The second batch of Rassvet satellites went up on 19 July. By the end of August, public catalogues showed something they had not shown in July: not one spacecraft from that batch has climbed above 360 kilometres, and several are now falling. At the same time, Ukrainian strikes have for the first time reached the factories and launch pads on which the deployment schedule depends. The contest for low orbit has shifted to the place where it is easiest to lose — the ground.
Analysis. Data as of 25 August 2026; some orbital parameters are estimates by independent observers and remain subject to revision.
The big picture
Between July and late August, the story of Russia's low-orbit constellation stopped being a story about orbit. The engineering question — whether the satellites are gaining altitude — has not gone away, and has in fact sharpened. But a second question has appeared alongside it, and over the coming months it is likely to prove decisive: the programme's manufacturing and launch base has become a target, and the ability to hold to schedule now depends on the state of specific workshops in Samara and on how often drone alerts are declared over the Arkhangelsk region.
That changes how the programme should be read. Until July, Rassvet was best assessed as an engineering project carrying inflated political expectations, with the distance between declaration and working infrastructure as the main story — the ground FACTUM covered in its earlier examination of the constellation. Now an external constraint has been added to the internal ones, and it cannot be fixed by budget or political priority: a bottleneck consisting of a single serial rocket assembly plant and a single operational military spaceport.
The same shift is visible from the other direction. Public rhetoric about Russia's ability to destroy Starlink satellites runs separately from practice: in four and a half years of war, not one Starlink spacecraft has been shot down, and the actual instruments of interference remain terrestrial — jamming and strikes on infrastructure. Neither side of the satellite argument is fighting in orbit. Seen that way, Rassvet's current losses are telling: they were caused not by an adversary but by its own propulsion units.
What the catalogues show since 19 July
The second batch launch took place on 19 July at around 20:51 Moscow time from Site 43 at the Plesetsk cosmodrome aboard a Soyuz-2.1b. The timing was reconstructed from an observed pass of the spacecraft; it was never officially announced. Bureau 1440 reported successful separation the following day without naming either the number of satellites or the launch site, and neither Roscosmos nor the defence ministry commented.
The count came from the catalogues. On 24 July the US Space Force logged 17 objects — 16 satellites and an upper stage — in an orbit averaging 290 by 327 kilometres at an inclination of 82.3 degrees. By 1 August the satellites had been designated RASSVET-3 17 to RASSVET-3 32 and listed as in orbit; the "satellite debris" tag that the n2yo service had initially applied was removed. The remaining object was identified as a Fregat upper stage.
From there the divergence with the first batch begins. According to Anatoly Zak of Russian Space Web, in an update dated 24 August, nine spacecraft in the second group began orbit-raising manoeuvres in late July and early August, rising to eleven by 5 August. By the end of the month, however, three more had abandoned the climb and were drifting down at 310 to 330 kilometres, while a pair that never manoeuvred at all had dropped below 300 kilometres and will most likely re-enter the atmosphere. The headline figure: no satellite in the second batch has risen above 360 kilometres, whereas the first batch stood roughly 200 kilometres higher at the equivalent point in flight.
The first batch, as of 3 August, looked like this: twelve spacecraft settled at 518 to 541 kilometres, three lagging badly at 414, 376 and 370 kilometres, with the last two effectively stalled. One satellite burned up on 6 June. In five months, not a single spacecraft has reached the declared target altitude of 800 kilometres.
Andrei Gritsenko, chief executive of the Severnaya Korona space information centre, has calculated what that delivers in practice: the chain of twelve satellites gives a user in the Moscow region up to 3.2 hours of radio contact per day, with a maximum unbroken session of 52 minutes and gaps of up to 10.7 hours. This is a test regime rather than a service — and Bureau 1440 describes it in much the same terms, referring to routine maintenance and equipment tuning ahead of first trials with customers.
The schedule that has to be met
The passport of the federal project on internet access infrastructure sets hard markers: 156 spacecraft in orbit by the end of 2026, 292 operational by the end of 2027, and 383 launched by 2030 including 91 spares. Nine launches were budgeted for 2026, one of them carried over from the previous year.
Two have taken place. As ComNews calculates, roughly 125 more spacecraft must reach orbit before the end of December — seven launches in five months, or a lift-off roughly every three weeks. The rocket can take it: at a stated satellite mass of 370 kilograms, sixteen spacecraft amount to 5,920 kilograms against the Soyuz-2.1b's rated capacity of 7,470 kilograms from Plesetsk, and even eighteen would stay within the vehicle's limits.
The constraint lies elsewhere. Anton Alekseyev, head of the Novy Kosmos company, assesses on-time delivery as possible but carrying a high risk of slipping to the right: the issue is not only rocket availability but satellite readiness, acceptance testing, cosmodrome operations and the ground control complex. The next two or three launches, in his view, will be the test — if a three-week interval holds without technical postponements, the deployment schedule remains achievable.
It is worth recalling that the programme's numbers have already moved. Roscosmos head Dmitry Bakanov said in June 2025 that the first phase would comprise 350 spacecraft, expanding later to 900, at a project cost of 430 billion roubles; the federal project passport works with different figures — 292 operational satellites and 383 launched. The discrepancy is not in itself evidence of failure, but it shows that target parameters have been revised and will probably be revised again.
The ground matters more than the orbit
Here the factor absent from the July picture enters.
On the night of 15 August, a strike hit the Progress Rocket and Space Centre in Samara — Russia's only serial assembly plant for the Soyuz-2 family, and therefore the manufacturer of the rocket that carries Rassvet. Ukraine's general staff confirmed the hit, reporting damage to a workshop assembling engines for Soyuz vehicles and a fire covering some 5,000 square metres; Ukrainian monitoring channels published satellite imagery of the destruction. Among the affected buildings is one where onboard electrical systems and instruments are assembled under high cleanroom standards. Analysts estimate that restoring such facilities takes months: fire and firefighting water in a clean zone mean requalification and recertification of production rather than roof repairs. The Ukrainian side's account of the damage has not been independently confirmed, and Russian official bodies have not commented on it.
On 23 August the Ukrainian monitoring channel Supernova+ reported a drone attack on Plesetsk — Russia's only operational military spaceport and the sole launch site for Rassvet. There is no independent confirmation of a strike; the governor of the Arkhangelsk region twice declared a drone threat, the administration of the town of Mirny said an attack had been repelled, and on the morning of 24 August a Soyuz-2.1b lifted off from Plesetsk with a payload for the defence ministry. This is the first such episode to become public, but not the first attempt: in April, Bakanov told President Vladimir Putin that drones had tried to hit the cosmodrome on the day of a broadband satellite launch and had failed to disrupt it.
Indirect evidence that the threat is taken seriously comes from how the July launch was organised. Russian Space Web notes that navigational warnings for air and sea traffic were issued in unusually wide 24-hour windows, alongside overlapping notices matching several other possible launches. The most straightforward explanation is an effort to obscure the exact lift-off time and the fuelling period, when the rocket is most vulnerable.
The combined effect can be stated without exaggeration. A programme whose schedule was already stretched to the limit now depends on a plant with part of its production floor out of action and on a cosmodrome under recurring drone alerts. That does not cancel the programme, but it moves the risk of failure out of engineering and into the domain where industrial and energy capacity becomes the binding constraint on any technological ambition — a pattern FACTUM has traced in Russia's artificial intelligence build-out.
Where the "downed Starlinks" come from
In parallel, a narrative has taken hold in Russian-language social media that Russia is destroying Starlink satellites. Tracing its origins is worthwhile, because it is assembled from three genuine elements, each true on its own, which together produce a false sum.
The first element is the statements. On 9 July, Alexei Zhuravlev, first deputy chairman of the State Duma defence committee, said Russia could in theory shoot down Starlink satellites from the ground, and named the Peresvet complex. The difficulty lies in what the system is for: Peresvet is a ground-based laser designed to dazzle the optical-electronic sensors of reconnaissance satellites and thereby shield the deployment areas of mobile intercontinental ballistic missiles. It acts on a sensor; it does not destroy a spacecraft. Starlink satellites are communications relays and carry no imaging optics — there is nothing there to dazzle. Western specialists, including the authors of the Secure World Foundation's annual counterspace survey, treat Peresvet's claimed performance with caution in any case: its power, wavelength and beam quality are unknown, and dazzling a sensor is a different order of task from disabling a satellite.
The second element is jamming. Over the summer of 2026, Russian military Telegram channels described an electronic warfare complex called Volna-Kupol-Garant which, they claimed, works against the satellites themselves rather than user terminals, across eight frequency bands; an assertion then followed that Starlink spacecraft had suffered irreversible damage, citing the conclusions of unnamed foreign researchers. This is material from one party to the conflict, traceable to anonymous channels, and it is not independently confirmed. Even if the system exists and is in use, radio interference disables a service for the duration of a pass, not the satellite.
The third element is the video. Footage of burning objects crossing the sky is indeed recorded regularly, and these are indeed Starlinks. They are simply falling of their own accord: the spacecraft are designed for a service life of around five years and for complete burn-up on re-entry, and SpaceX de-orbits them through planned manoeuvres. In a filing to the US regulator, the company reported that between December 2025 and May 2026 alone it conducted controlled de-orbits of 260 spacecraft, with a further 349 retired and queued for disposal. The astronomer Jonathan McDowell puts the current rate at one to two re-entries a day, rising towards roughly five. Against a constellation of some 10,600 active spacecraft out of more than 12,000 launched, that is background noise rather than the result of anyone's action.
There is also a substantive reason why kinetic destruction remains rhetoric. The anti-satellite test of November 2021, which destroyed Russia's own Kosmos-1408, generated around 1,500 trackable fragments by US Space Command's count and triggered an international outcry. Destroying spacecraft en masse at Starlink's operating altitudes would render those orbits marginal for the Russian constellation too — the constellation now trying to reach them. The constraint is acknowledged even within Russian public debate, where the space debris argument surfaces in articles that otherwise call for a hard response to Elon Musk.
The cost and the arguments on the other side
Several qualifications are needed, without which the picture is incomplete.
First, dependence on a single private network is a real vulnerability, and not only for Russia. The episode in February 2026, when Russian military access to Starlink was closed off through whitelist enforcement, showed both sides that the switch sits with a company in a foreign jurisdiction. The logic by which a state builds a channel of its own that cannot be switched off from outside — the theme of FACTUM's earlier study of Bureau 1440 — is rational in that respect, and not merely a propaganda construct.
Second, lowering the operating altitude is not in itself a failure. Vasily Leonov, deputy chairman of the board at Kazakhstan's Republican Centre for Space Communications, points out that 500 kilometres is the most popular altitude among low-orbit operators: a lower orbit means a stronger signal at the terminal and higher throughput, a higher orbit means a larger service area. SpaceX, conversely, is lowering its own spacecraft in 2026 from roughly 550 to 480 kilometres. The question is not whether 500 kilometres is worse, but whether the actual architecture matches the declared one: dropping from 800 to 500 kilometres would, on Anton Alekseyev's geometric estimate, require roughly 1.8 to 2.2 times more spacecraft for comparable continuous coverage — and the original plan is already running behind.
Third, the strikes on Progress fall on a dual-use enterprise that builds both civil launch vehicles and remote-sensing spacecraft. Ukraine justifies them by the reconnaissance use of the plant's output; the practical effect reaches the whole of Russia's space programme, not only its military component.
Fourth, some of the observed anomalies may reflect the normal logic of testing. Sergei Pekhterev, a shareholder in KA-Internet, suggests that the company, having satisfied itself that the spacecraft can be controlled, has moved on to the hardest problem in low-orbit networks — seamless handover of a user between satellites — which is precisely why the spacecraft are being strung into a chain at an intermediate altitude. On his assessment, the transfer from 516 to 800 kilometres will in time become routine.
What could prove this assessment wrong
The cautious conclusion — that the programme is stalling and the risk has moved to the ground — would be wrong under several developments.
If the next two or three launches proceed at roughly three-week intervals without postponement, then neither the damage to Progress nor the threat to Plesetsk has become a constraint: the industry either holds a stock of rockets or has repaired the plant faster than forecast. Talk of a terrestrial bottleneck would then be premature.
If the second-batch spacecraft now sitting at 310 to 330 kilometres resume climbing, the malfunction reading weakens: a slow pace may reflect the sequencing of thruster activation rather than failures. The first batch also climbed slowly and still put three-quarters of its complement at 520 kilometres.
If the reduced operating altitude is a deliberate design decision rather than a consequence of propulsion trouble, the complaint about the unreached 800 kilometres loses its force: the network is simply being built to a different scheme, with satellite numbers and timelines recalculated accordingly. Only an official revision of the orbital architecture, or a fresh filing with the International Telecommunication Union, could confirm this.
If the company is running a closed portion of the programme — and it has twice withheld both the satellite count and the launch site — then the public picture understates progress. Catalogue-based assessment reveals altitude and manoeuvres, but not the condition of the payload or the results of communications testing.
Finally, if Russia does demonstrate the destruction of a foreign satellite in orbit, the proposition that this contest remains terrestrial will be refuted directly, along with all the consequences for orbital debris and for Russia's own constellation.
Conclusions: established and hypothetical
Established. The second batch launch took place on 19 July from Plesetsk and carried 16 spacecraft; by the end of August none had climbed above 360 kilometres, and at least two are descending below 300 kilometres and will probably be lost. Of the first batch, twelve spacecraft have settled at 518 to 541 kilometres, three are lagging and one burned up on 6 June; none reached the declared 800 kilometres. Two of the nine launches scheduled for 2026 have been carried out. On 15 August the Progress centre — the only serial assembly plant for the Soyuz-2 — was hit. No case of Russia destroying a Starlink satellite has been recorded.
Hypothetical. The reasons the spacecraft are failing to gain altitude — propulsion failures, a deliberate change of target orbit, or the sequencing of tests — cannot be distinguished from open data. The scale of damage at Progress and the time needed to repair it rest on Ukrainian accounts and satellite imagery, with no Russian comment. The effect of the August strikes on the launch schedule has yet to show up in observable events.
If current trends hold, the next two to three months will answer the question that matters more than any statement: whether the launch tempo can be sustained. At three-week intervals the programme stays on schedule and accusations of imitation look unfair. At the first slip of a month or six weeks, the 2026 plan becomes arithmetically unreachable — and the binding constraint on sovereign communications will turn out to be not the gap with Starlink, but the condition of Russia's own infrastructure on the ground.
Sources
ComNews, Russian Space Web (Anatoly Zak), The Moscow Times, The Insider, Current Time, Lenta.ru, Kommersant, Habr, TAdviser, RBC Trends, the Foreign Policy Research Institute, Breaking Defense, the Secure World Foundation, EarthSky, SpaceX filings with the FCC, independent orbital tracking data (n2yo.com, NORAD), statements by Bureau 1440, Roscosmos and Russian officials, and reports from Ukraine's general staff and Ukrainian monitoring channels. Estimates of satellite numbers, deployment timelines and damage differ between sources and are given as ranges. Material from parties to the conflict — statements by the Russian defence ministry, Russian military Telegram channels, Ukraine's general staff and Ukrainian monitoring channels — is not independently confirmed. Technical specifications of the constellation are given as stated by the manufacturer.