UBS initiated coverage at $80 target price for $PLTR.
17 major customers have revealed important details about the success of AIP. Key points to note include:
1. Palantir is a perfect all-in-one platform with a unique front end. Palantir can run everything in one place, from data lakes to analysis and implementation, truly a perfect all-in-one platform.
2. Foundry and AIP coexist. Because Foundry and AIP are 100% integrated, you cannot run only one. Also, these two should not be considered as separate products. Without Foundry's data, you cannot run AIP. The actual data applied to AIP is literally stored in Foundry, so if you need AIP, you need Foundry.
3. The price of AIP is based on consumption "The contract execution amount of AIP probably accounts for 30-35% of Palantir's expenses. Because AIP is completely consumption-based pricing, you will be charged for storage and computing every time you make an LLM API call, and additional charges will be incurred for additional AIP features you are using. AIP is based on consumption and linked to data volume, but what you can actually consume is the absolute maximum available in Foundry. There are other variables based on the number of users, data volume, and access to AI prompts, making it very complex. However, it is likely that AIP expenses will be only a small part of Foundry expenses, around 30-40% over 2-3 years. So if you are spending 10 million dollars annually on Foundry, adding AIP will increase it to 13-14 million dollars in the third year."
4. AIP is not actually "free" "Customers may not explicitly pay for AIP. I have spoken with customers who do not explicitly pay for AIP. AIP is 'free', but ultimately it will result in spending more on Foundry. Based on experience, it is still early, but if existing Foundry customers add AIP, actual spending on AIP SKUs could increase overall Palantir spending by 10-20% and possibly another 20% increase through the pull-through effect to Foundry. Therefore, the impact is likely to be equally significant. If you do not pay explicitly for AIP, Foundry spending will only increase by 30-40%."
5. AIP triggers Foundry expansion "AIP will lead a large part of the increase in the number of customers, but it does not lead to revenue growth. This is because to actually leverage AIP, all data has to be loaded into Foundry and a robust data model needs to be built. While LLM ultimately only supports customers, the data model needs to be robust. Loading all data and use cases into Foundry will lead to significant changes from a revenue perspective. Therefore, AIP may be the trigger, but the spending is actually not on AIP. Spending comes from loading additional data sets and use cases into Foundry."
Palantir is expensive, but there is no equivalent. In terms of cost, we are spending about $4,000 to $45 million annually on Palantir, while Azure can build the same thing for less than $20 million annually, less than half. In terms of closed ecosystem, because there is no way to consolidate all users into a single data platform, we try to invest as much as possible in open source, open ecosystem. That's where Palantir really excels. If all data is stored in Palantir, its performance and functionality are unmatched by any other data platform. Having data in various places decreases its value. We consider this to be an all or nothing platform. However, Palantir is a very expensive solution, so we plan to introduce it only when we can achieve efficiency improvements. Our Palantir business will grow by 15-20% in 2022 and 2023, and by 30% in 2024. Most of this increase is due to AIP. I think you can understand that.
Palantir has no other options. The little-known secret is that there is actually no alternative to Palantir, even when combining other software products. The technology is not complex, and the backend is very simple. What makes Palantir unique is the ontology as mentioned earlier. You would need to reconstruct applications and gradually imitate what Palantir is doing fundamentally, but I think that is impossible. When migrating platforms, business users would not want to waste the work done in Palantir. Also, please note that Palantir does not communicate with any other systems, so even if you try to introduce new workloads, they won't be interconnected. To be frank, Palantir has no direct competitors. In my experience, Palantir hardly directly competes with other vendors. They usually come for a demo and start running use cases immediately. For organizations holding massive amounts of complex data, Palantir is the fastest and simplest way to leverage AI, and that is precisely its selling point.
Palantir is not just a data platform. In terms of data, Palantir is not just a data lake, but is working on operationalizing data. So, it is more than just a data platform and not a SaaS service. However, essentially, the entire technical stack is implemented, and then front-end engineers participate to develop user use cases/apps on the Foundry platform. In that sense, Palantir actually addresses the entire application technical stack.
Palantir is expensive but trustworthy. In any use case, Palantir always turns out to be the most expensive option. Organizations have large amounts of unstructured data, with too many copies of the same data, leading to the need to duplicate the data essentially each time a project is executed. What makes Palantir excellent is that they come, have one set of data, and create the 'golden' record of your data. This is basically the only reliable data source.
Palantir excels in data governance and lineage. If you only want to use Palantir, you should use AIP. Data governance and data lineage are top-notch. However, if you frequently use multiple solutions in large enterprises, you need to run data across multiple systems. Palantir's primary differentiator within AIP is related to data security, as it is literally built with very high confidentiality for military purposes. Therefore, using Palantir's data, you can be confident that you can enjoy all the benefits of LLM without risking IP and protect the information.
Palantir caters to emergencies and all use cases. Currently, we are a major customer of Foundry and are using only Foundry at the moment. The main use cases for Palantir include predictive analysis on the supply chain side, many analytical tasks related to production and manufacturing, adjusting all inventory, and ensuring the maximization of production per shift. During the pandemic, Palantir was crucial for us. It adjusted the supply chain in real-time and executed all the measures to access semiconductors for vehicles. All of these were done through Palantir. We are currently considering utilizing Palantir for driving an intelligence engine related to used cars, pricing when buying back used cars from customers, and which used cars sell the most.
They are fully committed to the boot camp. It's amazing that "I think they've done nearly 2,000 AIP boot camps, which is really important for their pipeline. They no longer need to invest months of an engineer's time. We eventually built a complete digital twin at Foundry and AIP, and the model has been expanded to 7 million unique finished products. In addition, Palantir has basically built a traversable table that connects the supply chain end-to-end. Literally, all 1.5 billion rows of tables are connected, allowing you to click on any customer, material, etc. This is amazing."
AIP saves 6-12 in all AI use cases "In our trials, the labor-intensive process was shortened from 3 weeks to 1 day. This AI app leverages Anthropic and AWS Titan models, AWS infrastructure, Palantir for data preparation, Accenture for consulting, and MuleSoft/Salesforce for API orchestration. "Time to market is also very important. By using Palantir's AIP, you may save 6 to 12 months in getting a specific AI use case to market."
Some customers fear being locked in by Palantir. "From a commercial standpoint, Palantir has been very flexible, essentially providing AIP for free and charging customers for the use of Foundry, but ultimately, customers end up being more tied to Palantir."
AIP enables low-risk AI DIY "When discussing AI use cases, many risks that are typically talked about are eliminated, and in reality, anything related to data integration, governance, and data lineage is very smoothly and DIY-like handled by AIP. There is no need for heavy implementations or calling in expensive Palantir engineers for the build."
AIP can replace 30% of analysts "Our supply chain department has about 700 business analysts, but by leveraging AIP, we could potentially replace around 30% of those employees in the next 2 to 3 years. Many of these analysts only answer very basic questions about products, so replacing them is quite easy."
AIP brings a 30-35% increase "One is customers whose spending doubles or triples by adding AIP, and the other is customers whose spending increases by 5-10% annually by adding AIP. The first type are customers who have implemented all the targeted AI use cases, causing Palantir spending to balloon. In this case, it's the IT department that needs to meet AI initiatives as quickly as possible. The second category depends heavily on business strategy and focuses on individually adding specific use cases and data sets. In these cases, customers may increase data sets from 7 to 8, or add a single use case, resulting in an actual spending increase of around 5-10%, limited to a few occurrences per year, restricting the actual increase in AIP spending to around 20-30%. Since most customers likely fall into this second category, the overall increase to Palantir would be as follows: When an average Foundry customer adds AIP, the cost will be 30-35%."
C3ai is a loser "C3 might say they are most similar to Palantir, but in my opinion, they are not even close. I've seen them once or twice, but at this point, none of our customers are using it. I have never seen them beat Palantir."...This feedback seems to tell its own story...Is Palantir worth $80?
Disclaimer: Community is offered by Moomoo Technologies Inc. and is for educational purposes only.
Read more
Comment
Sign in to post a comment
