Our Government Is Spying

An in-depth stock analysis on Palantir Technologies Inc.

Picture by BNN Bloomberg | Source

Greetings fellow investor 👋,

I’ve done the research so you don’t have to. This week’s stock report for premium subscribers: Palantir. All premium subscriber perks, here.

There’s a scene in The Lord of the Rings where Gandalf tells Saruman how the Palantír (one of the seeing-stones) is corrupting him, exposing him to danger by letting Sauron conquer him with evil through the stone (more on this below if you’re not familiar with The Lord of the Rings). It turns out that one of the largest data technology companies in the world took its name from this exact stone in the J.R.R. Tolkien story. The same company that was funded and currently works to help U.S. intelligence manage and learn from its data.

This company, being Palantir. Palantir is currently a $155 billion global technology behemoth, with its stock price fuelled by the AI craze. Since 2022, its shares have surged over 500%. You may have seen its name around social media (as I have), where investors and followers of the stock have been making overly bullish posts. But what does this company do? How does it make money? More importantly, is its valuation justifiable, even accounting for its high free cash flow growth? Buckle up, today’s a long one.

Let’s get into it. (32 min read)

---

Today at a glance:

1. The seeing-stones (history)

2. Business model & analysis

3. Management quality

4. Stock performance & financials

5. Risks, moat & competition

6. Growth aspects

7. Valuation

---

The seeing-stones (history)

Picture by Game Rant (edited by author) | Source

One of the best ways to locate great, down-to-earth founders of a high-growth, innovative technology company is if the backstory of the company’s name is creative. Not only is Palantir’s name creative, but it has possibly the most intriguing history I’ve written about on any business to date, while also being the least detailed history write-up I’ve ever wrote.

Founding

Screenshot | (Google Query: “Palantir founders”)

Palantir’s official founding was in 2003 by Peter Thiel, after the now-billionaire venture capitalist left PayPal. Thiel created the company with the goal of countering terrorism threats after 9/11. Eventually, Thiel would recruit colleagues and friends of his: Nathan Gettings, Joe Lonsdale, Stephen Cohen, and Alex Karp (who would soon take the role of CEO). Together, they worked to grow a company designed to manage and streamline large amounts of data. Peter Thiel at the time had amassed a $55 million fortune from his role as co-founder of PayPal and brought his venture capital expertise to the forefront of Palantir. (Many of these founders were wealthy and/or had wealthy connections, so there is sadly no “rags-to-riches” story today.)

All of these men, including Thiel, were motivated by the idea that government agencies were not making the best use of their available data and information. It was a simple idea they believed they could solve, and thus, Palantir was born. Palantir’s name comes from the “seeing-stones” in The Lord of the Rings trilogy, which, in the story, were a group of stones called the Palantíri that gave whoever held them the ability to see and uncover anything they desired. (Hence the name “seeing-stones.”) Palantir’s product was a piece of software that had the potential to help organizations unlock efficiencies, predict trends, and uncover insights from data using algorithms and AI that would otherwise not be known. Palantir’s correlation to this fantasy fiction ball, let’s say, was based on its ability to uncover the hidden and see what not many can see using data. Fun fact, Palantir’s logo closely resembles what a Palantíri looks like.

Now, as mentioned, Palantir’s original founding was based on the idea of helping governments manage their data more efficiently. In fact, the CIA (Central Intelligence Agency) was an early investor in Palantir via its venture capital subsidiary In-Q-Tel. Consequently, the first product released by Palantir in 2008 was one targeting government data management, under the name Gotham. This was a customizable subscription platform (both in price and product offering) helping government agencies by facilitating the tracking and analysis of intelligence data. Many of Palantir’s first clients were the FBI, CIA, and even the U.S. Department of Defense. Eventually, this product would be so beneficial to these government agencies that it helped in the success of preventing terrorist threats during the American-Middle East tensions post-9/11.

By this time, Palantir’s founders knew they could be doing something bigger. Governments are consistent, reliable, and they pay well over a long period of time, but private and public companies… now that’s where the dollars are. So, from here, more work was focused on leveraging Palantir’s success with government agencies to create what would be known as Palantir Foundry. This was a product focused on commercial clients, and it applied to a wide variety of commercial industries. Finance, healthcare, energy, manufacturing. The thought process was that companies have lots of data, and streamlining data would make them more money… and what company doesn’t want bigger profits?

This new product was a perfect fit, and officially, with Gotham and Foundry, Palantir’s core business model was now complete. By 2015, eleven years after its initial founding, Palantir reached a $1 billion valuation.

Palantir would remain focused on its two main products as the years went on (Gotham for governments; Foundry for commercial clients), and its partnerships continued to rise. Their early success in commercial markets and continued demand for Palantir’s government product gave them influence and dominance in the data analytics sector they operated. Clients included Boeing, Morgan Stanley, and Fiat Chrysler (now Stellantis). Palantir’s ability to assist organizations with their vast amounts of data was a clear working strategy, and companies resonated with it. This would quickly become a cornerstone of their value proposition.

Up until this point, Palantir had operated very secretly. As many of its current customers were high-profile government agencies, being open with its operations wasn’t a priority for the company. That is, until its IPO.

In 2020, after years of speculation, Palantir officially went public on the New York Stock Exchange through a direct listing, valuing the business at over $22 billion. Since going public, Palantir has continued to make significant moves in both the public and private sectors and continues to uphold its dominance in its own niche data analytics industry. Palantir even started expanding its reach globally, offering its products to foreign governments and even large multinational corporations. (Palantir only works with businesses and institutions that align with the company’s values. So far, they’ve rejected the Chinese and Saudi Arabian governments from using their services.)

In recent years, Palantir has focused on expanding its offerings to include artificial intelligence (AI) and machine learning capabilities (more so than they already had). As data becomes more and more valuable to businesses and organizations around the world, Palantir’s role in this revolution plays a larger and more important role. The company has continued its commitment to innovation, particularly in AI and machine learning as I explained, and continues to position itself as a leader in the ever-evolving world of data analytics. (More about the business and competitive advantage later on in this post.)

Palantir as a company begins on a secretive journey from starting as essentially a government contractor to becoming a global player in data analytics. As Palantir continues to grow, its influence with governments and companies around the world across industries is expected to increase along with it. PS: Because of Palantir’s secrecy over its history due to working with intelligence agencies, much of what I can find about its history is vague.

Business model & analysis

Palantir is a software company that specializes in a niche of big data analytics, providing tools for organizations to integrate, manage, and analyze complex data. It hosts two main products, Gotham and Foundry, that sell to government agencies and commercial companies respectively. These products allow its clients to make decisions using insights and predictive analysis (from the data).

Business model

Palantir generates revenue primarily through software subscriptions and service contracts from the government. These revenue streams support the business model, and they’re reported as two major segments on the company’s financial statements¹:

1. Government Segment (55% of revenue): Revenue generated from Palantir’s Gotham platform, which specializes in government clients (intelligence agencies, military). This platform provides software for counter-terrorism, threat detection, mission planning, and helps by leveraging these agencies’ vast amounts of data to come up with usable intelligence.

2. Commercial Segment (45% of revenue): Revenue generated from Palantir’s Foundry platform, which specializes in private and public companies (commercial clients). It serves all industries, from healthcare to manufacturing and finance, allowing companies to better understand their data through systems, helping to streamline operations and costs.

Palantir is an interesting business.

Have you ever watched a “how to make a delicious chocolate chip cookie” video on YouTube and thought to yourself over and over again that you wouldn’t be able to create such an appetizing cookie, only to try out the recipe and end up baking the best cookies of your life? Well, that’s the process of learning about Palantir’s business model. You read about the company, see posts online about how it’s going to the moon, and hear the words “data analytics,” and then you back out entirely from wanting to learn about the business. “It’s too complex,” you say. But then you actually research and learn about it, and everything isn’t so complex.

For a clear, short definition: Palantir sells data analytics software to governments and corporations, leveraging AI to help find and solve problems using an organization’s data.

For a more accessible definition, you can think of Palantir as a company that sells a sort of “puzzle software.” Palantir’s clients—governments and hospitals, for example—have access to many puzzle pieces throughout their vast amount of data. The problem is, these organizations don’t know how to put these puzzle pieces together to understand the finished result, and sometimes don’t even know if these puzzle pieces exist in the first place. For governments, this means piecing together puzzles to streamline intelligence capabilities, such as tracking the Taliban or predicting its next movement. Or for hospitals, this would involve correctly and effectively organizing data on, say, the coronavirus, so they can track and predict disease spread and plan accordingly.

Palantir’s products use artificial intelligence and complex algorithms to solve these “puzzles” on behalf of these institutions, gathering already-existing data from the client using the product to help them streamline operations or identify problems or threats. Palantir does not control or have access to this data; instead, it provides the software that allows its clients to work with their own data securely. (Continuing with our hospital and government examples): Palantir’s platform can process everything within a client’s data—spreadsheets, PDFs, emails, video and audio recordings, satellite images, military movements, doctors’ notes, patient logs, and more—collecting and compiling all of this data in a way that makes sense and helps organizations make better decisions.

But then there’s AIP.

AIP is an optional add-on for the consumers of Gotham or Foundry that integrates large language models (LLMs) into their existing software. This enhances their ability not just to piece together information with machine learning but to use the power of LLMs to understand whether a particular “puzzle piece” fits. Instead of simply compiling information and drawing conclusions, this add-on would assess whether the conclusion is plausible. If you’re unfamiliar with the recent LLM and GenAI craze, a massive distinction between basic AI and machine learning is that “basic AI” runs on procedure and algorithms, while LLMs (which I’ll call advanced AI in this case) run on intelligent understanding to figure out things for themselves, rather than being explicitly told what to do.

(As a reminder, these platforms and/or products that Palantir offers don’t just analyze data and do nothing—they provide insights that are acted upon. Palantir’s products drive real-world outcomes. It’s a very powerful product offering.)

Goverment surveillance

It’s fascinating technology, but let’s be pessimistic for a second. From this, Palantir reveals the complex, detailed access to data governments have on their citizens. Here in Canada and the United States, we have a little something called privacy law (as do all free countries), which grants us the right to, you guessed it, have privacy. Intelligence agencies work according to this law but operate with legal exceptions. CSIS or the CIA surveil their citizens in one way or another in the name of national security or to mitigate terrorism threats. That’s the exception. Government agencies like the CIA, FBI (and the corresponding intelligence and investigative agencies in every country) have access to GPS data from personal cell phones, phone logs detailing who called whom, from what time, for how long, and for what reason, as well as email logs, social media activity, transportation patterns logged from your car, any airplane you step foot on, and even details on where you travel and what you do in a foreign country—all in the name of monitoring national security.

This isn’t a conspiracy; it has been known for a while that most intelligence agencies do this. I bring this issue up because one of the main uses of Palantir Gotham is the product’s ability to cross-reference classified data for governments, across different agencies, to draw conclusions on certain threats. I used the example of the Taliban earlier, so let’s continue with that.

The way the CIA or FBI might conclude there is an incoming Taliban threat, for example, would be by compiling data from the FBI, U.S. Army, CIA, and various intelligence departments—including phone logs, emails, and all possible surveillance efforts—and then using Palantir Gotham, which processes this information and generates insights. The FBI would then come to a conclusion on whether or not there’s a threat, perhaps because the platform pointed out specific irregularities. These are the types of “puzzles” Palantir solves for its customers.

Pricing and flywheel

Palantir’s pricing model isn’t as set-in-stone as other software companies typically have. This means the pricing for businesses or organizations using Palantir is highly subjective, depending on each client. For example, an organization like the CIA requires much more data to be processed, and as a result, Palantir will need to work with a larger data set for the company. On the other hand, a business like Boeing might only need data on its engines or doors to learn and improve more quickly with new designs. Pricing is then set accordingly.

Governments tend to have much larger amounts of data that need to be processed, whereas businesses, hospitals, and commercial clients for Palantir tend to be more conservative with the data they need to handle. To make sense of this, the U.S. Army might pay Palantir $400 million in their specific agreement, based on the amount of data that needs to be processed and analyzed, while a company like Boeing might only pay $2 million. (I should mention Palantir’s pricing is typically negotiated on an annual basis. These contract figures are annual. Some contracts are paid monthly or quarterly, but most are annual, and these annual contracts are generally multi-year.)

The government, especially the U.S. government, tends to offer very lucrative contracts. These contracts are profitable, consistent, and usually require the most from Palantir’s product. Not to say that some commercial clients don’t pay over $400 million a year as well (there are likely some companies that make large payments), but Palantir’s government revenue is growing at a much higher pace than its commercial business, and these are the reasons why. (It’s not far-fetched to assume Palantir’s revenue will come from 60-70% government contracts by the end of the decade.)

Like most technology businesses, Palantir operates with a flywheel business model. Here’s what that looks like (click to enlarge):

An eye-sore graphic (I’m sorry) showcasing Palantir’s flywheel (as per the research analysis done by The J. Nicholas).

As Palantir grows and gains more clients, it gains more experience with different types of data and their cases. Palantir uses its customizability, proven track record, and working product to advertise and capture clients → having access to a larger variety of data types improves their models and algorithms → this delivers better results to clients (better, more accurate insights on the data) → which increases customer retention → and leads to more clients staying with Palantir’s service → which in turn, leads to more clients (proof of concept to potential clients) → and subsequently more clients using Palantir’s software.

As these clients become more and more accustomed to the product and the results from Palantir’s platform, they will likely want to use it more broadly across their operations, leading to higher contract sizes and longer annual contract durations. Like we saw with the U.S. Army in 2019: after years of working with Palantir on a smaller scale ($10–20 million), it announced a multi-year, $800 million contract with Palantir to build the U.S. battlefield software system. I expect these types of growing contracts to play a large role in the company’s business over the long term.

Palantir has an unusually concentrated focus on long-term partnerships with its clients to ensure high-margin contracts. This is why their software solutions are highly customizable, and why they provide clients with the flexibility to adapt the tools to their specific needs. In fact, when Palantir’s software is used for the first time by a new business or government agency client, Palantir will send its own engineers to guide and operate the software until the new client understands how to use it effectively. (This type of long-term focus on its clients reminds me of a mix of Berkshire Hathaway and Amazon. Palantir takes after the consumer-first mindset of Jeff Bezos but shares a long-term approach similar to Warren Buffett’s in some respects.)

Palantir has highly dependent clients using a highly effective product, with a long-term focus on generating high margins through high-priced government and commercial contracts that are built to compound over time based on reliability and trust. Palantir’s product is customizable and specific to each organization, making every client unique. Palantir’s prioritization of client satisfaction is what has led them to this point, and it’s an excellent foundation for a sustainable business (more below).

Although we don’t have exact access to the specific clientele using Palantir services, we do know that the total number of customers is 629 as of the latest quarter. This number continues to grow.

Management quality