Text and Image | Tangjie

If we were to list the beneficiaries of the generative AI revolution, for a long time, the list would only include upstream and downstream enterprises, such as nuclear power companies that accelerate AI computing capabilities, chip manufacturers, and so on. However, it has been difficult to find reliable targets for AI large models in application scenarios; even if there are any, it is hard to convince people of their capabilities.

But now, with Palantir continuously releasing better-than-expected results, market expectations for AI applications are gradually rising. As one of the earliest AI enterprises led and invested by Silicon Valley tycoon Peter Thiel, Palantir initially gained fame through big data analysis and decision-making in the military field. In 2011, it became well-known after using big data to help the U.S. military successfully locate and neutralize Osama bin Laden.

Nowadays, the company's capabilities are no longer limited to military and big data analysis. The emergence of the AIP platform allows them to effectively integrate the capabilities brought by large models, offering a richer and more diverse range of products.

In the third quarter of this year, the company's revenue was $726 million, a year-on-year increase of 30%, exceeding market expectations; net profit was $144 million, a record high; Non-GAAP earnings per share were $0.10, higher than market expectations. As a result, the company's share price surged by more than 14% from after-hours trading to the overnight session, demonstrating the market's enthusiasm for these results.

In terms of regional revenue distribution, the company is primarily focused on the U.S. market, with revenue of approximately $500 million, accounting for over 68%. Regarding income sources, 64% of the company's revenue in the U.S. market comes from the government ($320 million), with the remainder coming from commercial companies ($180 million). The core driver of growth originates from the U.S., where revenue surged by 44% in the third quarter.

Among the revenue growth in the U.S., commercial revenue increased by 54%, and government revenue increased by 40%. Revenue from commercial sources has become a driving force for growth. This logic is not difficult to understand. Generally speaking, technical capabilities honed in real battlefields often have higher safety and accuracy than those in daily commercial applications, as the consequences of errors can be fatal rather than merely economic losses.

In the U.S., the majority of defense and defense-related businesses are provided by private enterprises. When these enterprises choose partners, their needs tend to be more specialized, making it less likely for them to turn to non-industry players like OpenAI. Therefore, a considerable portion of the so-called "commercial revenue" should also come from defense industry companies.

Similarly, it is reasonable to infer that a significant portion of the company's overseas revenue, which accounts for approximately 32% of its total, should also come from G-end and defense enterprises. Based on this, we can provide a basic definition of the company: an AI large model company centered on providing military AI capabilities, with government defense cooperation as its competitive barrier, and leveraging a good reputation to attract defense companies and cooperate with the governments of U.S. allies.

In the entire U.S. stock market, Palantir is the only company of its kind. From the perspective of the company's over 20-year history since its establishment, it possesses strong scarcity and technological barriers.

With a solid foundation and the current unstoppable trend of AI, the company has provided Q4 performance guidance far exceeding market expectations. The full-year revenue forecast has been revised upwards to $2.805 billion to $2.809 billion, and the adjusted operating profit for the year is expected to be $1.054 billion to $1.058 billion. Based on the third-quarter level, this represents a significant increase in quarter-on-quarter growth.

As for achieving these goals, it is currently the least of the company's concerns. As a private enterprise providing defense technology, choosing Palantir provides a foundation of trust from government suppliers. Coupled with the AIP platform launched by the latter, it enables rapid application and deployment of large models in enterprises. It can also help train AI technicians for clients, allowing them to expand their application scope rapidly.

As the AI market continues to expand, it is not difficult for the company to achieve double-digit growth each quarter, and the likelihood of continuously exceeding market expectations is also high. However, in our view, the company's current valuation multiple is relatively high. The recommended approach is to add a small position and continue observing before making significant purchases during significant market downturns.

As is well-known, during earnings seasons, compared to fully disclosed financial reports, the information revealed during earnings calls is more critical for assessing a company's value. For example, Palantir may not specify the development of AIP in its financial reports but will clearly articulate the expansion logic of AIP during earnings calls:

We serve the United States and its allies, and using our products will facilitate direct cooperation between enterprises and the U.S. government. We have started providing these products to various defense technology startups and begun cooperating with more mature large integrators like L3.

Understanding this information clarifies one thing: in the enormous defense business, Palantir is the only large model enterprise related to AI technology and maintaining long-term cooperation with the U.S. government. With such a market position, it is relatively easy for the company to maintain its current high growth rate.

Finally, regarding AI, the evolution of its applications is actually much faster than we all imagined. During the earnings season, we discovered a tool application called 'Simple Transcription,' an audio transcription tool developed by Baidu Netdisk, a leading domestic large model enterprise. It helped us transcribe company earnings call documents in a short period, improving our coverage efficiency during earnings seasons.

Disclaimer: This article is for learning and exchange purposes only and does not constitute investment advice.

We welcome likes, views, and shares. Your support is our motivation for updates.