Who Owns Ukraine’s War? Palantir, battlefield data and the hidden economy of artificial intelligence

By Matthew Parish

Sunday 6 September 2026

Wars have always produced things as well as destroying them. They produce new weapons, new institutions, new political arrangements and new forms of knowledge. The First World War accelerated aviation, radio communications, blood transfusion and industrial chemistry. The Second produced radar, ballistic missiles, nuclear weapons and the first electronic computers. The Cold War gave us satellites, packet-switched computer networks and much of the technological infrastructure upon which the contemporary world depends.

The war in Ukraine is producing something different. It is producing data. Not merely ordinary data, of the sort accumulated by governments, banks and social-media companies, but perhaps the most valuable collection of empirical information about twenty-first-century warfare that has ever existed. Every drone flight, electronic-warfare encounter, missile interception, Russian glide-bomb attack, artillery strike, satellite image, thermal signature, reconnaissance mission and battlefield movement potentially generates information. When these observations are combined across millions of events, they become something more than a historical record of a war. They become a means of teaching machines how wars are fought.

Ukraine has understood this better than virtually any other country. Yet there is an uncomfortable question that has received much less attention. Who owns this knowledge? And, in particular, how much of the extraordinary commercial value generated by four years of Ukrainian suffering has accrued not to Ukraine herself, but to the foreign technology companies that came to help her? One company makes the question especially acute: ⁠Palantir Technologies.

The visitor to Kyiv

The story begins remarkably early in the full-scale war. On 2 June 2022, barely three months after Russia’s invasion, Palantir chief executive Alex Karp travelled to Kyiv. He was reportedly the first chief executive of a major Western corporation to visit President Volodymyr Zelenskyy after the invasion. Also present was Mykhailo Fedorov, then Minister of Digital Transformation.

The Ukrainian Presidency was explicit about what was being discussed: cooperation in defence, security and digitalisation. Zelenskyy welcomed Palantir’s willingness to invest in Ukraine and to help her fight Russia on what he called the “digital frontline”. Karp proposed an office in Ukraine and joint development with Ukrainian specialists.

There was nothing obviously sinister about this. Quite the contrary. Ukraine was fighting for her survival. Western governments had not yet supplied many of the weapons they would eventually provide. Russian forces occupied vast areas of Ukrainian territory. Kyiv needed every technological advantage she could obtain. Palantir had something useful to offer.

The company specialises in making enormous quantities of disparate information intelligible. Its software is designed to ingest material from different databases and sensors, establish relationships between it and present those relationships in ways that human decision-makers can use. In civilian life this can mean supply chains, manufacturing or healthcare. In war it can mean intelligence, targeting, logistics and battlefield awareness.

Palantir’s relationship with Ukraine subsequently expanded. In May 2023 the Ministry of Digital Transformation signed a memorandum with the company covering defence and reconstruction. The Ukrainian government described Palantir as already cooperating with multiple state institutions and referred expressly to the exchange of “data and experience”. Its technology was envisaged not only for military purposes but for assessing damage, planning reconstruction, infrastructure and urban development. What began as wartime assistance was becoming an unusually intimate technological relationship between a sovereign state at war and an American corporation.

The world’s greatest laboratory

To understand why this matters, we must understand what Ukraine has become.

There have been other recent wars involving advanced technology. The United States employed extraordinarily sophisticated systems in Iraq and Afghanistan. Israel has developed formidable experience in drones, intelligence fusion and air defence. Azerbaijan’s 2020 victory over Armenia demonstrated the transformative potential of unmanned systems. Ukraine is different.

Here two large states are fighting a prolonged industrial war while simultaneously deploying technologies that did not exist, or existed only experimentally, a decade ago. Electronic warfare interacts with autonomous navigation. Cheap FPV drones hunt individual soldiers and vehicles. Fibre-optic drones defeat radio jamming. Interceptors pursue Shahed attack drones. Satellite imagery is combined with human intelligence and signals intelligence. Artificial intelligence assists image recognition. Both sides innovate and counter-innovate at extraordinary speed.

A weapons-testing range can tell an engineer whether a system works under controlled circumstances. Ukraine can tell the engineer whether it works after the enemy has spent six months trying to make it fail. That distinction is worth a great deal of money.

The importance of the resulting information is no longer speculative. In January 2026 Ukraine’s Ministry of Defence announced Brave1 Dataroom, developed with the Ministry of Digital Transformation, Armed Forces, military intelligence researchers and Palantir. Its purpose is expressly to allow artificial-intelligence systems to be trained and tested using real battlefield data.

By June, more than 100 Ukrainian companies had access. The datasets included visual and thermal observations of aerial targets obtained in real-world conditions, across different weather, times of day and sensor systems. The Ministry said that the data were being used to improve autonomous recognition and interception of Russian drones. It also described millions of annotated frames from actual combat sorties being made available for AI training through a related programme.

This is extraordinarily important. Artificial intelligence depends upon data, but not all data are equally valuable. Ten million photographs of an aircraft against an uncluttered blue sky are much less useful to a military vision system than examples showing the same aircraft at night, in cloud, through thermal sensors, partly obscured, viewed from unexpected angles and mixed amongst false positives. Better still is information showing what happened afterwards: whether the object was correctly identified, whether an interceptor acquired it and whether the engagement succeeded.

The Ukrainian battlefield produces precisely this sort of information. A terrible inversion has therefore occurred. The destruction of Ukraine has itself become an economic resource.

The blood price of information

This should not be expressed melodramatically. Data do not belong to a civilian merely because a missile passed over his house, nor does every Ukrainian acquire intellectual-property rights in an algorithm because the war in which he lives contributed observations used to train it. Nevertheless, there is a moral fact that should not be obscured by the technical vocabulary of artificial intelligence.

The “real-world conditions” that make Ukrainian data so valuable consist of people being killed. A thermal image of a Shahed is valuable because a real Shahed was flying towards a real Ukrainian city. An algorithm learns to distinguish destroyed from operational infrastructure because real buildings have been destroyed. Battlefield-management systems improve because real soldiers die when information arrives too late or is interpreted incorrectly. Air-defence models learn because Russia repeatedly launches weapons at Ukrainian civilians.

In the twentieth century, an invaded country might lose coal, steel, agricultural land and factories to exploitation. In the twenty-first, she can lose something less tangible: information generated by the process of resisting her own destruction. This does not mean that Palantir has stolen Ukrainian data. There is presently no public evidence establishing anything so simple. It does mean that Ukraine should know precisely what she has agreed to.

The bargain of 2022

It would be facile to criticise the Ukrainian government for accepting technological assistance in 2022. The relevant comparison is not between the agreement Ukraine made and some immaculate commercial bargain negotiated by investment bankers at leisure. The comparison is between accepting assistance and possibly losing the war.

If Palantir technology helped Ukraine identify Russian forces, integrate intelligence or allocate scarce military resources more effectively, then software supplied without conventional commercial payment may have been worth vastly more to Ukraine than any licensing fee.

But emergency bargains have consequences after the emergency in which they were made.

Four years later it is reasonable to ask what rights govern the resulting information. Who owns raw Ukrainian battlefield data? Who owns structured versions of it? Who owns annotations? Who may retain them? Who may train algorithms upon them? Who owns the resulting models? Can knowledge obtained from Ukrainian deployments be incorporated into Palantir products sold to the United States, Britain or other governments? What happens when an agreement terminates? What auditing rights does Ukraine possess? What restrictions exist upon onward use?

And perhaps the most important question: did anybody in 2022 understand what these rights might eventually be worth? These are not accusations. They are questions of public administration.

Some answers may quite properly remain secret while the war continues. Operational data cannot simply be published in the interests of transparency. But secrecy concerning military information is not the same thing as an absence of democratic accountability concerning the legal arrangements governing that information. Ukraine should at least possess an authoritative inventory of what she has shared, upon what terms and with whom.

From $30 billion to hundreds of billions

The extraordinary financial history of Palantir inevitably hangs over these questions. The crude argument should be resisted. One cannot take the increase in Palantir’s market capitalisation since 2022, subtract some hypothetical valuation based upon revenue and declare the remainder to have been stolen from Ukraine. Stock markets do not work like that.

Palantir’s revenue has itself increased dramatically. Its SEC filing records revenue of $4.48 billion in 2025, compared with $2.87 billion in 2024, a rise of 56 per cent in a single year. Government customers provided 54 per cent of 2025 revenue and US government customers alone accounted for approximately $1.9 billion.

Its valuation nevertheless became extraordinary. By September 2026 the company had reached a valuation in the region of $440 billion, notwithstanding subsequent volatility; contemporary reporting places this at roughly 28 times the valuation at its 2020 market debut. Even after a substantial correction, Palantir recently traded at approximately 82 times estimated forward earnings, having exceeded 230 times at its 2025 peak.

No sensible analysis can attribute all, or even a specified percentage, of that increase to Ukraine. But something subtler has happened. Palantir entered the Ukrainian war as a company claiming that its software could transform military decision-making. It emerges from four years of war able to point to experience in the most technologically sophisticated major conflict of the age.

That is not merely data acquisition. It is validation. For a defence-technology company, proof that a system works in combat can be worth incomparably more than another demonstration video, laboratory experiment or PowerPoint presentation. Ukraine has provided a place in which technologies encounter Russian electronic warfare, deception, missiles, drones, countermeasures and human ingenuity every day. The result is something Silicon Valley normally struggles to manufacture: reality.

The American customer

The importance of this becomes clearer when we look westwards. Palantir’s largest market is now the United States. In 2025, 74 per cent of its revenue came from American customers and its US government revenue rose from approximately $1.2 billion to $1.9 billion in a year. The company’s political prominence has increased accordingly.

On 10 April 2026 President Donald Trump publicly praised Palantir by name on Truth Social, referring to its war-fighting capabilities. The effect upon the market was immediate: after the shares had fallen sharply, their losses narrowed substantially following the President’s intervention.

There are legitimate separate questions about the relationship between American political power, defence procurement and Palantir. Those questions should be investigated upon their own evidence rather than folded into a theory of conspiracy. For Ukraine, the more important observation is straightforward. A company that has spent years operating in the world’s most important contemporary battlefield is now an increasingly important supplier to the government commanding the world’s largest military establishment. Experience gained in Ukraine therefore possesses a potential commercial afterlife lasting decades.

Then came Fedorov

This would already be sufficient reason for Ukraine to examine the bargain she has made. Events of the last few days make the question considerably more urgent. Mykhailo Fedorov is no longer in government. On 1 September 2026 it was announced that he is establishing a new defence-technology company. Its first major investor will be Alex Karp.

Reuters reported the arrangement on 1 September. Fedorov and Karp had met at Palantir’s Washington office and the former minister described his ambition as using the unique technological experience accumulated during the full-scale war to build technologies for Ukraine and her allies.

The Washington Post put the commercial proposition still more starkly: the venture intends to transform the lessons Ukrainians have learned fighting Russia into technology available in Ukraine and other countries. Karp spoke of bringing Ukraine’s knowledge of warfare as a product to allied states.

There is no basis presently to say that this arrangement is corrupt. There is no evidence presently available that Fedorov improperly transferred state property to Palantir, that decisions he made in office were influenced by the prospect of subsequent private investment, or that his new company will receive confidential Ukrainian military data to which it is not entitled. It would be irresponsible to assert any of those things without evidence.

But democracies have conflict-of-interest rules precisely because impropriety and the appearance of impropriety are different things. Fedorov was centrally involved in the Ukrainian state’s technological relationship with Palantir. He publicly welcomed the company into Ukraine. His ministry signed cooperation arrangements involving the exchange of data and experience. Ukraine and Palantir subsequently built infrastructure expressly intended to turn battlefield information into improved artificial intelligence. He has now left government and Karp is becoming the first major investor in his private defence-technology venture.

These facts justify questions even if every participant has behaved impeccably. What restrictions apply to Fedorov’s use of knowledge acquired in public office? What Ukrainian state information may his new company use? What safeguards prevent confidential military datasets migrating into private ventures? Who will own technologies developed from experience obtained while Fedorov was a minister? What financial interest will Karp acquire? What rights will Ukraine receive?

Above all, where is the boundary between Ukraine’s experience and the private commercial experience of the people and companies who participated in her war? That boundary desperately needs definition.

Data sovereignty

We are accustomed to thinking of sovereignty territorially. A state owns public buildings, military equipment, land and other assets. It guards classified documents and military secrets. Artificial intelligence makes sovereignty more complicated. Imagine that a foreign company receives no Ukrainian military secrets in the traditional sense. Instead, over several years its systems participate in millions of interactions involving Ukrainian sensors, operators, weapons and adversaries. The company improves its software as a consequence. What exactly has crossed the border?

Perhaps nothing capable of being identified as conventional property. Yet knowledge has plainly been transferred. This is why the old vocabulary of ownership may prove inadequate. The important distinction may not be between data that were “stolen” and data that were “given”. The important questions concern rights of use, derivative knowledge and economic participation.

Ukraine needs a doctrine of wartime data sovereignty. Such a doctrine need not prevent cooperation with foreign companies. That would be disastrous. Ukraine’s extraordinary defence-technology ecosystem exists precisely because she has combined domestic ingenuity with foreign capital, software, components and expertise. Instead, the principle should be reciprocity.

If Ukrainian battlefield data train a foreign company’s model, Ukraine should understand what rights she retains in the resulting system. If battlefield experience materially improves a commercially valuable product, agreements should specify how that improvement may subsequently be exploited. If private companies obtain access to state-generated combat datasets, the state should know what happens to derivative models. If officials involved in administering these relationships later enter commercial ventures with the same companies, stringent transparency rules should apply.

None of this requires hostility towards Palantir. Indeed, the best agreements are those in which both parties acknowledge openly what each contributes. Palantir contributes extraordinary software and engineering expertise. Ukraine contributes something that money cannot buy. She contributes the war.

The invisible export

This leads us to a broader problem. For four years Ukraine has been described as a vast recipient of Western assistance. The arithmetic is familiar: so many billions in American weapons, so many billions in European financial support, so many air-defence systems, artillery shells and armoured vehicles. That accounting is incomplete.

Ukraine has also been giving things to the West. She has demonstrated which Western weapons work and which do not. She has revealed vulnerabilities in Russian military systems. She has developed drone tactics that armed forces across NATO are now studying. She has discovered methods of operating under extraordinary electronic-warfare conditions. She has taught Western militaries about the relationship between cheap autonomous systems and expensive traditional platforms. She has provided a laboratory in which artificial intelligence meets industrial warfare.

The West’s defence industries will make enormous amounts of money from these lessons.

Some of that is entirely proper. Companies that develop useful technologies should profit from them. Investors who take risks should receive returns. Without the prospect of profit, many of the technologies keeping Ukrainians alive would never have been developed.

But Ukraine should not imagine that the flow of value has travelled in only one direction. She has been exporting knowledge throughout the war. Much of it was purchased at an appalling price.

Who owns the next war?

There is a final irony. Eventually the war in Ukraine will end. The databases will remain.

Algorithms trained upon this conflict may be used in wars that have not yet begun. Autonomous interceptors developed against Shaheds may defend European cities. Target-recognition systems trained upon Ukrainian imagery may be sold across the world. Battlefield-management software refined through Ukrainian experience may become standard equipment in Western armies. Twenty years from now, soldiers who have never visited Ukraine may survive because machines learned lessons here.

That would be one of the more extraordinary legacies of Ukrainian resistance and there is nothing inherently objectionable about it. Ukraine may rightly take pride in having transformed modern warfare.

But pride is not a substitute for ownership. The immediate requirement is therefore not a fantastical $200 billion lawsuit against Palantir. Nor is it an assumption that every foreign company operating in Ukraine has exploited her. What is required is something simultaneously more modest and more important: an independent accounting of Ukraine’s wartime digital relationships.

Parliament, appropriate security-cleared auditors and competent Ukrainian authorities should establish what categories of battlefield and civilian data have been made available to private companies since February 2022; upon what contractual terms; what rights exist over derivative models and improvements; what retention and deletion obligations apply; what commercial-use restrictions exist; and what post-government restrictions apply to officials who negotiated or supervised these relationships.

Operational secrets need not be disclosed publicly. The legal architecture should be. Palantir may emerge from such scrutiny as an exceptionally generous ally that supplied Ukraine with enormously valuable technology while observing exemplary safeguards over Ukrainian information. If so, that conclusion would deserve to be published as prominently as the questions raised here.

But the questions must nevertheless be asked. The industrial economy treated coal, iron, oil and uranium as strategic resources. The information economy added personal and commercial data. The age of artificial intelligence adds something still more valuable: vast quantities of accurately labelled experience from situations that cannot ethically, economically or practically be reproduced.

War is one such situation. Ukraine has accumulated more of this experience than perhaps any society on Earth. She accumulated it because millions of her citizens fought, fled, suffered, adapted and died. Every night that a Shahed crosses the country, another observation may enter a dataset. Every successful interception teaches a machine something. Every failure may teach it something more.

The tragedy of Ukraine is therefore generating an asset whose value may persist long after the ruins have been rebuilt. Ukraine should welcome allies who help her turn that knowledge into weapons capable of keeping her alive. She should welcome foreign investment and share lessons with countries that may one day face the same enemy. But she should also remember a principle older than artificial intelligence and older than Silicon Valley.

Before giving away something precious, it is wise to know what it is worth. And before the knowledge of this war becomes the foundation of somebody else’s fortune, Ukraine should decide how much of it belongs to her.