem too different from when we hear Snowflake Ventures, and they want to sort of invest in early-stage companies to build upon their platform. Or I was talking to Databricks today, and they have a similar kind of initiative going forward. Is that the similar ilk in terms of what you guys are trying to do with these strategic investments?
Yeah. Let's talk about what is strategic about the strategic investment program. I think the business, the aspiration, like what is the highest potential use of Foundry? I think it's application development infrastructure. It should be where you go to build the applications of the future, not AWS, not Azure. Like, these things are great and they're fine and they work, but they're rather highly unopinionated. If you're going to go build something like Skywise, which is the aviation ecosystem that Airbus built on top of Foundry, it has 160 airlines that run their operations on this software. It's integrated with their top 15 suppliers, and it generates over $3 billion worth of value for them a year.
You know, we started working on Skywise two years after Boeing and Microsoft announced their own aviation ecosystem, which is nowhere.
Right.
It's not because Microsoft's tech doesn't work. It does work. It's because most of the hard work is in front of you. You know, just saying that you're going to use this storage and this compute is not how you get to value. There's so much risk and execution that's left for these companies. What's strategic about it is, like, we have invested so much opinion. If you are building applications that require you to integrate disparate data to generate like business logic and decision-making interfaces for normal human beings, I don't know how you could do better than starting and building on top of Foundry there.
Okay.
I am so excited to work with all my customers who have supply chain problems or want to work on pricing or integration in production, but that's not the strategic investment program.
Okay.
The strategic part of this is Foundry is our AWS.
Right.
Finding that, it really pushes us in both ways. Like, we have accomplished so much in terms of what our product can do in that category. We've also been able to partner with companies to bring the cutting-edge technology we first created in government around Edge AI and bring that to the commercial world. There's clearly if you're building autonomous vehicles, you're generating about 10 TB of data an hour on the vehicle.
Okay.
It is completely cost prohibitive to bring all that data to the cloud, so you need to do the AI edge inferencing, the data analytics at the edge. Being able to have the right sort of partnerships to transition that and expand the surface area of what our platforms do is incredibly valuable.
Got it. On a go-forward basis, should investors expect a similar kind of level of strategic investment on a go-forward basis in those programs?
I think we're still going to do strategic things. Like, the SPAC market doesn't exist anymore. It's going to manifest in different ways.
Right.
It won't always involve putting capital to work in quite the same way.
Right.
Like, we're deeply committed to being the infrastructure that people build their business on top of. And that will manifest in many ways. I think you should expect the SPAC program to basically peak now-
Right.
-of the 30-some-odd customers we added in Q4, 0 were SPACs.
Right. Got it. That's helpful. Last thing I want to touch on the commercial side of the equation is, evolution in kind of the pricing model and how you guys are thinking about that with the increasing modularity and also sort of the evolving kind of use cases that you guys are bringing up. Can you talk to us about sort of where pricing was 18, 24 months ago and how you see that pricing model evolving?
Yeah. There's been a lot of evolution there. Historically, like 18 months ago, there's really only two ways that you could think about licensing something like Foundry. An enterprise license, that's all you can eat.
Right.
Rather expensive, or you could license it per use case to kind of ring-fence where the aspiration was really after two or three use cases, you would aspire to have an enterprise license.
Right.
That created a lot of barriers to entry, obviously. The big thing we've been working on is consumption-based pricing that allows folks to start with essentially no commit. Start where they want to start on problems that they're scoping to, and then slowly over time consume more and more. We're rewarded for the consumption that they have, for the value they're getting out of the product. It motivates us to invest in those capabilities. It also allows us to activate the partner ecosystem in a way that we really couldn't before with use case-based pricing.
Right. Got it. I'm a big fan of consumption-based pricing. From my perspective, it kind of solves the original sin of software of selling something that they're not using, right?
Yeah.
Shelfware. You don't get paid until the customer starts to see value. How should investors think about the ramp there? Like, it's not really a big part of sort of revenues today. What's the expectation of sort of how quickly that could ramp within that commercial business?
Yeah. I think it's too soon to give, like, something concrete on the ramp there. Like, we're in the market now. We've closed usage-based pricing deals really starting this quarter, Q1. I think you'll start to see it layer in, and expect some commentary from us.
Okay.
As that starts to happen.
Got it. I'd love your opinion on sort of the one of the big investor bases that have kind of come up over the last two weeks post Snowflake, right? Within these consumption models, price performance improvements that you see from sort of getting your underlying platform to be more efficient and more effective on a going forward basis, should that be passed through to the end customer, or should the vendor kind of accrue that in margin and see the benefit today? How do you guys think about those price performance improvements and sort of whether that's important to pass on to the customer?
If you're building the company for the long term, I think there's no doubt that you want to continuously invest in price performance improvements.
Right.
You're gonna be able to expand the number of workloads. You're gonna kind of go from the highest end, most exotic, hardest problems to the preponderance of the enterprise, almost to the point where you've hit everything that's kind of banal. There are real network effects in the software. To the extent you are a company with real network effects in your software, you want to maximize the adoption that you can really have. That creates a kind of deeper gravity well, what you're doing. I think you have to be really thoughtful about that because it's just clear that cloud deflation has to come. Unless there's only going to be three companies left in the world, the revenue can't keep growing more slowly for your customers than their cloud bills.
Right.
If you're not investing in cloud deflation. And by the way, when you're trying to deal with this price performance, there is no overnight solve for that.
Right.
That's a lot of engineering work. It's real effort. If you're thinking about how am I going to continue growing my relationship with this customer over the next 5, 10 years, I think it's crucial.
Yeah. I agree. I wanna switch gears to the government business. The government side of the equation is one that I've always been very optimistic about and very positive on. I think, to a certain extent, I think investors overlook your positioning there, right?
Right.
Can you talk to us about almost like the concept of lighthouse accounts and how that occurs in the federal government, particularly the U.S. federal government, and sort of the permission it gives you to sort of spread out into additional agencies and do more?
Yeah. You have this interesting dynamic in government where, you know, I think in the commercial world, largely it's all competitive.
Yeah.
You know, in government, these agencies are kind of naturally, as humans we're competitive, but they're fundamentally cooperative as well. If you have immense network effects as a result. When we are able to solve some of these very hard problems in under incredible timelines, it sets the standard for what is the requirement that the acquisition executives across services, across different agencies, even if we've done this project in one place, are gonna be converging on and going after. I think it plays to a lot of our strengths and our ability to invest outside of the government's own expectation or vision of what can be done. You know, I think we seize those opportunities really well.
If you look at how much product that we have created since the listing, we have our Edge AI capability, which did not exist at the listing. That's gonna be launched into a hardware bus of a satellite this month in space. The work that we've done with MetaConstellation, we were really, you know, obviously, we had no knowledge of what's going on in Ukraine, but it is being used in anger by multiple Western allied services to really observe and to take space from an intelligence domain, where you're looking at what's happening every week, to how can I use this in a real-time decision-making sort of basis?
Right.
How do I pair the imagery I'm getting from space with my AI models to drive my common operating picture? How does that feed into Gotham in terms of how I'm then running operations? It's been really exciting to see that. We've also seen a lot of acceleration based on these sorts of exigent events. The work that we did with CDID, I mean, that Army program was designed for a land war with Russia.
Okay.
You know, when the moment came, it was ready to be used. Like, okay, where can these tanks go? What's the restricted terrain based on the elevation? What's navigable? How long will it take them to get there? How do we help plan around these things? Really, you know, in so many different dimensions, we were there to meet the moment. That's really exciting. It's obviously exciting from a mission perspective. It's what we get up and are excited about working on as engineers. It catalyzes the growth of the business. It creates. It doesn't show up in a quarter.
Right.
It creates all these tentacles. It defines what good looks like, whether you're the Air Force or the Army or the Marines.
Right.
It starts informing the acquisition philosophy. In our case, this is commercial product that exists.
Right.
It's not something you have to go build 2 years from now. You could get it tomorrow.
Right. Got it. On the government side of the business, I think one of the reasons that investors perhaps underestimated and perhaps ignored is it's so complicated to understand how the government sources, sort of, how the funds start to flow through. I wanna kind of rewind the clock a little bit and start back in 2018 when you guys won a court decision that really opened up a lot of these federal opportunities to you. One, let's start there. Just like, why was that so important, and what's been the follow-through? Like how good of a follow-through have you seen on sort of that initial kind of entry point?
Yeah. The court decision was really based around, you can think about it as build or buy.
Yeah.
You may remember, I think it was in the 1990s, there was this issue where we were buying like $700 toilet seats.
Yeah
That had gone to Home Depot. Basically, the law was, you need to buy things that exist, and to the extent you have really exotic requirements, you first need to see if you can change your requirement to buy things that exist rather than custom building it.
Right.
As you can imagine, with the incumbents in government, there's no interest in them enforcing that. That's against their business model.
Right.
Having some teeth around that was very helpful. It also coincides with the time where the department and government more broadly wants to transform. They can't keep affording to spend this much. They can't have things take as long. There's an interest in bringing the kind of greater power and innovation of America to bear on these problems and finding non-traditionals. That is as important to the acceleration that we've seen on the back of that. That's created a lot of new opportunities where we can start, you know, I would say something like Space Force, we started at $100K a couple of years ago, and now it's $24 million a year and growing. You have this opportunity where they really view what we've done, they call it Warp Core, as a software factory.
Okay.
This vision of Foundry as infrastructure to build applications on top of. It's where they go to build their own applications that drive their dominance in space.
Right.
Having these lighthouse customers then reflects on how other agencies and organizations can adopt the same thing. I think the whole world has changed. You know, Europe is not the same place it was two years ago.
Right.
This sort of, to the degree that some of those militaries were kind of performative, they're not gonna be performative going forward.
Right.
You see that with the Germans committing EUR 100 billion to modernizing their force because they realize the threats are real. I think that's also gonna create a lot of market access, not just because they need it, but they also are gonna need it in the context of collaborating with allied forces.
Right. You're jumping ahead to part three of my question, but part two of the question before we get there. Government business in 2018 unlocks, and the growth rates are extraordinary. Continuing resolutions hit, right? Can you explain to us sort of what that means from, like, a Palantir side of the equation? You have the existing contracts. What is that, sort of, what's the impact in terms of being able to get new contracts rolling versus executing on existing contracts?
Yeah.
What kind of drag does that exert?
The principal drag that these continuing resolutions create is that they prevent new starts from occurring.
Yeah.
Existing programs can continue. Things that have been budgeted for more or less can continue, but things that are new awards they get delayed until this happens.
Right.
Now, I think there's really complicated dynamics that happen when a budget does get passed. If we get a budget imminently here, that whole money needs to be spent by September 30th.
Yeah.
In reality, you can't spend money on September 30th. That means everything has to more or less be obligated in July.
Yeah.
You have from April to July to go spend an entire year's worth of funding. That creates volatility on the upside, where lots of projects are gonna get additional capital to go even faster.
Right.
Those projects can be shaped by real world events. Other projects may kind of get caught up in this sort of blip to get the money out the door.
Got it. Part three, there's exogenous events, and you have a geopolitical crisis like Ukraine. Emergency spending bills start getting passed. How does that change the dynamic? Does that free up spending and free up dollars that could be put to use for solutions like Palantir?
It absolutely does, and I think it plays to the strength of our business model where, you know, we're not waiting for that money to be obligated to start innovating.
Okay.
Like, we are there literally on the front lines building as much features as we can for the generals who are running these campaigns.
Right.
Creating as much visibility across the allied force. That innovation is like a punctuated equilibrium, and it creates a bunch of IP, a bunch of interesting products that then can be systematized and brought to bear against the government's problems through the funding that is available.
Got it. That sounds very bullish to me.
Well.
A whole really-
The thing I'm most excited about in the government is I think we have the opportunity to be the first software prime. Depending on how you count it.
Yeah
Either there's, you know, 4-5 primes.
Yeah.
They're principally hardware companies.
Yeah.
With TITAN, this project that we've taken on to build a tactical ground station, we are the prime contractor as a software company.
Yeah.
We're using that unique position to understand how do we totally transform this asset, which, you know, connects to space constellations from the ground using software to fundamentally reimagine this. It's a little bit like what Tesla did to the car.
Sure.
I think that's gonna create profound opportunities in the government space.
Yeah. I don't think there's much debate that you guys get to work on the coolest shit. The uncool part of the equation. Let's talk about margins a little bit. Operating margins had a tremendous expansion ahead of the listing, through the listing, into 2021. For 2022, you are looking for operating margins to come in a little bit, and investors are kinda trying to understand the shape of that curve, right? It's not 30% in the near term. Is it in the long term? Is 27% at the bottom? Like, how should we think about where those operating margins level out and maybe trend to over time?
Yeah. I mean, 2021 shows, you know, high growth and high free cash flow, and that's what we're doing, and that's what we think we're gonna keep doing here.
Right.
We're giving ourselves some space as we invest. We saw incredible movement on the margin just given the investments in the product and how they paid off. Frankly, they drove more improvement faster than we actually thought they might.
Right.
Now we're investing significantly in the go-to-market. We're giving ourselves a little space there to invest as aggressively as possible. I think we're in a range I wouldn't expect it to deviate too much.
Okay. Got it. Stable margins on a go-forward basis. In the 36 seconds we have left, any questions from the audience? Outstanding. Maybe just to leave on kinda one last question from me. I think you touched on this a little bit. We talked a lot about sort of the product innovations on the commercial side of the equation around Apollo and Foundry. What's most exciting in terms of the evolution of Gotham on a go-forward basis on the federal side of the business?
Well, I mean, Gotham has taken so much surface area in terms of operations, so.
Yeah.
You know, Gotham used to be used in anger every single night on counterterrorism operations. As the U.S. and our focus as the West has pivoted to near-peer adversaries, to Russia, to China, and seeing these things, the innovation has really been on integrating the entire suite of sensor and shooter capabilities from space to mud. How do we better leverage space? How do we bring it to these domains? How do we AI-enable that so it's helping commanders make decisions, not just understand what's going on, but understanding what they can do about what's going on there. That is the single pane of data that creates the single pane of glass to run your operations on. You see that with MetaConstellation bringing-
Right
- space in. You see that with Edge AI. Like, how do I push that inferencing to the edge? How do these things operate in survivable disconnected modes? How do we kind of upgrade all the hardware we do have to be AI-enabled?
Right. Outstanding. Well, thank you so much for joining us. This has been a super interesting conversation and very exciting to see the evolution of Palantir continue.
Yeah, thank you. Thanks for having me.
Thanks.