Palantir: The Most Important Company in the West
Welcome back! Today, I bring you my deep dive on PLTR 0.00 , a company that I believe has the potential to own the cloud computing space (skip to section 3.0 for that). I hope that you enjoy it as much as I have enjoyed researching and writing about it.
Below is the format for this week´s write up. Feel free to skip to whatever section may be of your interest.
1.0 Introduction to Palantir´s Potential
2.0 Investment Opportunity
3.0 Palantir´s Offerings are the Future of Cloud Computing
3.1 Palantir Will Own the Carbon Market
4.0 (Non) Competition and Culture
5.0 Karp´s Hair and Palantir´s Organizational Structure
6.0 Key Performance Indicators: Product Distribution and Stickiness
7.0 Financials, Reliance on Governments and Customer Concentration
8.0 Thoughts on ESG
Disclaimer: the information contained in this write up is not intended to serve as financial advice. It is just my opinion and remember, there is no substitute to doing your own research. Also note that my minimum investment horizon is 5 years.
Also, if you would like to stay even more up to date with my investing related thoughts, feel free to follow me on Twitter:
1.0 Introduction to Palantir´s Potential
Palantir is radically advancing how organizations function, by enabling the adoption of AI and Blockchain technologies across the board. This will revolutionize how we create wealth.
Palantir is formidable in the sense that it taps right into humanity’s core wealth creation mechanism and meaningfully amplifies it. This is likely to make it a multi trillion dollar company going forward. To explain why, allow me to take you on a brief tour through some of my mental models.
Right since the beginning of time, all humans have done to create material wealth is figure out new ways to configure atoms / electrons to yield some kind of benefit. In other words, humanity creates wealth by acquiring insights that then translate into better decisions. We are all just playing an Information Game.
This is a strange abstraction, but very easy to understand by casting your mind back in history. Things like semiconductors, electricity, the internal combustion engine, the telephone and a long list of inventions that have upgraded our civilization are all examples of figuring out new ways to assemble the building blocks of the universe, to produce wealth.
In today´s economy, most of the wealth is produced through private enterprise. Companies are in essence optimization functions, that aim to maximize the useful output and minimize the inputs, which are resources and time. They are in effect compressed representations of the Information Game.
In general, humanity has been playing this game in beginner mode. We run companies based on very subjective observations. We do not understand 99.9% of the information flows within and around a company.
Along comes Palantir, which enables companies (and organizations of any kind) to create digital twins. Digital twins are exhaustive digital representations of organizations. They unleash two very powerful technologies into the enterprise, which are AI and Blockchain, just because you can only run either of them in the digital space.
This lays the foundation for a profound shift in terms of how we run companies:
With digital twins, organizations become data churning machines. Every little action inside an organization turns into additional data being generated.
This data can then be processed by machine learning (AI) models, to extract useful insights at a far higher pace and lower cost than a human ever could do.
In essence, this commoditizes predictions, which enables organizations to run simulations first and then execute, versus execute first and then see what happens.
Further, as the machine learning models get more accurate, much operational complexity can be delegated to smart contracts (blockchain), which can be trusted to execute as programmed.
This will all radically decrease costs relative to revenue in whatever industry it touches.
You can visualize a company that gets increasingly smarter as it operates, in which humans can almost forget about all the boring repetitive tasks. This is by no means science fiction anymore, thanks to Palantir´s offerings, which I will review in depth in section 3.
The above creates a whole new kind of organization, that is simply much better at unlocking insights and hence, far more efficient at producing wealth. Palantir enables this shift via an increasingly productized software suite, ideally positioning it to deliver and capture much of the value.
As industry evolves in this direction, we can expect the company to be responsible for the OPEX optimization of a growing number of companies across the world. In this sense, Palantir is likely to become a key asset for enterprise in general and in turn, this will make the company very valuable.
2.0 Investment Opportunity
Palantir is almost 70% off its ATHs and seems to be trading at a reasonable price, given its prospects.
Before I review the company in depth and going back to the market, technology / growth stocks have been selling off meaningfully since what seems like forever now, with Palantir trading just above its IPO price. As I am writing this, the company trades at PS ratio of 14.8, which means the market is not totally ignorant of the promise it holds.
Palantir´s future prospects are so phenomenal, however, that it may be a worthy portfolio addition today. On other hand, it is definitely worth having in your watch list, should we see it trade much lower due to some potential nuclear scares in the near future.
3.0 Palantir´s Offerings are the Future of Cloud Computing
Palantir has 3 core offerings, which it licenses to clients in exchange for money: Foundry, Gotham and Apollo. These are low key changing the way computation is sold in the market place, which may turn Palantir into a gatekeeper for cloud providers.
Foundry is the operating system that enables companies to create their digital twins. Gotham is the same, but for governments. Lastly, Apollo enables Palantir´s clients to run Foundry / Gotham on any environment they may choose. It is that “simple”.
As I explained in the first section, these offerings pave the way for Industry´s next evolutionary step. Today, industry consists mostly of isolated machines and workflows, whilst Palantir´s offerings are advancing industry towards a state of interconnection and automation.
The end game of this is far more meaningful than you can imagine. As it moves in this direction, Palantir is bringing to market “useful compute”, which is going to fundamentally change how computation is sold in the market place. Allow me to walk you through a very simple example, that you will never forget:
Consider a scenario in which we have 1 autonomous company (no humans) that produces wooden chairs, which operates 2 factories in proximity with 1 machine each.
In essence, the company has to maximize the number of quality chairs that it outputs and minimize the time the machines are running for and the cost of raw materials (wood).
In traditional industry, the machines and factories are not aware of each other. For instance, the machine in factory 1 may not know that the machine in factory 2 broke down. As such, raw material inventory would build up in factory 2 and factory 1 would just keep buying wood as normal.
This would be a waste of money, because factory 1 could just ask factory 2 to send over some already payed for wood whilst factor 2 fixes the machine, if it turns out to be cost effective. This would help to control cost of goods sold. However, with machines and factories being in isolation, this opportunity does not exist.
Now suppose we have the same company, with 20 factories with 100 machines each, dispersed across the globe. The operational complexity goes up exponentially. In a state of isolation, the machines and factories cannot leverage insights, but in a state of interconnection, in which the whole infrastructure is getting smarter day by day, there are plenty of ways to unlock value.
*Consider Palantir´s Meta-Constellation as a real world example.
Suppose that now we have 10,000 other wood chair companies in the market place with 20 factories and 100 machines per factory. Naturally, they all want to optimize their operations, to produce the maximum number of chairs with the least input.
Would they be interested in buying the ready made software suite that enables insights of the above sort to be unlocked systematically, or would they prefer to buy raw computation, that they then have to adapt to their needs?
If competitively priced, they would most likely prefer the more valuable form of computation. Alex Karp has coined this as valuable compute. You can make quite a fitting analogy with gasoline for your car, for instance. Do you want to buy an oil rig, or do you want to buy gasoline?
This is what Karp talked about in Q4 2021 ER:
“And so, people can stop just buying pure compute, they can buy valuable compute… Wanting a standardized, productized version of what we did at Airbus, what we've done internally at BP, what we did with William. And so, what I can tell you is like of the very big deals we're working on now, they're almost all this.”
There is a strong flywheel forming here. As Palantir continues to onboard customers from different industries (50 and counting, per their latest 10-K), it gets better at providing valuable compute to each industry. For instance, the more clients it deals with in the healthcare space, the better it gets at providing customers in that industry with computation that is directly relevant to their goals.
As the two KPIs that I explore in section 6 continue to progress, this flywheel is likely to have Palantir acting as gatekeeper in the cloud computing space. In 10 years time, it may be common practice to build a company starting with the digital twin first, much like is the case with websites today. Buying raw compute will likely seem sub-optimal, to say the least.
The second derivative effect of this is that we will end up seeing entire industries running on Palantir´s software. This is quite clear when looking at Skywise, for instance. Down the line, this can have Palantir not only making a lot of money, but being quite powerful, Google / Alphabet style.
3.1 Palantir Will Own the Carbon Market
As valuable compute extends to the carbon market, Palantir will become as essential as email.
A relatively immediate example of the above is Palantir´s new focus on helping customers reduce their carbon emissions. As a quick macro overview of the carbon situation, a growing number of end customers prefer to buy products and services from companies that take steps to minimize emissions.
This means that you can “beat your competition and win in the market”, in Palantir´s words, by setting up your organization to understand how it can tweak its operations to minimize the carbon emissions (I bet the wooden chair analogy now makes extra sense). According to Palantir:
“Imagine you'll be able to bluegreen upgrade your streaming pipelines across submarines, factory floors, five and 6G networks. This is a revolutionary capability that will give our customers the edge against their competition.”
“How do you understand the levers you have to pull and the implications of each lever? Do you have the ability to understand the consequences of changing a supplier or setting a different delivery routing or changing a production location on your emissions?”
The flywheel that I describe in section 3.0 will be very much at play here. As Palantir gains experience in serving customers to help them minimize CO2 offerings, they will be able to offer an increasingly productized solution. In 10 years time, most of us will buy “valuable carbon compute” as one of the first steps to set up a company, like today we set up corporate email accounts with Gmail.
When many companies adopt valuable compute for carbon, Palantir will effectively be the core element of the carbon market. If you do not run Palantir, you will have to buy plenty of credits. Else, you will not have to buy as many.
4.0 Competition and Culture
Studying Palantir´s (non) competition leads me to their culture, which is actually its core asset and its key competitive advantage. Valuable compute is / will be Palantir´s moat.
My biggest shortcoming when analyzing Palantir is that I have not tried their product personally. I believe this is the case for most investors aswell. However, there are two very strong market signals that suggest Palantir has the strongest offerings in their domain:
Palantir provides “the operating platform to safeguard America's nuclear stockpile”. The $90m 5 year contract was announced in Q1.
“Microsoft and Boeing launched a competitive ecosystem nearly a year before Palantir and Airbus launched Skywise” Now, “65% of the Boeing fleet is managed in Airbus' Skywise ecosystem”. Skywise now connects more than 100 airlines and 9,000 aircraft (the wooden chair analogy may come to mind).
Firstly, the fact that Palantir is in charge of the US nuclear stockpile I believe makes it the most important company in the west and probably in the world. It is literally in charge of our civilization´s non-extinction. Secondly, a contract of this nature sheds some light on what the US government thinks about Palantir´s software. If it is good enough for this task, it is good enough for me.
As it refers to the aviation quarrel with Microsoft, the Innovation Stack comes to mind again, which I have discussed many times, but here is a refresher. Block ($SQ) beat Amazon in 2014-2015, although Amazon was able to under-price Blocks point-of-sale offering. Why?
Because Block´s culture was 100% focused on building for merchants. The Innovation Stack is Jim McKelvey´s theory that intense cultural focus led to many thousands of little things that Amazon simply could not compete with. In 2015, Amazon sent $SQ point-of-sale hardware devices to all their merchant customers.
When I hear people ask what Palantir can do that Microsoft or Google cannot, I think about their culture and their organizational structure. What Palantir has that these two firms do not is a culture 100% focused in building the best data operating system. This is likely why Palantir beat $MSFT to the aviation race.
Culture is actually Palantir´s core asset. Through time, new alternative solutions can emerge in the market, but if the culture remains strong, you can expect Palantir to keep pushing ahead.
“We spend so much time thinking critically about our people. Who are they? Who can they be? How do we maximize them and their potential. We spend time thinking about exactly what gamma radiation your incoming Bruce Banner needs to turn into the Incredible Hulk. And then we irradiate them.”- Shyam Sankar, Chief Operating Officer @ Q2 2021 ER
“We at Palantir, we're an artist colony, extraordinarily and exquisitely flat. …But we're an artist colony. It makes no sense to tell Dalí to paint more like Monet.” - Q2 2021 ER
You can find many statements such as the above in their public documents. However, the point is that I infer two things about Palantir´s culture from them:
Not much politics involved and plenty of weird and highly talented people around.
Plenty of individual empowerment.
The two always yield an environment were people can be their own weird selves and do their best work. Further, Palantir´s stock option emissions further prove the company´s interest in empowering individuals. It has come at a considerable cost to shareholders in the form of dilution, but I believe is essential for the company´s long term success. It is the binding glue for culture.
The combination of relentless talent development and individual empowerment will have Palantir innovating for decades and the management knows this, so I believe this is the primary way to look at how Palantir will deal with competition going forward.
Having said that, it is also worth noting that Palantir´s management considers they have no actual competition and that if anything, they compete with the internal IT departments of their customers:
“Our competition is not any other company. It's really -- the competition is our customer, specifically our customers' IT department and their desire to build their own solution.” - Shyam Sankar, COO @ Q2 2021 ER
Management also talks repeatedly about how the business has to deal with complex and high risk installation processes, together with long term sales cycles. The truth is, this is a business with a high barrier of entry.
Palantir has invested $3b over 15 years in getting to where it is and to be fair, $MSFT or $GOOG could do this without blinking over the next few years. However, they will have a tougher time catching up with Palantir when valuable compute catches on, as follows:
Once the KPIs in Section 6 (mostly distribution speed) catch on fire, Palantir will become a learning machine.
Each day, it will learn more about how to provide valuable compute to companies in a whole range of industries.
This flywheel will become very powerful. It will be something similar to what we see in Spotify today, with it magically outpacing “competitors” with most people not fully understanding why.
5.0 Karp´s Hair, Management Style and Palantir´s Organizational Structure
Karp´s hair is phenomenal, but so are his management skills.
I have to say it has taken me a while to understand Karp. He talks in a way that seems a bit strange at first and his hair is very distracting and mesmerizing. Through time, however, I have come to appreciate three things about his management style:
A relentless long term focus, which was evident when he stopped issuing short term guidance, to my delight.
He ignores Wall Street.
He is comfortable running a decentralized / hierarchically flat organization and taking capital allocation decisions himself.
Let me explain the above. Earlier in January, I read a book called “the Outsiders: Eight Unconventional CEOs and Their Radically Rational Blueprint for Success” by William N. Thorndike Jr. Apparently, is Buffett´s favorite book.
The book goes through the main characteristics of CEOs that produced stellar returns through time. They are exactly the traits that I outline above about Karp´s management. Here´s my understanding on why these traits lead to great performance:
A long term focus and ignoring wall street go hand in hand and both are key to building value long term. Wall street wants nice shiny numbers every quarter, but as we have learnt from Jack Welch and $GE, this is not the best approach to run a business, at least for shareholders. Here is Karp clarifying this:
Operations are best when decentralized and people are empowered, because it is the optimal way to gather and process / act on information across the span of an enterprise in real time. However, as it refers to capital allocation, this is often the contrary. Very good capital allocation decisions barely ever make it through investment committees, because it requires judgment that is rare and / or gets distorted in the masses.
Whatever reasoning you may want to apply to this, business history strongly suggests that Karp´s traits are typical of a world class manager. Further, this conclusion would be worthless if there was nothing to show for it in terms of business results, but whilst the market has been busy swinging up and down, the company has made some incredible advances which further suggest that world class management is in action here.
Before moving on to review these advances, my take on Karp selling shares is that he is 54 now and it is about time he cashed in a little bit.
5.0 Key Performance Indicators
The key metrics to watch for Palantir are product distribution speed and stickiness. As both metric progress, Palantir will increasingly become a cash machine.
Software is like content in that you only really make serious money when it goes viral. For either to go viral, the marginal cost of replication has to be very low and perhaps more so in the case of software, then it has to stick.
revenue(software) = f (distribution speed, retention)
Palantir´s software is very complex and distributing it (installing it into client companies) is not that easy, making the cost of replication relatively high, in context with other SaaS offerings. On the other hand, retention is very high with a net dollar retention of 131% in FY 2021.
For Palantir to enter the kind of growth that the market loves, it has to totally productize its offerings. In doing so, it would make the marginal cost of replication go down enough for installation numbers to go through the roof, if indeed the product continues to be appealing in the marketplace.
Palantir is still some time away from making that happen. In fact, when I first looked at the company (just when it IPOed), I passed because I thought they would have a very tough time fixing the distribution speed. Now, there is a very strong signal that the company is powering ahead in this sense.
The company has clients in 3 different phases:
Acquire: < 100,000$ revenue per year per customer
Expand: > 100,000$ revenue per year per customer, with a negative contribution margin
Scale: > 100,000$ revenue per year per customer, with a positive contribution margin
The contribution margin of both commercial and government clients is going up fast. Numbers below:
Per the company´s definition of the metric, this implies that product distribution speed is also going up fast:
“We define contribution margin as revenue less our cost of revenue and sales and marketing expenses, excluding stock-based compensation, divided by revenue.”
In the numerator, sales and marketing expenses must be going up because the company is constantly expanding its sales force, so the only factor that can be making the numerator larger relative to overall revenue is a declining cost of revenue, which includes distribution costs.
The cost of revenue and sales and marketing costs include both the costs associated with the deployment and operation of our software as well as expenses associated with identifying new customers and expanding partnerships with existing ones.
In other words, contribution margins are going up because Palantir is getting better each quarter at succesfully putting their offerings into their clients hands.
Digging further into the company, I have found that this improvement in distribution speed is due to increased modularization of the offerings, which enables Palantir to meet customers where they actually are, instead of trying to sell a big piece of software that can scare most IT managers away.
“…the product investments that we've made around modularization and archetypes, those are what are translating also into increased activity.”
“And we were adversarial because they were like, ‘OK, this Foundry thing, yes, great. But we've already built these things. You would replace this, it may honestly could also make us look bad. No one wants it also.’”
In other words, Palantir can now approach customers and say, “I know you have all these legacy systems installed and changing them suddenly is a big risk for you, so here is a little piece of software that can add to your stack”. This enables them to land and expand.
Further, the modularization of their offerings seems to have resulted in predictable price points for end customers which is, almost needless to say, quite important to get sales going.
“…we've been able to really use these modules to meet our customers where they are, predictable price points. It also enables us to really scale channels and enable our partners to help us go to market. So predictable price points, clear problems to go after.”
Through time, I think we will see the company continue to innovate and eventually, they will work out how to deliver a frictionless distribution. When that day comes, Palantir will likely become a cash machine, so long as retention continues to be high.
7.0 Financials and Reliance on Governments
Despite the market´s concerns about profitability, my focus is on Palantir´s notable ability to produce free cashflow. It also a very healthy balance sheet.
Palantir is growing revenue fantastically well. The market has recently been disenchanted by a slow down in government revenue growth, but this seems to be just a little blip in the long term growth trend.
Incidentally, the average revenue per customer has trended down, but this is due to Palantir onboarding more of the little clients, so I am not worried about this trend.
As per usual, we see the market fixating on Palantir not being profitable, but the company is producing formidable levels of FCF, having gone from just $44.4m in FY2019 to $474.4m in the TTM.
On top of this, the company has $2.29b in cash and $0 in debt. Palantir´s financials look very strong to me and so long as FCF keeps going up, I would not be too worried about it.
I have two issues regarding Palantir´s top line, however:
It has a heavy reliance on government contracts.
About half of its revenue is concentrated in its top 20 customers.
About number one, there is nothing inherently wrong with, but I myself do not understand how that part of the business works. I have never sold anything to a government, nor have I had to manage a commercial contract of any form. Thus, I have a bit of a blindspot here.
However, in Q3 2021, management did mention that
“We've worked under four administrations, and have seen consistent continuity across that.”
Palantir´s continued growth in the government business going from one administration to the next I think is relevant qualitative point to have in mind. Another point to have in mind is that Karp mentioned in Q4 2021 ER that European governments had slowed down during the pandemic, generally:
“Europe, in general, has like grew slower and that's -- if you assume that comes back online, it's just going to be bombastic. One of the places that we're actually very strong in Europe is Switzerland.”
It also seems that up until 2018, it was very common for the US government to spend a lot of money on building their own products. This has changed with Palantir´s victory in Federal Court in 2018, although this is worth watching because we do not know when a judge can come up with some new resolution:
“We intend to capture an even greater share of U.S. federal government spending on software systems, following our 2018 legal victory in federal court. The ruling requires the government to consider commercially available products, such as our software, before attempting to build its own.”
About number two, consumer concentration is always an added element of risk that is hard to model, except that it tends to always lead to some unexpected volatility, often at the worse time possible. As Palantir continues to speed up its distribution and acquire more customers in the earlier phases, I do expect this risk to be mitigated incrementally.
Going back to the idea of software / content going viral, I think Palantir´s income statement will take on a very different shape once distribution begins to really take off. For the next year or two, I think we are going to see OPEX continue to cast a shadow on the top line, but once organic / viral growth kicks in, this will be history.
Once this happens, I think we will see the price per share go parabolic, so long as shareholder dilution ceases. Palantir´s FY2018 revenue was $595.4m and $1,541.9m in the TTM, yielding a 2.5 fold increase in revenue during the period. Meanwhile, revenue per share was down 30%.
This dynamic can completely destroy shareholder value in the long term, as has been the case for Twitter.
On the plus side, Karp seems to hint in the Q4 2021 ER call that stock option emissions will normalize through time:
“And because our primary client was not what someone with a hedge fund would think, we didn't actually think of these things from inception. And so, now there's a process of normalization.”
8.0 Thoughts on ESG
Defense companies are now the basis for ESG, in general.
I can totally understand that most people do not like to invest in a company that, well, is involved in killing other humans. I myself am a pacifist. The thing I enjoy most in life is being in nature with loved ones, but I must say that the war in Ukraine has updated my view on how necessary defense is.
The truth is, if we have “bad guys” (it is never black and white) shooting missiles at us, there is no ESG. I think the general public has come to this realization aswell and so generally, I would expect the market to understand that defense companies are necessary to enable any kind of environment and/or social governance.
When the war subsides, people will likely forget, so this is a concern that I think long term Palantir investors need to weigh in on. However, I think there is much to be learnt here from Exxon and Chevron:
The world agreed for a while that oil was not good in general. Now it is evident to everyone that it is actually quite essential and that in fact, we cannot count on 11% of it because we have a relatively mad dictator waging war.
The point is that the market forms narratives (more detailed thoughts on this) that come and go, but unfortunately defense will continue to be essential for a long time, much like oil. In truth, both are the basis of ESG today.
Palantir will become one of the most (if not the most) important software companies in the world and its valuation will likely reflect that one day. I have one major obstacle left before I reach a conviction, however.
I have never tried Palantir´s product and to be fair, this is the case with Blackberry too, which is I a company I am long on. Both companies exhibit very strong signals of the superiority of their offerings aswell. However, Blackberry´s dominance in the auto space is so underpriced by the market, that the investment at today´s prices has more than enough margin of safety built into it.
Palantir´s future, regarding how it is changing / is going to change computing in general is widely disregarded by the market today. However, its existing business I think is quite generously priced at a PS ratio of 14.8 and I somewhat do not think the market is not paying me enough to take on the risk of not fully grasping Palantir´s offerings.
As is often the case with some very promising companies that I look at, I may slowly get more comfortable with the company and eventually, I may start a position. This definitely happened with Amyris, which is another world changing company lurking in the shadows.
If you made it all the way here, thank you very much for reading my deep dive and until next time!
⚡ If you enjoyed the post, please feel free to share with friends, drop a like and leave me a comment.
You can also reach me at:
Antonio, me ha parecido un magnifico artículo. Lo acabo de leer por primera vez y sé que debo leerlo un par o tres de veces más para poder aprovecharlo. Interesante la comparación con Walmart (otra empresa a estudiar), IBM una trampa a no caer en ello, el planteamiento sobre los peligros internos existentes (importante). Un estudio cualitativo impecable y muy interesante (como en Meta). A ver si vuelves a animarte a hacer una presentación en youtube o en un canal en directo.
Hi Antonio, very insightful write-up. Loonshot ideas being bred or even just incubated by companies such as AMD is what investing in these companies is all about. You mentioned edge AI a few times and how chiplets are more efficient to handle pervasive AI especially at the edge. I was wondering if you’ve ever heard the term neuromorphic chips (from what I read, sounds like a new type of chiplet) which makes better “inferences” by filtering out more noise and with 2x less power consumption. Extremely important for battery operated devices! With your knowledge of AMD, do you know if they are working on such a thing? Should they be? Intel is working on one, but I believe ARM, together with their technological partner Brainchip LTD., has the only commercially available neuromorphic chip on the market so far. This may be the next loonshot idea for moving electrons efficiently. Would love your thoughts.