SEC filings and investor presentations have pulled back the curtain on Palantir Technologies' (PLTR) opaque operations, and we are intrigued by the company’s prospects as an emerging leader in data integration, analytics, and artificial intelligence. We anticipate the company will come public on 30 September via a direct listing; we value it at about US$28.2 billion ($40 billion) in equity value, or about US$13 per share based on an estimate of 2.17 billion shares.

We believe Palantir has a narrow economic moat and a positive moat trend based on customer switching costs, intangible assets, and a network effect. In our view, Palantir’s government and enterprise software has become mission-critical and deeply engrained with clients, the company has unique customer and market knowledge to pair with its AI acumen, and its software platforms become stronger as more users, institutions, and industries rely on its solutions. We believe these sources will strengthen, and despite a lack of profitability today, we anticipate that Palantir will generate robust operating leverage and excess returns in the long run.

Our estimated 34 per cent five-year revenue compound annual growth rate is driven by robust growth in government and commercial sectors. Providing a holistic view of legacy data from disparate locations and various types, integrating the influx of new data, and then using AI to extract actionable insights can give organisations an advantage over peers. However, for firms and governments attacking the challenge in-house can be unsurmountable. From assisting top-secret intelligence operations to extracting the optimal amount of oil out of wells to predictive maintenance on airplanes, Palantir is a disruptive force by selling commercial software platforms versus highly customized, consulting-led efforts.

We view Palantir as a land-and-expand story: After converting trial customers to its software platform, its software subscriptions proliferate in organisations and the incremental revenue outpaces associated investments required. With improvements in up-front installation costs and the cadence of subsequent software updates, we expect substantial operating margin improvement to the mid- to high 20s on an adjusted basis by 2025, up from negative 45 per cent in 2019.

Palantir aims to solve customers’ problems with Big Data

Palantir’s products are aimed at first enabling institutions to integrate all their legacy and constant influx of new data, regardless of format, device, or location. The software provides a holistic view of data for testing, insight, pattern analysis, and what-if scenarios and allows technical and nontechnical users to all collaborate within the same platform.

Next, its data operating system provides customers with the ability to extract insights from seemingly disparate data patterns or connections with AI. Palantir’s software platforms can be deployed in any environment and buck the trend of requiring highly customised solutions. In addition to the functionality its software platforms are providing, Palantir aims to solve the historically terrible rate of success in data integration and insight-producing projects, while also not requiring customers to displace legacy investments. Its software platforms enable self-sufficient customers whereby Palantir is providing software updates, but insights can be gleaned internally.

The company sells two software platforms: Gotham, mostly targeted at government operations, and Foundry, aimed at commercial markets. Launched in 2008, Gotham’s provenance was assisting US counterterrorism operations to find insight from their data, and alleviating the data organisational issues and dangers of soldiers in Afghanistan and Iraq mapping insurgent networks and roadside bomb makers manually.

Over time, Palantir expanded into the commercial segment. While the stakes may be different, the challenges of integrating new and legacy data sets and gleaning insight from copious amounts of data are ubiquitous. Foundry was developed to make the integration of data sets routine in addition to gaining insights through AI. These platforms are sold across their customer segments as well, with financial institutions using Gotham for fraud investigation and Foundry being used across government divisions.

Under the hood, Palantir’s Apollo platform is software that controls and updates its offerings, also enabling Gotham and Foundry to be deployed in clouds, on-premises, classified networks, edge networks, and devices. Palantir credits innovation in the Apollo platform as its way to achieve margins resembling those of software as a service instead of operating like a consulting business, all hidden to the end users of Gotham and Foundry.

Strategy has evolved

Starting as a company specialising in government intelligence software in 2003, Palantir’s business model more likely resembled a consulting firm employed for defence contracting work. This model can require a lot of manual labour, working to build customised solutions to solve a specific problem at a point in time. Issues arise when attempting to scale that particular solution to other related challenges, whether across particular institutions or market segments.

With its foray into the commercial markets, Palantir was also initially operating a consulting model upon releasing Foundry in 2016. The following year, the company migrated commercial customers to a software platform instead of bespoke applications. While the firm may have lost customers during this process, we believe having a commercialised platform is the correct strategic play in the long run. Palantir has evolved toward a SaaS company, with all its customers using its software platforms.

Palantir’s development team rotates into the field to understand customer- and industry-specific challenges firsthand, and then updates its platforms to meet the tasks. The company retains the intellectual property gained through these improvements. It typically provides its solutions at no cost for a trial period. Importantly, Palantir is coming to companies with a solution from the get-go, instead of a plan of integrating data before tackling the analytics implementation. These loss-leading efforts turn into customers as Palantir supports prospects with engineers for installation and software training. Palantir’s engineers return from the field as customers become self-sufficient with the software.

Palantir’s revenue streams are more globally and sector diversified than most investors may suspect, in our view. The company sells its platforms to the US and allies on the government side and into a variety of commercial industries, in total 36 industries across 150 countries. While we expect US government customers to remain a cornerstone of Palantir’s business, 53 per cent of revenue was commercial-based and 60 per cent of revenue was derived outside the US at the end of 2019.

The impact of these strategic changes is accelerating revenue streams alongside an improved operating profile. Margin expansion comes from lower deployment costs, proliferation of its platforms throughout existing client organisations, and properly pricing its software platform to maximize the value that customers receive.

Some notable risks

We assign Palantir a high fair value uncertainty rating.

The integration of disparate data sources and uncovering actionable insights with the assistance of AI is challenging and may not be a widespread problem that customers require to be solved. While Palantir may be able to execute on this task, a big enough market may not exist, pushing the company into a niche.

Palantir’s top 20 customers represented 67 per cent of revenue in 2019. As of its IPO filing, Palantir had a small customer base of 125 institutions, so losing any key government or commercial customers, which have flexibility not to expand contracts with the firm, could be overly damaging. Palantir has also stated its intention to not enter contracts with organisations that it considers inconsistent with its mission to support Western liberal democracy and its strategic allies, including the Chinese market. This limits overall market size and could increase cyberattacks, and working with military operations could cause potential investors to avoid the name.

A differentiated selling point is the commercial aspect of Palantir’s software platforms; however, customers may choose to keep their historical method of employing consultants and system integrators to assist in-house developments for data projects.

Alternatively, large established firms may attempt to copy its commercial approach and limit any premium Palantir can demand. Without strong revenue growth and an improving operating profile, Palantir may not achieve profitability. The company went through many rounds of funding before announcing its direct listing, and investors may not be willing to support a long-term, arduous journey to profitability and possible shareholder returns.

Morningstar's Global Best Ideas list is out now. Morningstar Premium subscribers can view the list here.