Russia announced on May 12 that the long-delayed Sarmat heavy intercontinental ballistic missile had completed a successful test from the Plesetsk Cosmodrome and would enter combat duty later in the year. The announcement came in a televised videoconference at the Kremlin, in which Strategic Missile Forces commander Col. Gen. Sergei Karakayev reported to Vladimir Putin that the launch mission had been accomplished. It was, depending on how one counted, the second publicly confirmed successful flight of a weapon that had been declared operational nearly three years earlier.
The launch sat in the news cycle alongside the usual run of strategic-rocketry items: Starship development, NASA’s SLS hardware moving through Kennedy, an Indian startup nearing its first orbital launch, a European suborbital demonstration. Sarmat appeared among them as if it were a comparable item of technical news, a rocket that flew, more or less, as advertised.
The interesting fact about Sarmat is not that it flew. It is that the institutional pathway by which a Russian heavy ICBM moves from concept to “combat duty” appears to have changed in ways that have very little to do with engineering and a great deal to do with how strategic capability is now declared rather than demonstrated.
The numbers that sit alongside each other
The missile Sarmat is designed to replace, the R-36M2 Voevoda, known in NATO parlance as the SS-18 Satan, went through a sustained multi-year flight-test campaign before the Soviet Union accepted it for deployment. Flight tests began at Baikonur in March 1986. The first three attempts that year were failures, with at least one of them destroying its silo. Testing continued through 1987 and into 1988, with the missile carrying its ten-warhead bus accepted for service in August 1988. Flight testing of the single-warhead variant continued until September 1989. The Soviet strategic-rocket complex required a statistical record across multiple years, accumulated through both successes and failures, before it would call the weapon ready.
Sarmat was placed on “combat duty” in September 2023 with a publicly confirmed flight-test record consisting of one full success, from April 2022. A subsequent test attempt in February 2023 was reported by U.S. officials to have failed. A test in September 2024 ended in what satellite imagery captured by Maxar and Planet Labs showed as the destruction of a fuelled missile in its silo at Plesetsk, leaving a crater approximately sixty metres wide. A further failure was reported near Yasnyy in late November 2025. The count of publicly confirmed successful tests has now reached two. A week before the May 2026 announcement, the director of the plant that produces the missile was reported to have been detained on embezzlement charges.
The point in setting these histories alongside each other is not to indict the Russian Federation for cutting corners on a nuclear weapon, though one could. It is to ask a narrower structural question: what does the word “operational” mean inside a strategic-weapons bureaucracy when the institutional definition has migrated from a Soviet engineering standard to a contemporary political-signalling one? The same complex that demanded a years-long flight-test campaign in the late 1980s accepted a single full success as sufficient in 2023. Nothing about the physics has gotten easier.
This is not a budget argument. Soviet strategic spending was enormous in absolute and relative terms; modern Russian strategic spending is constrained but not negligible. The gap is not money. The gap is in what counts as proof.
How this actually works historically
Strategic-weapons programs have always had a political surface and an engineering interior, and the two have always pulled against each other. What is striking historically is how often the engineering interior has won, even, sometimes especially, in authoritarian systems.
The Soviet R-7, the world’s first ICBM and the rocket that lofted Sputnik, was tested repeatedly through 1957 and was never a particularly useful weapon. Khrushchev nonetheless rattled it loudly at the West. But the Strategic Rocket Forces, when they actually fielded successors, did so with conservative test programs. The R-16, the R-36, the UR-100, each went through extensive flight testing. The 1960 Nedelin catastrophe, in which a rushed R-16 launch attempt killed personnel including a marshal of the Soviet Union, became an institutional memory that shaped procurement discipline for decades.
The American pattern was similar. Minuteman I conducted extensive test flights before initial operational capability in the early 1960s. Peacekeeper, the closest American analogue to Voevoda, flew successfully through a substantial development program. The reason is not that engineers in either superpower were unusually scrupulous. The reason is that strategic deterrence depends on the adversary believing the weapon will work, and that belief, historically, has been generated by accumulated flight data that intelligence services on both sides could observe and count.
What is unusual about the Sarmat trajectory is not authoritarian propaganda, that has always existed, but the apparent uncoupling of the propaganda layer from the verification layer. In the Soviet era, Khrushchev could lie about how many missiles the USSR had, but the missiles themselves still had to go through their test campaigns before the Strategic Rocket Forces would accept them. The lie was at the inventory level. The engineering threshold held.
Reality-check on the technical claim
Sarmat is a real missile. The program, formally known as RS-28, has been in development since the early 2010s under a contract awarded to the Makeyev State Rocket Centre. It is a liquid-fuelled, silo-based heavy ICBM designed to carry multiple MIRVed warheads or hypersonic glide vehicles such as the Avangard. On paper it is a credible successor to Voevoda.
What it has not done is demonstrate, through repeated flight, the reliability that the term “combat duty” historically implied. Of the publicly known flight tests, the April 2022 test was a clear success. Subsequent attempts produced an alleged failure in February 2023, a silo-destroying explosion in September 2024, a further reported failure in late 2025, and now the May 2026 success. Two publicly confirmed full successes in roughly four years of attempts is not a deployment-ready statistic by any historical standard for a strategic nuclear weapon. The structural difficulties of the program, supply-chain stress under sanctions, the loss of the Ukrainian engineering inputs that fed the Soviet missile complex, corruption indictments at the manufacturing plant, are documented and serious. None of this means Sarmat will not eventually become a reliable weapon. It means the gap between the announcement and the engineering reality is wider than the announcement admits, and the announcement is happening anyway.
The politics of the artefact
Langdon Winner, in his 1980 essay “Do Artifacts Have Politics?”, argued that technologies embed and express choices about social order, that the question of who a technology is for, and what institutional arrangements it ratifies, is built into the object itself rather than added on later. The argument applies with peculiar force to strategic weapons, because a nuclear missile is almost entirely a political object. Its actual use is the failure case. Its purpose is to be believed in.
Which means that the production of belief, the test cadence, the announcement schedule, the parade footage, the satellite-visible silo activity, is not ancillary to the weapon. It is the weapon. And the question of how much real engineering must underwrite the belief becomes a structural question about the political system doing the believing and the political system doing the signalling.
The Soviet strategic complex, for all its pathologies, contained internal constituencies, the Strategic Rocket Forces, the design bureaus, the General Staff, that could insist on engineering thresholds even against political pressure. The flight-test record was a currency these institutions used among themselves and against the Politburo. A sustained multi-year test campaign was not a number imposed by the Kremlin; it was a standard the rocket forces demanded.
The contemporary Russian system appears to have hollowed out that internal contestation, or at least to have reordered it. When the political signal, operational by year’s end, can be issued without the engineering currency to back it, something has changed about the internal balance of the institution. The weapon, in Winner’s sense, has acquired a different politics. It is for the announcement more than for the silo.
What this means, and what it does not
It would be easy, and wrong, to read this as a story about Russian decline or about authoritarian mendacity in general. Authoritarian systems are perfectly capable of running disciplined long-cycle engineering programs. China’s strategic modernization, whatever else one thinks of it, has not been characterised by premature operational declarations. And democratic systems generate their own forms of capability theatre; the F-35 program spent years being described as combat-ready in configurations that the operational test community considered nothing of the sort.
The narrower question Sarmat raises is what happens to the meaning of “operational” inside any strategic-weapons bureaucracy when the political layer’s appetite for announcement outruns the engineering layer’s capacity to verify. The Soviet system held those layers apart with institutional walls that have, in the Russian successor system, apparently thinned. Other systems have their own walls and their own thinning points.
There is a second-order question, which is what foreign intelligence services and arms-control negotiators are supposed to do with capability claims that have decoupled from test data. The traditional logic of deterrence assumed that an adversary’s announcements could be roughly cross-checked against observable engineering reality, and that the two would converge over time. If they do not converge, if the announcement is the capability, in the sense that matters politically, then the strategic conversation is being conducted in a different register than the one arms-control regimes were designed for. That is a problem for any New START successor negotiations and for the analysts trying to write threat assessments in the meantime.
None of which tells us whether Sarmat will, in fact, work. It probably will, eventually; heavy liquid-fuelled ICBMs are within the reach of the Russian industrial base, and more years of testing will produce a more credible weapon than two confirmed successes have. The structural fact is that the bureaucracy producing the missile no longer needs it to work in order to call it operational. So the question worth holding is not whether Russia has a new ICBM. The question is: when a strategic-weapons institution can issue the announcement without the engineering, what exactly has been deployed?
