


Palantir(minutes): hopes to become the cornerstone company of this AI revolution
The following is the summary of the 4Q24 earnings conference call for $Palantir Tech(PLTR.US). For a review of the financial report, please refer to “Palantir: Beating the Shorts, the Unattainable AI Faith”
1. Overview of Key Performance Indicators
1. Revenue
(1) Operating revenue reached $828 million, a year-on-year increase of 36% and a quarter-on-quarter increase of 14%. Annual revenue was $2.87 billion, a year-on-year increase of 29%.
(2) The U.S. business performed strongly, with fourth-quarter revenue increasing by 52% year-on-year and 12% quarter-on-quarter; annual revenue increased by 38%, reaching $1.9 billion.
(3) The U.S. commercial business grew rapidly, with fourth-quarter revenue increasing by 64% year-on-year and 20% quarter-on-quarter; annual revenue increased by 54%, reaching $702 million.
(4) International commercial business saw fourth-quarter revenue increase by 3% year-on-year and 15% quarter-on-quarter; annual revenue increased by 9%, reaching $594 million.
(6) Government business saw fourth-quarter revenue increase by 40% year-on-year and 11% quarter-on-quarter; annual revenue increased by 28%, reaching $1.57 billion.
2. Profit
(1) Fourth-quarter adjusted operating profit was $373 million, with an adjusted operating profit margin of 45%, the highest level in the company's history.
(2) Annual adjusted operating profit was $1.13 billion, with an adjusted operating profit margin of 39%.
(3) Fourth-quarter adjusted earnings per share (EPS) was $0.14, with annual adjusted EPS of $0.41.
(4) Fourth-quarter free cash flow was $517 million, with annual free cash flow of $1.25 billion.
3. Customer Growth
At the end of the fourth quarter, the number of customers reached 711, a year-on-year increase of 43% and a quarter-on-quarter increase of 13%.
The number of U.S. commercial customers reached 382, a year-on-year increase of 73% and a quarter-on-quarter increase of 19%.
The net dollar retention rate for the fourth quarter was 120%, an increase of 200 basis points from the previous quarter. This growth was mainly due to the expansion of existing customers and the acquisition of new customers in the fourth quarter of last year, reflecting the acceleration effect brought by the AI revolution.
4. TCV Performance (Total Contract Value)
In the fourth quarter, we achieved a record TCV booking in the commercial business, reaching $995 million, a year-on-year increase of 42% and a quarter-on-quarter increase of 63%The U.S. commercial business achieved record TCV bookings in the fourth quarter, reaching $803 million, a year-on-year increase of 134% and a quarter-on-quarter increase of 170%. This figure surpasses our previous highest quarterly U.S. commercial TCV bookings by nearly $400 million, highlighting the significant growth in demand for AI production use cases.
The total remaining deal value of the U.S. commercial business increased by 99% year-on-year and 47% quarter-on-quarter.
The TCV booking performance of the commercial business for the full year of 2024 is strong, with the robust performance in the fourth quarter driving annual growth.
The annual TCV bookings for the U.S. commercial business showed significant growth, reflecting the high demand for AI solutions in the market.
5. Customer Growth and Retention:
At the end of the fourth quarter, our customer count reached 711, a year-on-year increase of 43% and a quarter-on-quarter increase of 13%.
The number of U.S. commercial customers reached 382, a year-on-year increase of 73% and a quarter-on-quarter increase of 19%.
David A. Glazer (Chief Financial Officer and Financial Executive):
II. Management Report
1. AI Revolution and AIP (AI Production)
The AI revolution is transforming industries and redefining how organizations operate. AI is key to driving innovation and efficiency, and businesses need to embrace AI or risk falling behind.
Palantir focuses on transforming AI models into high-value finished services, achieving "quantitative excellence" through the AIP platform.
In the fourth quarter, we achieved a 64% year-on-year growth in the U.S. commercial business, significantly driven by the AIP platform.
For the past two years, we have emphasized that although large language models (LLMs) are continuously improving, both open-source and closed-source models are converging in performance, and inference costs are significantly decreasing. With the release of DeepSeek-R1, this view has become consensus. The market's attention on AI is primarily focused on model supply, while we focus on the demand side of AI.
From the beginning, AIP was built to address this reality. We do not view LLMs as chat tools but as AI labor. To capture the productive value of AI labor, an intermediary layer is needed to enable AI to interact with enterprises practically, such as how to allocate inventory and how to handle claims. This intermediary layer is the ontology, which is also the secret to our rapid development.
We believe the normative value of AI lies in enterprise autonomy, akin to "autonomous driving." Users shift from executing workflows to supervising AI labor, teaching it how to handle edge cases and reducing processing time.
For example, we collaborated with a large multinational bank to reduce the processing time of core backend processes from 5 days to 3 minutes. This is not just about labor savings; more importantly, it eliminates historical constraints in the middle and back office, enabling the bank to create entirely new differentiated financial products.
In addition, we are working with a top engineering and construction company to leverage AI labor to automatically identify risks in tens of thousands of pages of technical documents, reducing what would have taken months of manual review to just a few minutesWe are collaborating with a large power system company to transform the automatic understanding of technical drawings into quotes and orders.
We are working with an automotive supplier to analyze CAD files to verify engineering and manufacturability checks. This process, which originally required 100 hours of human engineering time, has now been automated, with exceptions submitted for human review.
2. Examples of Corporate Client Contracts
(1) A large American pharmacy became our client at the beginning of 2024 and signed a $67 million contract after a pilot program to optimize the prescription drug distribution and patient communication processes.
(2) A U.S. telecommunications company has been our client for about two years and recently signed a $40 million contract extension to help manage and accelerate the retirement of old network technologies, resulting in significant cost savings.
(3) A global leading insurance company shortened its underwriting process from two weeks to three hours through the AIP platform and signed a nearly $11 million contract extension in the fourth quarter.
(4) Panasonic Energy North America improved battery production efficiency and quality control through the AIP platform, reducing machine downtime, enhancing production efficiency, and accelerating the onboarding of new technicians.
(5) Warp Speed is a modern operating system built by Palantir for American manufacturing, integrating engineering, testing, production, quality, and operations, enabling manufacturers to produce better and deliver faster.
The first batch of Warp Speed clients includes Anduril and L3Harris, with positive customer feedback and rapid business expansion.
3. Government Business
Our U.S. government business revenue grew by 45% year-over-year in the fourth quarter and 30% for the full year.
We signed a long-term cooperation agreement with the U.S. Army and expanded our contract with the U.S. Special Operations Command, deploying Mission Manager to U.S. Special Operations Forces units for the first time.
Our deep investment in Joint All-Domain Command and Control (JADC2) continues to yield results. The Maven project introduced new AI capabilities into the kill chain, particularly integrating contested logistics into Maven's AI-driven kill chain during Q4 exercises.
The adoption rate of the Maven project within the military continues to increase, especially in the Army, Air Force, and Space Force, as well as in the Space Command, Southern Command, Africa Command, and Strategic Command.
The Maven project has also expanded to allies and partners through capabilities provided by the U.S. (Maven REL), especially in situations requiring real-time collaborative crisis response.
4. Competition
Everything we do at Palantir, including our culture, our products, and how we bring those products to market, is aimed at making America more competitive and making our adversaries more fearful of confronting American interests and the American people. We take pride in our ethical stance and are confident in America's futureWe believe that Palantir is at the beginning stage of the AI revolution, and we plan to become a cornerstone company driving this revolution in the next 3 to 5 years.
5. 2025 Outlook
In the first quarter of 2025, we expect revenue to be between $858 million and $862 million, with adjusted operating profit between $354 million and $358 million.
For the full year of 2025, we expect revenue to be between $3.741 billion and $3.757 billion, representing a year-on-year growth of 31%. Revenue from our U.S. commercial business is expected to exceed $1.079 billion, with a year-on-year growth of at least 54%. This growth expectation reflects our continued confidence in the AI solutions market and our positive response to customer demand.
Adjusted operating profit is expected to be between $1.551 billion and $1.567 billion, and adjusted free cash flow is expected to be between $1.5 billion and $1.7 billion. We expect to achieve GAAP operating profit and net profit in every quarter.
III. Analyst Q&A
Q1: What is unique about the ontology? How does it set you apart from the competition? Why can't other companies replicate something similar?
A: If I build software for a company to solve a problem, then I don't need an ontology. I just need to find the shortest path from A to B, which doesn't involve an ontology. But if I have the ambition to build software that spans governments and 50 different industries, then I must build an ontology. This is definitely the longest path from A to B, but it provides superpowers to the customers.
More importantly, AI makes the ontology essential. The ontology ultimately becomes the Intermediate Representation (IR), allowing your business to open up to AI in a controlled and secure manner, providing the observability you need so that you can trust AI, and more importantly, your regulators can trust your transition to an "autonomous driving company."
It also allows you to manage them, placing large language models in a more precise location through practical taxonomies so that they can be managed at scale. You can think of it as what we do on the backend of Foundry; we need to create pipelines that essentially de-segment the data. To make AI truly work in the enterprise, you need to re-segment and reconnect large language models so that they are thin enough to provide sufficiently precise information. The precision of this connection is enough to deliver actual results in an enterprise environment, meaning you can rely on large language models to accomplish tasks that would normally require 500 experts to process text with just 3 people.
Building this platform is difficult, so Palantir is continuously reaping the rewards, with the 40% rule reaching 81%, and there is hardly a truly professional sales team. The difficulty of building this product needs to be understood from each product we have built, including security models, how to interact with the actual data of enterprises, and how to implement branches. Then you need to interact with it and build content that can work with large language models, not what you think they are or what you wish they were, but what they actually are. This process may just be beginning for many companiesThis is a very long and difficult process to achieve in a productized form. Many people do not understand this because technology companies do not actually need this, as they only have one company, or they have something, but it is not a product. Expanding this from one company to a company that can be installed anywhere requires a very specialized, product-focused culture that has a deep understanding of the business.
Last but not least, in order to train large language models to provide more accurate information, you need to authorize access to the underlying secure and clean data of the business. No other company in the world has this access like Palantir, which is why we are in a unique position. Fortunately for the world, we are very optimistic about the United States, and we are primarily doing this here because it is very powerful.
Q2: How will the new Chinese AI model DeepSeek affect Palantir? How can Palantir mitigate its impact?
A: I think one obvious lesson from DeepSeek-R1 is what we have been saying for the past two years, that models are becoming commoditized. Large models, whether closed or open source, are getting better and better, but they are also becoming more similar. The price of inference is dropping sharply.
But I think a deeper lesson is that we are at war with China, and we are in an AI arms race. Over a year ago, I had a disagreement with Sam Altman at the Senate AI summit, where I argued that we are in an AI arms race, while there is a viewpoint that we can cooperate in this area, and the other side believes that intellectual property will be stolen in cooperation.
But this war began long ago in the form of an economic war, with China joining the World Trade Organization, which is the largest intellectual property theft in history and the largest transfer of wealth in history. This is an opium war. The number one cause of death for people aged 18 to 45 is fentanyl, which comes from China. This is a diplomatic war, and the Belt and Road Initiative essentially represents that other countries are in a state of "indentured servitude" to China.
China knows they are at war. The question is, as a peace-loving nation, we are somewhat ambiguous. But one thing I do hope we all realize is that the engineering in R1 is sophisticated, and the optimizations they have made are indeed impressive. I think you cannot dismiss it with a simple explanation that the Chinese are just copying, and we are the only innovators.
We must respect our opponents and realize that we are in competition. But they have absolutely stolen a lot through model distillation, and perhaps they have stolen more. Then you can look at the growth in GPU sales in Singapore, which is just a small island nation. So I am quite sure that there are operations going on there that violate sanctions. We must realize that the AI race is winner-takes-all, and this will be a national effort, far beyond the Department of Defense, for us to win as a country. M-Day was just yesterday, and the time for mobilization has comeQ3: Are you surprised by how fast the sales cycle is? Did the team find it unexpected from the first contact with customers to the deployment of AI programs?
A: I would say I’m not surprised because I can feel the feedback from customers on-site and the impact of AI. Shyam shared some examples, such as reducing the back-end processes of a large bank from 5 days to less than 3 minutes, and shortening the automation processes of an automotive supplier from originally needing 100 hours to just manual checks. Seeing the reactions of these customers after implementation, seeing their desire to launch quickly, scale rapidly, and stay ahead of competitors, that energy is palpable.
What surprises me greatly is the stark contrast in reactions to this revolution between customers and countries. In the United States, companies are responding to this revolution in a very pragmatic way. We have many experts saying you should buy a large language model and install it in your business. Customers say, “Okay, we’ll give it a try, but we also want to try Palantir.” In the U.S., there isn’t an ideology that defines how people do things. It’s about output and how quickly results appear.
In Europe, our business accounts for 13%, with a growth rate of only 4% last year, while the growth rate of U.S. commercial business exceeds ten times that. In the U.S., people are really focused on results. In Europe, it feels like people are reading a technical installation manual from the 1950s and just following it. What’s truly shocking is that you see some customers showing great enthusiasm for our solutions, while in a culturally similar region, some just say, “Well, that sounds good, but maybe it will interfere with my vacation.” This difference is really astonishing.
I think when investors look at Palantir, they find it hard to judge because people are moving in different directions. But in the U.S., the overall reaction is: “I expect this to be completed in X time, and you not only completed it but did it 1/10, 1/20, 1/50 faster than I expected.” This rapid delivery of results is something we have experienced in the past with Foundry and even some military-sensitive projects.
In the past, we might complete something in 6 months that would originally take 10 years, and these results are so powerful that people cannot ignore them. Now, due to the combination of time and power, these results are even harder to hide. In the past, we might do a pilot project and then expand it to the entire enterprise.
However, because people don’t always appreciate my charm, we didn’t hold dinners or cater to their preferences, so these projects often get stuck in a corner of the enterprise. Even though everyone knows this is the most important thing for the business, no one wants to admit it. And now, these results have become so powerful that they cannot be ignored. People might say, “I don’t particularly like them; they don’t hold dinners and don’t wear suits. Shyam is hard for me to understand. But their results are tremendous.” It is these results that drove our 54% growth last year and the expected growth this yearIt is these achievements that have driven our performance beyond expectations. This is also why we can truly maintain our company culture. By the way, this is also the reason we can achieve an 81% adherence to the 40 principles, which is a reflection of our company and because we can stick to our culture, our Palantir ontology. We do not need to add 500 employees every quarter; we can truly drive these projects because when people really see these results, they are enough to break through the institutional barriers and political interference that we could not overcome in the past.
Q4: My first question is about disruption. Palantir has thrived on disruption from the very beginning, whether it’s 9/11, the coronavirus, supply chains, or LLMs. Now, everyone is very concerned about the changes brought by DOGE (Department of Government Efficiency). I wonder if you think the focus on efficiency improvements, software modernization, and digitization could present a great opportunity for Palantir?
My second question is for Dave, but Shyam can also answer. When you see this demand, where will you invest to ensure you can truly execute and seize this opportunity? Is it in talent, sales, engineers, or partnerships?
A: Regarding the first question, I think no one really knows. We can feel the tremendous fear from traditional system integrators and vendors. We are optimistic about DOGE. I think if you look at my testimony in the Senate Armed Services Committee even last week, Palantir's real competitor is the lack of accountability in government. Some software projects cost a fortune but fail to deliver real results. They are the sacred cows protected by the bureaucrats behind the government. You can think of D6A, which cost $26 billion but does not work properly.
Soldiers in the field prefer Palantir because it is effective and it just costs millions of dollars. I believe DOGE will bring elitism and transparency to the government, which is exactly what our commercial business has. The commercial market is elitist and transparent, and you can see the results we achieve in this environment. That’s why we are optimistic about this. I think the work we do in government is deeply operational and highly valuable. We are very excited about excellent engineers being able to understand and see these changes deeply.
We love disruption; anything that benefits America will greatly benefit Americans and Palantir as well. I think you are absolutely right. Disruption will ultimately expose what does not work. There will be ups and downs, but this is destined to be a revolution. Some people may lose their positions, just as we expect to see truly unexpected things and achieve victories. Just as we expect to see truly unexpected things and achieve victories. We plan to do this, and we plan to report expected things and achieve victories. We plan to do this, and we are very optimistic about the American environmentIn terms of investment, the key lies in excellent engineers. You can see this from how we set up our communication plan. Of course, this means an increase in the number of employees, but what truly matters is having outstanding engineers, not just the quantity. You will continue to see this in the first quarter and beyond.
This time, it’s the first time clients genuinely want to collaborate with us. In the past, collaboration meetings were a complete waste of time, filled with empty rhetoric, mainly so clients could fill out a report stating they had met with us, like a nerdy high school date, “I met him, perfect. But I won’t go out with him.” That was our past experience with partners. Now, even in the past few weeks, I have had real collaborative discussions because those with vertical expertise need to deliver results. They are under significant pressure, and people know how good our products are.
What remains crucial is that we are truly investing in our unique culture. I visit the company every week and have been doing so for 20 years. I have never seen a company or culture like ours. Every Palantir employee is unique. Everyone is doing something distinctive. We need to continue attracting and retaining those who are different, who think differently, and keep them focused on the world’s most important missions, building products in the way we know how. That is our main investment story. It’s not just about money; it’s about how we beat margins. It’s about talent and truly building what you should build, not what you are asked to build, and pushing them to the forefront.
We are doing this. I believe you enjoy this process as much as I do. Let’s not discuss with analysts the burden of being right, the burden of investing in ontology, the burden of truly looking at the math, the burden of explaining what an enterprise software company is, or the burden of explaining to your friends that you are really happy. Perhaps we should stop talking about these things. I am thrilled to embark on this journey with you; you are our true partners.
Every Palantir employee, we are achieving great success, and others are listening. We have dedicated the company to serving the West, especially in the United States, and we are very proud of the role we play, especially in areas we cannot discuss. We love our success in the U.S., and globally, we have also succeeded in the UK and many other places. Palantir is committed to disrupting and making the institutions we collaborate with the best in the world, and when necessary, scaring off enemies and occasionally eliminating them. We hope you support this and enjoy being our partners; we are very focused on what we are doing.
Risk disclosure and statement of this article: Dolphin Investment Research Disclaimer and General Disclosure