Thursday, 6 August 2026
+22°758 mmHg
$ 469.85 ▼€ 542.16 ▼₽ 5.78 ▼
kzruen
Shanraq Shanraq
Sign in

← Economy

Daulet Baimurza (karma: 13) · 06.08.26 · 117 ·15 min read
Economy

Palantir: The Company That Cannot Be Outvoted

People have argued about Palantir for twenty years, and mostly about the wrong thing. From altitude: where the company came from in 2003 on money from the CIA's venture arm, what its founders said then and say now, what its numbers actually are — and above all how its charter is built.

Because the company’s future is decided not by its founders’ rhetoric but by one class of share that hands them 49.999999% of the votes no matter how much of the company they still own.

Let me say this up front: this is an opinion, not a verdict. Every checkable claim carries a source at the end. Everything else is my assessment, and arguing with it is not merely allowed but necessary.

Why the argument misses

The usual argument about Palantir runs like this. One side says: this is a company that helps the state watch people, it has worked with US immigration enforcement, its software is used in military operations. The other answers: this is the best data system in the world, it catches terrorists and fraudsters, and the complaints come from people who have never had to keep anyone safe.

Both sides are right on the facts and both miss the point. Because Palantir is that rare company which does not hide what it does. Its chief executive says it plainly, in public, into a microphone:

“Palantir is here to disrupt and make the institutions we partner with the very best in the world and, when it’s necessary, to scare enemies and on occasion kill them.”

That is not a leak and not an exposé. That is Alex Karp on an investor call. When a company says this about itself, denouncing it is pointless — it has already confessed. The interesting question is a different one: what follows, for the future, from the way it is built. That is what we will take apart.

Where they came from

  1. Peter Thiel, fresh from selling PayPal, notices something almost nobody else does: the algorithms PayPal used to catch fraudsters are, in essence, algorithms for finding terrorist networks. The only difference is what you are looking for — suspicious transactions or suspicious connections.

The company is founded by Thiel, Alex Karp, Joe Lonsdale and Stephen Cohen. The name comes from Tolkien: a palantír is a seeing-stone, through which you can look at distant places. A detail rarely mentioned: in Tolkien those stones are dangerous precisely because whoever looks into one never knows who is looking back.

The first money does not come from Silicon Valley. The venture funds passed: a product for intelligence agencies, a long sales cycle, no consumer market at all. The investor is In-Q-Tel, the CIA’s venture arm, which put in about $2 million.

Everything else is contained in that beginning. Palantir was never a company for people. From day one it was built as an instrument for institutions that hold the right to use force. So when it is reproached today for working with the state, the reproach is that the company is doing exactly what it was created to do more than twenty years ago.

The numbers you cannot ignore

The conversation about Palantir is usually conducted in the language of ethics, and misses that this is by now a very large economy.

Second quarter of 2026:

  • Revenue — $1.935 billion, up 93% year on year.
  • Government segment — $990 million, up 79%.
  • Commercial segment — $945 million, up 110%.
  • US commercial revenue — $764 million, up 149%.
  • GAAP net income — $1.062 billion, a 55% margin.
  • Full-year guidance raised to $8.154 billion — 82% growth.

Stop on two of these.

First: a 55% net margin. That is not “a software company”, that is a level almost nobody reaches. More than half of every dollar of revenue is profit.

Second, and more important: commercial is outrunning government. The company built for the CIA is growing faster on corporate customers than on state ones — 110% against 79%, and 149% in the United States. By revenue mix, Palantir is turning into an ordinary enterprise software vendor before our eyes.

And here it gets interesting. The rhetoric is moving in precisely the opposite direction. The more commercial the business becomes, the more martial the leadership sounds. That divergence is the company’s central puzzle, and we will come back to it.

And the number that complicates everything

Palantir’s market capitalisation is around $388 billion, the 36th most valuable company in the world.

Divide by the revenue guidance: 388 / 8.154 ≈ 47 times annual revenue.

For scale: mature enterprise software companies trade at somewhere between 5 and 10 times revenue. So the market is pricing in either a multiple-fold increase from here, or it is wrong.

Let us work out what would have to happen. To reach an ordinary industry multiple of 10 without losing that valuation, Palantir needs revenue of roughly $39 billion — nearly five times today’s. At 40% annual growth that takes about five years; at 25%, about seven; at 15%, more than ten.

That is the substance of the argument about the stock, with the emotion removed. Buying Palantir today is not buying the present business; it is buying the claim that the business will quintuple and hold its margin. Is that possible? Yes. Is it guaranteed? No. We have examined how chrematistics was passed off as economics: the price of a share and the worth of a business are different quantities, and confusing them is expensive.

What they say themselves — and how it changed

Working from primary sources is easy here: Karp writes long shareholder letters that read more like essays, with detours through the New Testament, Nixon and European philosophy. Let us trace the arc.

The early period. The public position was that technology allows security and privacy to be delivered together — that you need not choose between freedom and protection if you design access to data properly. Palantir stressed that it does not own the customer’s data, it supplies the instrument; responsibility lies with whoever wields it.

The middle period. A new theme appears: we work with the West because the West is better than the alternatives. The company stops answering critics with justifications and starts answering with an accusation of naivety — refusing to work with defence is not neutrality, it is a choice in favour of the other side.

Today, said plainly. Karp talks about scaring enemies and on occasion killing them. In 2025 his book with Nicholas Zamiska appears — “The Technological Republic: Hard Power, Soft Belief, and the Future of the West”, a New York Times bestseller. Its thesis is direct: what makes America great is its software industry, and the most important software companies are those whose products secure the supremacy of the West and its ability to do harm to its adversaries. Silicon Valley is reproached for trading serious problems for food-delivery apps.

What actually changed. Not the activity — that stayed the same. What changed is the willingness to call it by its name. The early Palantir explained itself. Today’s Palantir announces itself.

This deserves to be said honestly even by those who disagree: in a world where corporations disguise military contracts as “public safety solutions”, a company that speaks plainly gives society more to argue with, not less. You can dispute it on the merits, because it has stated the merits.

The build: why 49.999999%

And now the part this article exists for. A great deal is said about Palantir and almost nobody reads its charter. Yet that is where the answer about its future lies.

The company has a special class of share — Class F, just 1,005,000 of them, held in a founders’ voting trust and owned equally by three people: Thiel, Karp and Cohen.

Here is how they work. The number of votes attached to such a share is variable. It recalculates automatically so that the founders retain up to 49.999999% of all votes in the company — regardless of how much economic ownership they hold and how many new shares the company issues.

Consider the mechanics. Normally a founder’s control dilutes: the company issues shares, the stake falls, the influence drains away. Here dilution is impossible by construction. The founders can sell nearly all their shares, take the money, and keep the power. Economic interest and political control have been deliberately separated.

And why 49.999999% rather than 51? Because that is exactly enough to be unable to force a decision alone, while being able to block anyone else’s. It is not the power to command. It is the power to refuse to be changed.

The structure was disclosed at the 2020 listing and approved by a majority of investors — they bought the shares knowing it.

What follows

Put the three observations together, because separately they mean nothing and together they answer the question.

One. The company came out of an intelligence contract and by its nature serves institutions that apply force.

Two. In money terms it is rapidly becoming an ordinary corporate vendor: commercial is growing faster than government.

Three. Its founders are shielded from a change of course by the charter itself.

Now add them. An ordinary public company drifts towards wherever the money is: shareholders press, the board replaces management, the rhetoric softens to something safe. Palantir is protected from that drift by construction. Whatever the market thinks, the course is set by three people who cannot be outvoted.

From which follows a forecast, and it is not about “good” or “bad”:

The company will not soften as it commercialises. Normally a growing civilian revenue base forces a business to file down its sharp edges so as not to frighten corporate customers. Here that mechanism is absent — there is nobody who can demand the softening. Expecting maturity and moderation is expecting the wrong thing.

And equally it will not become purely military. Commercial money is growing faster, and nobody is going to walk away from it. So the divergence between martial rhetoric and increasingly civilian revenue will not resolve — it is the stable state.

The central risk lies not in ethics but in concentration. The infrastructure by which states and corporations make decisions is consolidating in a company whose direction is set by three private individuals answerable neither to voters nor to their own shareholders. That is not an accusation of ill intent — it is a description of the construction. The question “what if they are wrong” has no institutional answer.

Why this concerns us

The temptation is to read all this as a distant American story. It will not do, and here is why.

Palantir does not sell software. It sells the ability to see: to fold scattered databases into one picture and make decisions from it. Every state needs that ability, ours included — and buying it is faster than building it.

We have written that data is the new oil that we forget to refine, and that privacy is a form of freedom. Now combine that with the Tolkien detail we opened on: a seeing-stone always has two ends. Whoever looks through it at their citizens is themselves within the field of view of whoever made and maintains the stone.

Hence a practical conclusion, and it is the same one we reached examining Kazakhstan’s bet on artificial intelligence: a bought capability and an owned capability are different things. A bought one works exactly as long as the seller is willing to sell, and only as far as they let you look inside. That is why an accessible open frontier matters more to a country our size than any contract: it gives what a contract never gives — the ability to understand what is actually running.

And one last thing worth holding in mind while looking at 47 times revenue. We have shown that interest on the debt has overtaken military spending, and that the bill for a decade of cheap money has been presented. Companies valued at forty-seven times revenue are part of that bill too.

The searchlight: what can be checked, and when

  1. The commercial share of revenue will keep growing faster than the government share at least to the end of 2027, while the leadership’s rhetoric does not soften. Checkable against the quarterly reports and the shareholder letters — both are published.
  2. The Class F structure will not be abolished or diluted, whatever changes in the shareholder base. Checkable against the annual proxy statement.
  3. Should growth slow below 30% a year, the multiple will compress faster than revenue falls — the share price here rests on the rate of growth, not on the profit.
  4. By the end of 2027 at least one EU state will publicly raise the question of data sovereignty in connection with a Palantir contract. The mechanics of European regulation make this close to inevitable.
  5. A comparable European or Asian competitor is unlikely before 2030. The barrier is not technological but institutional: it takes twenty years of work with closed customers and clearances that money does not buy.

What a reader should do with this

  • Separate the judgement of the company from the judgement of the share. The first is about what it does. The second is about how many times over it must grow to justify its price. Different questions, and expensive to confuse.
  • Read the primary source. Karp’s shareholder letters are open. A company that speaks plainly spares you the need to trust anyone’s summary — including mine.
  • Look at the charter, not only the statements. Statements change; a charter does not. If you want to know where an organisation is heading, find out who is unable to stop it.
  • And ask about the other end of the stone. Not “is the instrument useful”, but “who sees me while I am seeing through it”.

That is my opinion. Check it — and decide for yourself.

Also on Shanraq

Sources

Financial results

Capitalisation and valuation

How the company is built

The founders’ own words

Comments (0)

No comments yet. Be the first.