Palantir (PLTR) has initiated its direct listing process and will trade on the NYSE. As we prepare for their “investor day” where they will present the investment case we can summarize some of the things we took away from the initial filing.

On the Product

Palantir is kind of a hybrid between a software company and a services provider. They stress that a business isn’t going to get a strategic advantage from a packaged application. Those systems may be necessary but do not serve to differentiate. Instead, Palantir has built what might be described as a new kind of “4GL” if anyone remembers that era of the 1990s. This one however is at a dramatically higher scale and scope. Their pitch is that you can keep your existing systems but need a kind of “digital operating system” that weaves it together and enables much more powerful capabilities.

Data is really the area Palantir focuses on to provide better analytics and decision making. But Palantir works toward allowing customers to take the next steps which are to implement those decisions, reflect them in the operating environment, and create feedback loops to measure results and make further changes. One popular manifestation of this idea is the “digital twin” which is where a complete virtual model of a real thing or real-world process is created. The digital twin can be used to simulate real-world conditions of all types and experiment with changes before implementing them in the real version.

An unusual aspect of their model is that they use software engineering resources deployed or “embedded” with customers to work on tailoring and implementing solutions to specific customer needs. When I worked at IBM in the 1980s this is how we won over customers back then too. Our support staff was on-site with the customer and if there was ever a serious operational issue the customer knew that we would “darken the sky with planes” to make sure business operations opened on time the next morning. It’s almost like Palantir enters long-term joint development relationships with their customers.

Palantir targets the largest, most complex, and technologically demand problems at large government and commercial organizations. In most of these cases, other solutions and custom development have been tried but failed to deliver results, often at cost of billions of dollars. With few exceptions companies and even government agencies do not have the resources required to pull off projects of this size and scope.

There are two different product platforms, Gotham, which came first in 2008, is targeted toward the intelligence community. Foundry was released in 2016 and is targeted towards commercial enterprises. Gotham is heavy on pattern matching, mission planning, surveillance, and intelligence gathering. Foundry is much more of a pure data science and analytics platform.

There are core features of the platform that include data integration, versioning, model management, orchestration, provenance, security & compliance, data modeling, collaboration, search and compute, and robust time series analysis. For those that appreciate visuals more than descriptions here is an array of product screenshots from the S1 Palantir Product Snapshots (PDF).

When examined different pieces of what Palantir provides have existing solutions like Splunk (SPLK) for monitoring and automating operations, like Elastic (ESTC) for data analysis, search and pattern matching, Talend (TLND) for data integration, Anaplan (PLAN) for cross-enterprise analysis and optimization, CrowdStrike (CRWD) for security monitoring, alerting and remediation. The list goes on. Palantir positions itself as the layer that sits on top of these technology components and integrates them (with real human engineering when required) into a solution and then maintains and updates that as client needs change.

While the cloud is typically part of any solution Palantir is geared to support large deployments of on-premise systems, even those that are air-gapped. They are capable of managing single-tenant disconnected systems in the same fashion as multi-tenant cloud instances at scale. This is a unique capability. They note that their system today handles over 41,000 upgrades per week across their 125 customers.

Palantir took the US Army to court for failing to comply with a previously obscure act passed in 1994 as the Federal Acquisition Streamline Act (FASA). Basically the point was to make federal agencies stop developing custom solutions for everything and consider more readily available and proven solutions from commercial markets. While FASA had an impact in some areas it didn’t move the needle much or at all in terms of software. In a landmark ruling in 2016 (later upheld in 2018 by the US Court of Appeals) the courts ordered that the Army must consider existing commercially available products before trying to build their own. Subsequently Palantir business with the US Amry took off as shown in the figure below. Palantir believes the same rules will help them grow their business in other parts of the US government.

On Business Model

The Palantir platform is priced for large customers. Their average annual (2019) revenue per customer is $5.6M and the average revenue for their top 20 customers is $24.9. But they are getting paid to provide mission-critical systems. Many technology vendors enjoy this level of annual revenue from large companies. Total revenues for 2019 were $742M.

They have their own very exponential version of “land and expand” in building their customer relationships. In their framework, they use Acquire, Expand, and Scale as the stages. They make a major investment in building the customer relationship in the first two and it is in Scale where revenue and margins expand. Management will be reporting “contribution margin” on these cohorts during major reporting periods.

Palantir charges very little during the first two phases but then at Scale pricing is based on “value to the customer” which is where they go to 7 or 8 figures of annual revenue from the customer and a multi-year contract. This is also where the contribution margin swings positive and generates a profitable customer. We hope to see more data on the customer journey when they host their investor day. Meanwhile, it’s clear that they are performing against this metric.


There’s a longer-term goal to take a customer solution and extend it into an industry platform. Airbus is an example of a customer that ultimately led to the introduction of the Skywise “aviation platform” that many different organizations now use if they are flying Airbus planes. Palantir hopes to leverage this into relationships with those other organizations over time. They also want to repeat the Skywise approach in other industries like healthcare and financial services.

On Management

Palantir works hard to incorporate culture and principles into their work. In particular, they insist on working through security, privacy, and transparency at the start of the design process. They also believe that humans and human decision making need to remain part of any technology-driven solution or process.

The company also makes a major point about “choosing sides” which means they restrict who they will and will not do business with. The US and their allies are fair game, communist governments and regimes are not. The company also believes that it is within their span of responsibility to safeguard privacy and civil liberties.

Peter Thiel is a co-founder and major shareholder. Alex Karp is well-known as the CEO. What’s surprising is the relative youth of the rest of the senior team. The other five executive officers are tightly clustered in age from 36 to 39. They are just reaching their time of peak capacity.

Alexandra Schiff is on the board and this is the first time we see an author and journalist in such a position at a major technology company. I welcome the refreshing recognition of the liberal arts with our organizations that are too driven by technological and financial considerations alone.

On Valuation

It’s early days on valuation but their last round was said to value the company at $20B or about 20x sales. The other consideration is that they put their TAM at $119B ($63B for government and $56B for commercial.)

Sales clearly have lots of room to grow over time and as more customers reach “scale” the company can generate profits.

As technology valuations stretch ever higher one might ask a question about what happens when the market value = TAM? Or even if it just hits 50% of TAM. That’s discounting a lot of future goodness.

Once the full investor presentation is available we will follow up with Parsing Palantir – Part II


Leave a Reply