Palantir (PLTR)

In the early 2000s, my dad worked as a data architect at the Department of Defense. Their data infrastructure was a mess (like most things in the public sector) and leadership was finally starting to recognize the impact poor data management was having on their operations. Defense officials had heard of a small company called Palantir with a software that could revolutionize data management and called them for a demo. Palantir didn’t pick up. Desperate, the team called again and again until Palantir responded: “we don’t think our product is ready yet.”

Knowing my dad this anecdote is probably exaggerated. I’m also writing it from memory 4-5 years after hearing it, so it’s likely inaccurate. But the essence of the anecdote is twofold: Palantir’s software was so valuable that the DoD was chasing it and Palantir’s early R&D was so strong that they initially rejected this advance. I was fascinated by the idea of an offering so valuable that customers had to have it (how would we survive without AWS or iPhones?) – this is how trillion-dollar companies are born.

Today, Palantir is a meme stock trading at 10x sales (mind you, down 75% from ATHs) that most serious investors wouldn’t touch with a 10-foot pole. If you ask the average, retail Palantir investor for their thesis, they’ll probably say some gibberish about AI, a $100 billion TAM, and Cathie Wood. On top of all this, Palantir’s public image is terrible – there’s a slew of articles online about how Palantir enables racism and at one point in 2013, CEO Alex Karp had a bodyguard follow him around 24/7 to protect from death threats. But very of few of these people – the public, investors, journalists, politicians, and possibly even tech bros – understand what Palantir really does.

I think I have a better understanding than most regarding Palantir’s product and vision, so I’ll start by talking about what they do. After that, it’s natural to ask whether Palantir is a compelling investment opportunity especially since it’s down 60+% YTD and given there’s “extreme fear” in the markets. I’ll give my thoughts and a price target.

Before we talk about Palantir, we need to talk a little bit about data management. Let’s say you’re an executive at large healthcare company that manages a few hundred hospitals. The COVID-19 outbreak has just begun. You immediately need to know answers to complex questions like: What hospitals are understaffed or don’t have enough equipment relative to the expected number of incoming COVID patients, given the severity of the COVID outbreak in their vicinity? If we’re short of equipment or staff at some locations, can we reroute surpluses or incoming shipments headed for other locations? Who are our key suppliers and to what extent will they be able to meet our demand? Say you have 10-15 questions like this, all of which inform important decisions. How do you answer them? You pass the questions to a team of analysts who dig into some data and find answers.

But here’s the problem: your healthcare company has thousands of different data sources in various formats (different types of databases, flat files, etc.) scattered across teams and hospitals. There’s lots of missing or low-quality data, sensitive data that only certain individuals have access to, and duplicate data in different sources. As a team of analysts, understanding, modelling, finding, ingesting, cleaning, joining, and finally analyzing the data necessary to answer your questions is a HARD task and can take weeks or months to complete. Most people are familiar with the last step of this process (analysis) but don’t understand how hard it is to find and get data in a shape ready for analysis. This is why the discipline of data management exists – data management is an umbrella term for all the “in-between” work that enables analysts to do their job.

Back to our fictional scenario – you need answers to your questions now not in a few weeks or months. If your hospital had proper data management practices in place, then there would be data catalogs that describe where different data sources are, what they include, and who has access to them, data pipelines that transform data from raw sources into a more useful shape, dashboards for common analytics executives may want to see, etc. You can probably see now that data management is key to making quick and accurate operational decisions. Said otherwise, improvements in data management generate incremental alpha. Every enterprise would benefit from data-driven decision making, so it seems natural that there should be a standardized framework for data management that organizations follow. Unfortunately, there’s not. Instead, enterprises tend to throw together shoddy, fragmented in-house data management solutions that mix up different tools and code to eventually get something “that works.” If this organic approach seems medieval and unintuitive, that’s because it is – this is where Palantir comes in.

Palantir Foundry (their core offering) is an operating system for an enterprise’s data. It’s an end-to-end solution that does all the “in-between” work (examples: connecting to data sources, modelling them and mapping them to a common vocabulary so everyone knows where different data is, monitoring data quality, building consistent data pipelines, dealing with security, pushing analyst insights and deploying AI/ML models across the organization, …). There is no other platform out there like this – Palantir has a 15+ year head-start and their competition are ineffective, bloated, fragmented in-house data management practices and software at customers. I have been blown away by each of Palantir’s Foundry demos – there’s so many small details and features that all come together to create a seamless experience for different end users in the enterprise, be it executives, analysts, or others. Aside from that, the UI is awesome and if you’re a SWE or data person, you can tell how holistic and user-oriented the team’s vision is. In 50 years, I think all organizations will be using some sort of “data operating system” that makes the data tools of today look like assembly.

Hopefully this long-winded introduction gives you an idea of what Palantir is focused on and why their software is disruptive. Their value proposition is at best tangential to AI/ML, its rather a novel approach to data management, helping unlock alpha for organizations. With all this in mind, we’re ready to approach Palantir from an investment perspective.

We’ve seen Palantir has a quality software offering with a large moat and market. Palantir’s TAM has been pegged at ~$100 billion which is cool, but the best companies create their own market. As enterprises start to use Foundry and generate alpha from it, I believe it will force their competitors to adopt Foundry as well because otherwise, they would be at a competitive disadvantage. I wouldn’t be surprised if there’s an inflection point at which mass adoption of Foundry within entire industries begins because a) in-house teams can’t compete and b) the only alternative is Foundry. This is by definition “creating your own market.”

The hurdle to initial adoption of Foundry is that it’s expensive and takes a lot of time to transition from in-house data management software to Foundry. Thinking about buying Foundry as an investment from the customer point of view, executives need to see that the initial CAPEX required for Foundry will yield large enough future alpha to justify a transition. A lot of times, if an enterprise has something “that works,” it’s hard to convince them to transition away from that. But, when there are black swan disasters like a mass pandemic or world war (or on a micro level like Boeing 737 incident), it tests organizational data capabilities – if they break, it’s a wakeup call for executives that catalyzes a transition to software like Foundry. I have no problem waiting around for tail events that catalyze customer acquisition (alongside continued sales efforts of course) until we eventually hit critical mass 2.0. All the signs here point to exponential growth.

One big question remains – what about valuation? I passed on Palantir months back because it was trading at a $50billionmarketcap,fartooexpensivegivenitscurrentfinancials.ButgrowthstockshavebeencrashingrecentlyandnowPalantirhasa$15 billion market cap. At what point does it become cheap enough to scoop up? I’m not going to walk through Palantir’s financials or a DCF (you can find and build that on your own), but with conservative estimates – 9% growth in government customers, 15% in commercial customers, 10% growth in number of “large” accounts (>$40,000,000inrevenue),Igetpricetargetsinthe$7-10 range. There are a few numbers that don’t sit right with me, especially the $1Binstock-basedcompensationPalantirpaidout.Ofcourse,thatshouldbemuchlessthenextfewyearswiththestockpricecratering,butPalantirpayingthatmuchouttoitsemployeesmakesmeabituncomfortable.Maybeit’snormalatsoftwarecompaniesandIjusthaven’tanalyzedenoughofthem.Withthecurrentmacroenvironment,Ithinkyouneedalargemarginofsafety.Saywehaveamarginofsafetyof50$3.50-5 range.

There are a lot of other risks here – an important one I haven’t touched on is image. Palantir has faced a lot of backlash for enabling “racist” systems especially in the defense and homeland security context. My take on this: Palantir enables organizations to have better access into their data. If organizations unethically collect or interpret/analyze data, then that’s a problem with the organization not Palantir. Some may argue that the drive towards efficiency and alpha generation is problematic in a society that is systematically oppressive, but then I’d say the best course of action is to attack the root of that oppression, not neutral software like Foundry. Regardless of what I think, Palantir hasn’t done much to reform its public image and there’s no reason that changes in the near future.

Overall, I love Foundry – the engineer side of me is obsessed with it. The investment case has become more compelling as Palantir stock has plummeted, but it isn’t time to buy yet. Tech is a long way from bottoming but when it does, I believe Palantir will rebound and separate itself as a winner.