The National Grammar

Last week, two German companies were competing for the top valuation in European defense tech, and Wire #1 showed how, within that same window, narrative decided between them. It left one question open: why were those two in the running, and not the French company the press had nicknamed "the European Anduril," which was seriously funded and technically sound. The answer fits in one line, and it governs everything that follows: before you can win the room, you have to get into it. Every country guards its door with a language.

Wire #1 described a closed loop: in defense, the same actor, the state, writes the threat and signs the check. That loop has a consequence almost nobody states out loud: the threat is written in a national language. A French white paper, a German Weißbuch, a Turkish doctrine do not describe the same world with different words. They describe different worlds. And the company that wants to answer the threat has to speak the language it was written in.

Call that language the national grammar: the set of codes a company must speak for its own state to recognize it as legitimate. Political scientists call it strategic culture. Every state filters its defense choices through its national culture; Jack Snyder set it out in 1977, Colin Gray extended it afterward. We take the concept and move it: from states to the companies speaking to them.

The architecture has three floors. The closed loop opens the market: the state names the threat and pays for it. The grammar decides who gets into the room. The selector, narrative, from Wire #1, decides who wins once inside. The grammar does not decide who wins. It decides who gets heard. And the company that speaks an imported grammar is not contradicted. It is not heard. No committee rejects its file. The file does not exist in the committee's language.

This frame also disposes of a stubborn cliché. Europe supposedly "has no narrative." The diagnosis comes back at every panel, in every op-ed. It is wrong. Europe does not have a narrative deficit. It has distinct grammars, one per state, where the United States has only one. The continent's typical loser is not the company that tells its story badly. It is the company speaking a language its own state does not recognize.

Four grammars are enough to cover the theater. France speaks state sovereignty carried by the prime. Éric Trappier, at Dassault Aviation, hammers design autonomy and the independence of national choices in every public appearance. The French state hears an industrial house that guarantees sovereignty across decades, not a new entrant that promises it. Germany speaks engineering credibility: you prove before you proclaim. Turkey speaks state-borne national genius: technology as the work of the nation, staged by the state. The United States speaks the Arsenal of Democracy: out-produce the threat, in the name of the free world. One sentence each. The demonstration lives in the cases.

A grammar is not a uniform. It allows dialects: distinct ways of speaking the same state language. The German case gives the cleanest demonstration available right now. Two companies, two dialects, one grammar of engineering credibility.

Helsing speaks the dialect of European technological sovereignty. The chronology is the decisive fact, and it checks out on the record: as early as 2021, the company's public materials install that vocabulary, European technological sovereignty framed as its mission, before any structuring contract, before any political mandate. 2021. On 27 February 2022, Olaf Scholz takes the Bundestag podium and delivers the Zeitenwende. That morning, the chancellor does not create Helsing's dialect. He validates it. The company did not ride the wave. It was swimming before the wave. The order of events rules out the opportunistic reading: you do not tune your vocabulary to a speech that does not yet exist. Today, Germany entrusts it with its combat cloud, a programme of roughly 580 million euros according to the German press. The pre-positioned dialect has become infrastructure.

Quantum Systems speaks the other dialect: combat-proven engineering. The company does not talk about a Europe of power. It counts. More than 19,000 missions claimed for its systems, per its own announcement of 2 July. Its narrative is measured in flight hours, not in political horizon. It is the oldest dialect of the German grammar, the one where the machine proves itself, and it remains perfectly admissible: the German state hears the continental promise and the logbook equally well, because both rest on the same demand, prove before you ask.

Two dialects pulling in opposite directions while holding the same string. Palintropos harmonia, the harmony of opposed tensions. Both companies are audible. This is where Wire #1 takes over: once both are in the room, the selector decides. But the lesson of #2 comes first, and it is chronological. The winning dialect is built before the state validates it. Waiting for the chancellor's speech to learn the language means arriving at the competition after the jury has picked its favourite.

There is a level above the dialect: marrying the grammar. Literally.

The Turkish grammar is state-borne national genius. Technology as a demonstration of what the nation can do, the state as both stage and guarantor. Baykar does not merely speak it. It has built it into its structure. The entanglement runs as far as a family alliance: Selçuk Bayraktar married the president's daughter in 2016. A public fact, not an explanation. The TB2 sells into more than thirty countries because it works. But the grammar in which that success is told is the grammar of national genius. The Bayraktar family founded TEKNOFEST, the national technology festival, and still chairs it through the T3 foundation. The company does not take part in the nation's technological narrative. It stages it, sets its calendar, and builds its audience, the generation of engineers who will carry the story forward. The Kızılelma, its uncrewed combat aircraft, in development and testing rather than mass deployment, carries in its very name an old motif of the Turkish national imagination: the red apple, the horizon the nation is meant to reach. At that level of integration, the boundary between company and state narrative is no longer a line. It is a membrane.

Describing this regime is not judging it. It is noting the most complete alignment in the current theater between a grammar and the actor who speaks it.

And that grammar pays in its own currency: state export. The TB2 equips more than thirty countries, and that is the hard fact. Baykar claims, citing SIPRI, roughly 65 percent of the combat drone export market. The figure is self-reported and should be taken as such, but the hard fact carries the demonstration: the Turkish drone travels with Turkish diplomacy, and every export contract is an act of foreign policy. This victory is not read in a private valuation. It is read on a map.

That leaves the uncomfortable case. It is the French one, and it forces a requalification.

"European Anduril": the phrase comes from the press, never from the company itself, and that nuance matters. Harmattan grew up under an imported nickname. The problem is not that the label is flattering or inaccurate. It is that the category "European Anduril" exists in no French doctrinal document. The French grammar hears houses. Dassault, and behind it the chain of national industrial firms, because a house guarantees autonomy over thirty years, survives changes of government, and sustains a production plan. An independent new entrant promising to do it without the primes speaks a language the French administration does not recognize. It is not fought. It is not heard. Its arguments do not meet objections. They fail to meet a category.

What does exist is the answer the French grammar gave, in its own terms. In January 2026, Harmattan closed its Series B: roughly 200 million dollars, with Dassault Aviation leading.

The lazy reading says: the young wolf got absorbed, the old industry won, France missed its Anduril. That reading is wrong, and it misses the mechanism. Harmattan did not lose. It cashed its victory in French currency. Let us put the table down once and for all: each grammar pays in its own currency. Germany pays in valuation. Turkey pays in state export. France pays in adossement, in being taken under a prime's wing. And Ukraine, which has been writing its grammar in real time since 2022, pays in a currency no other grammar demands: proof at the front. That case, a grammar born in war rather than inherited, deserves its own issue. It will get one.

A French founder backed by a prime, with 200 million on the balance sheet, did not fail to become Anduril. He succeeded in becoming audible, and he was paid at the rate of his own grammar.

The mirror completes the demonstration: Anduril itself. Even the company that rewrote the American grammar has to translate itself to be audible in London. It opened a subsidiary there, set up an industrial site, signed with the MoD. Anduril does not tell Whitehall about the Arsenal of Democracy. It translates: local entity, local industrial footprint, local contracts. When the company that best understood the defense narrative of its generation localizes itself in order to be heard elsewhere, the grammar stops being a French hypothesis. It is the condition of entry, everywhere.

A boundary, before the objection, and it keeps the thesis from explaining everything and therefore nothing. The grammar is not always sovereigntist. Finland, Estonia and Portugal have no grammar of industrial sovereignty: their domestic market is too narrow to fund one. Their grammar is export first. The state there recognizes the company that wins abroad and brings it home. ICEYE, despite its international face, stays anchored as Finnish, and it is that anchor that makes it audible in Helsinki. Tekever wins in London more than in Lisbon, but London is a foreign state: the case falls outside the scope of the thesis, which is about a company's audibility with its own state. The grammar is not what nationalism demands. It is what the state recognizes. Naming this boundary makes the thesis refutable rather than total, the same move as the two regimes of narrative in Wire #1.

The serious objection deserves its strongest form: winners win through product, timing and capital, and the narrative is scenery rationalized after the fact. Helsing wins because its systems work and the German budget opened. Baykar wins because the TB2 flies, strikes and costs little. The grammar would be nothing more than the legend analysts embroider onto industrial victories.

Answer: correct. The grammar does not explain the victory, and this piece never claims it does. Product, timing and capital make the winner; Wire #1 described how narrative decides between actors of comparable substance. The grammar does something else: it decides who is allowed to compete, and in what currency the victory is paid. Hence a distinctive, testable prediction: at equal technical merit, a company speaking a grammar its state does not recognize will not appear on the shortlist. Audibility is measurable before the signature: who is shortlisted, who is invited to tender, who is called to hearings, who appears in programming documents. The badly-grammared company does not lose in the final round. It never appears at that stage.

The test is stated plainly. Find a company that captures durable demand from its own state while speaking an imported grammar without translation. Criteria set in advance: category and legitimacy borrowed from another theater, zero local anchoring, zero backing by a national prime, zero localization. We looked for that case: ICEYE, Milrem, Tekever, the London companies founded by Palantir alumni. ICEYE holds through its Finnish anchor. Tekever wins with a foreign state. None of the candidates meets all four criteria. If the case exists, the inbox is open. It will be worth a published correction.

For the founder, three questions come before the pitch deck. One: what is your country's grammar, sovereignty carried by the prime, engineering credibility, national genius, export first? Two: which dialect do you speak, native, translated, or imported without translation? Three: in what currency does your grammar pay for victory, valuation, adossement, state export? Anyone who answers "I speak like Anduril" without checking that his state hears that language is not preparing a fundraise. He is preparing a silence.

For the investor, the criterion comes before the spec sheet: is this founder audible to his own state? A native positioning, a translated positioning and an imported positioning without translation are three different assets at equal technical maturity, and the currency in which each will be paid appears on no standard term sheet. Due diligence that does not test audibility is assessing a product without checking that a counter exists to buy it.

The bow and the lyre: one taut string, two instruments. Each state tunes its own. Whoever wants to play it starts by hearing it.

Polemos pater panton.