STATUS NOTE AND WHAT THIS PAGE IS

This page was written on the evening of 18 October 2026 (US Central time), while events were still unfolding, and every "current" statement below is only as fresh as that. The cluster is Kosmos 2609, 2610, 2611, 2612, 2613 and 2614, six Russian Ministry of Defence satellites that went up on one Soyuz-2.1b rocket. Their purpose has never been stated by Russia. What the outside world knows comes almost entirely from orbital tracking data (what the satellites do, not what they are) plus a handful of analysts who read that data carefully.

One satellite of the six, Kosmos 2612, is the reason this page exists: in the last three weeks it has manoeuvred to within a few kilometres of, and possibly much closer than that to, ICEYE-X36, a US-registered commercial radar-imaging satellite whose operator supplies imagery to Ukraine. ICEYE-X36 is a commercial satellite. It is not a NATO military satellite. The US registration and the Finnish founding of its operator are what make the situation sensitive for NATO countries, but it is not a military asset of any alliance member. [DOCUMENTED]

Plain summary of the state of knowledge: a deliberate, fuel-expensive, patient approach has been documented by several independent trackers. Nobody outside Russia knows what the satellite is for. No hostile act has been documented, no government has publicly characterised the event on the record (beyond a Space Force remark in May, see below), and ICEYE has declined to comment. Everything about purpose is speculation, and this page labels it that way.

HOW TO READ THIS PAGE

Every substantive statement carries a label.

• [DOCUMENTED] — stated by an official source, or by tracking data/analysts in a way multiple independent sources agree on.
• [SINGLE-SOURCE] — one secondary source only; plausible but not cross-checked.
• [UNVERIFIED] — claimed in circulation but not confirmable from public material.
• [ESTIMATE] — my own arithmetic or engineering reasoning from standard physics, shown so you can check it.
• [SPECULATION] — a hypothesis about intent or capability. Used heavily in this article because intent is the unknown.

No images are used on this page. No photograph or official image of any satellite in this cluster exists in public sources, and drawing a diagram that implies we know what they look like would be misleading. Orbital geometry is explained in words and small tables instead.

Note on the "five of six" and "which one" confusion. Reports disagree on exactly which satellites made the May plane changes and which one passed ICEYE at 2.5 km. This page sets the disagreement out in the individual satellite sections rather than hiding it. The US catalogue also only presumes which NORAD object carries which Kosmos number (see the numbering section), which is one reason such confusion is possible.

CURRENT EVENTS: WHAT KOSMOS 2612 IS DOING RIGHT NOW (as of the evening of 18 Oct 2026 US Central)

This section is the most perishable part of the page. Times are given in UTC as the sources give them, with the US Central equivalent where useful (Central is UTC-5 in October).

The approach, in one paragraph. From 30 September, Kosmos 2612 began a short, efficient sequence of burns (details in the timeline section). By 12 October it sat in ICEYE-X36's orbital plane at almost exactly its altitude, about 475 km behind it. Burns on 15 and 16 October closed that gap quickly. On Friday 16 October it was roughly three miles (about 5 km) from ICEYE-X36 according to LeoLabs, a commercial radar-tracking firm, after being more than 200 miles away the day before (as reported by Axios). [DOCUMENTED, via LeoLabs/Axios, and consistent with tracker estimates of 3 to 5 km on 16 Oct]

What the independent trackers say happened next.

• Jonathan McDowell (astrophysicist who maintains the main public satellite catalogue) said Kosmos 2612 made "a single slow pass" of ICEYE-X36 "at a distance of a few km" on 16 October, at a relative speed of about 20 km/h, and on 17 October said new data showed it had manoeuvred on 16 October and was "stationkeeping within 5 km or less". He called it, to his knowledge, the first rendezvous in low orbit of two active satellites where one owner did not give approval, and does not think it "quite violates the Outer Space Treaty". [SINGLE-SOURCE: his statements on X as quoted by the America 2.0 "Space War Watch" page; I could not open the posts directly]
• Marco Langbroek (TU Delft lecturer, SatTrackCam Leiden blog) wrote on 17 Oct (18:15 UTC) that the two were within 1 to 2 km, that the burns from 30 Sep were "clearly designed to bring the two objects very close and synchronise their orbital periods, orbital altitude and eccentricity", and that he expected a pass within metres early on 18 October. His update of 18 Oct 14:30 UTC, using orbital elements with an epoch near 07:39 UTC, says there was a close approach early on 18 October, possibly around 01:35 UTC, "to within a few hundred meters or possibly (much) closer", that Kosmos 2612 appears to have made one or more small manoeuvres around or just after that time, and that he sees no sign in ICEYE-X36's orbital elements that it was hit or disturbed. He adds: "Whether it functions after the close approach, only the users can tell us." [DOCUMENTED: his blog, read directly]
• America 2.0's "Space War Watch" tracker (an aggregator that recomputes separation from public US data every two hours; not a primary source) reported at 22:12 UTC on 18 October (5:12 PM Central) that public US data show an apparent small burn early on 18 October that put Kosmos 2612 on a slightly higher, drifting orbit, with the gap about 12 km by 15:36 UTC and growing. It says independent analysts had not yet commented on that, and that no hostile act has been documented and no government has commented. It sets a "hold test" for 24 October: whether the satellite comes back. [SINGLE-SOURCE; derived from public CelesTrak data; read directly]

Conflicting data point: the "26 m" prediction. The commercial tracker orbitalradar.com shows a "26 m miss distance, closest approach 2026-10-19 16:07 UTC, relative speed 0.00 km/s, altitude 541 km" page. Its own text says miss distances are computed by propagating the latest published orbital element sets and "values update as new element sets are published". That makes it a prediction from public elements, not a measurement, and it appears to be inconsistent with the later data (a burn on 18 Oct and a growing gap). Treat the 26 m figure as superseded or at best unreliable. The same site was quoted in a Washington Times article of 16 October as showing a "miss distance of about 85 feet" (about 26 m), which the article said occurred "Sunday"; since on Sunday 11 October the satellites were hundreds of kilometres apart, that wording appears to have confused a predicted future value with an event that had happened. Likewise the "136 m and 48 m" figures that circulated earlier were predictions for different passes. Predictions swing wildly because a few metres per second of burn, or a few hundred metres of error in the orbit estimate, changes the closest-approach number completely. [DOCUMENTED that the page says what it says; the interpretation is mine, flagged as ESTIMATE]

Why the numbers disagree: the data limits. The public orbital data (two-line element sets from the US Space Force's Space-Track, republished by CelesTrak) have uncertainties of roughly a kilometre or a few. Below about 2 km the public data cannot reliably say how far apart two objects are, and two objects flying in formation can even be confused in the catalogue. Radar trackers (LeoLabs) and the US military's own systems are far more precise, but their precise numbers are not public. So "a few hundred metres", "26 m" and "a few km" can all be honest statements by honest people looking at different data of different quality. [DOCUMENTED as a general property; ESTIMATE for the specific application]

What ICEYE and governments have said. ICEYE declined to comment to Axios (17 Oct); its press-release page showed no related statement as of my check (latest releases 8 and 1 October, the latter an unrelated Nokia partnership). The Pentagon declined to comment on Friday and Saturday. A Ukrainian official told Axios that Ukraine's military works with ICEYE but said it was unclear whether this specific satellite supports Ukraine, and that Ukraine was not aware of the recent movements but had been informed of similar manoeuvres in spring and summer. In May, the US Space Force said the Cosmos satellites were positioned close to a "Western commercial radar satellite used by Ukraine for imagery intelligence" (as quoted by Axios). [DOCUMENTED via Axios; I did not find the original Space Force text]

What has not happened (as far as public sources show). No object has been reported released from Kosmos 2612. No debris event. No reported loss of ICEYE-X36, no manoeuvre by ICEYE-X36, no ICEYE report of interference. No government has attributed any hostile act. [DOCUMENTED as absence in the sources I could find; absence of public reporting is not evidence that nothing happened]

What to expect next (ESTIMATE/SPECULATION). If the satellite really is drifting away on a slightly higher orbit, the likeliest readings are (a) the approach was a one-time "pass" for collection or a demonstration, now complete, (b) it is repositioning for a closer, better-lit pass, or (c) the burn was a safety or phasing trim. The decisive observation is whether it comes back, which is why the 24 October date matters. Any new catalogued object from the April launch, or any ICEYE statement, would change the picture immediately.

THE SHORT VERSION: KNOWN AND UNKNOWN

Known (strong confidence).

• Six payloads Kosmos 2609 to 2614 reached orbit on 16 April 2026 at 23:17 UTC (02:17 Moscow time on 17 April), on a Soyuz-2.1b with a Volga upper stage from Plesetsk, site 43/4. Russia's Defence Ministry announced "several" satellites for the ministry, and withheld designations and missions. [DOCUMENTED]
• They sit in near-circular, sun-synchronous orbits at roughly 540 to 556 km altitude, around 96.95 to 97.9 degrees inclination, circling the Earth about every 95 to 96 minutes. [DOCUMENTED from US tracking data via RussianSpaceWeb]
• In mid-May, several of them made large inclination changes (about 0.8 degrees, costing on the order of 105 m/s each) that put them in the plane of ICEYE-X36. [DOCUMENTED; the exact count is disputed, see below]
• From 30 September, Kosmos 2612 alone made a precise sequence of burns that again matched ICEYE-X36's plane, altitude, eccentricity and period, then closed the gap to a few km by 16 October. [DOCUMENTED, multiple independent analysts]

Unknown.

• What any of the six are for. Russia has not said. [DOCUMENTED]
• Whether any carries a weapon, jammer, grapple or sub-satellite. [UNKNOWN]
• Whether the six are identical. Three different orbital "families" are visible at launch (see below), which might mean different roles. [SPECULATION]
• Whether the approach to ICEYE-X36 is the mission or one chapter of a rehearsal. [UNKNOWN]
• Whether ICEYE-X36 is the specific satellite serving Ukraine. It shares its orbital plane with sister satellites X37 and X38, and which ICEYE satellites serve which customer is not public. [SINGLE-SOURCE via COMSPOC as quoted by America 2.0]

THE LAUNCH AND THE GROUP AS A GROUP

The launch. A Soyuz-2.1b rocket with a Volga upper stage lifted off from Plesetsk (site 43, pad 4) at 02:17 Moscow time on 17 April 2026, which is 23:17 UTC on 16 April. That is why some sources say 16 April and others 17 April; both are correct in their own time zone. Russia announced success around 09:30 Moscow time. Residents in Finland reported seeing the ascent. [DOCUMENTED, Anatoly Zak / RussianSpaceWeb]

Zak notes it was the first flight of the Soyuz-2.1b with Volga in this role (one aggregator lists it as the only such flight). Volga is a small, restartable "space tug" stage, meant to deliver several payloads to different orbits from one launch. No safety notice had been issued for the sea area used to de-orbit a Volga stage, which is unusual. [DOCUMENTED for the observations; the "first flight" claim is SINGLE-SOURCE (flightatlas.org) and conflicts with the idea that Volga flew before, so treat it as unverified]

What the US catalogue first saw. The US Space Force initially tracked ten objects (2026-083A to 083K, with two NORAD numbers skipped because a Chinese launch took them). By early May, only six were identified as payloads; the other four were the rocket body, the Volga stage, and two bits of debris. Zak and Russian forces blogger Pavel Podvig's site (russianforces.org) both record this. [DOCUMENTED]

Two orbital families at insertion. At first the objects sat in two sets of planes: Object A and B (the Soyuz third stage and what became Kosmos 2609) near 98.25 degrees, and the other five satellites (F, G, H, J, K, i.e. 2610 to 2614) near 96.95 degrees at about 540 to 556 km. Zak records that the Volga stage made an unexplained inclination-change manoeuvre from about 98.26 to 96.95 degrees before releasing the second group, which probably used up the fuel it would have needed to de-orbit itself later. [DOCUMENTED]

What that tells us (ESTIMATE/SPECULATION). A tug that deliberately changes plane to release a batch means the batch's final plane mattered. The target plane (96.95 degrees at that RAAN) is within about 0.5 degrees of ICEYE-X36's right ascension, per Integrity ISR's May report, which is a surprisingly good match for a "random" sun-synchronous orbit. That makes it plausible the cluster was designed from the start to be near ICEYE-X36's plane. But it is also true that sun-synchronous orbits at similar altitudes and similar launch times naturally end up near each other's planes, and McDowell and Bart Hendrickx initially suspected coincidence after the May events, so this is not proven. [SPECULATION; the initial doubts are SINGLE-SOURCE via Meduza as quoted by America 2.0]

The number of satellites is itself unclear. RussianSpaceWeb's summary table says "Kosmos (8 satellites, including an experimental Okulus)". On 6 June 2026 the Russian startup Novy Kosmos announced that its first satellite, an experimental radar-imaging technology demonstrator, had been launched in April 2026 and was in flight tests. That suggests there may have been rideshare or non-military payloads on the same launch, which would explain why more objects were tracked than the US counted as Kosmos payloads. Whether Okulus is one of the six Kosmos satellites is not stated by Zak. Update (Bart Hendrickx, The Space Review, 12 Oct 2026): Novy Kosmos's own customer guide gives Okulus-01 and Okulus-02 orbits of about 530 and 535 km, 97.5 degrees, crossing the descending node at 10:30 local time. The Plesetsk satellites are in higher, more inclined orbits with a mid-afternoon node, so neither Okulus-01 nor -02 is among Kosmos 2609 to 2614. Hendrickx identifies Okulus-01 as the Chinese Taijing-4-03 (MinoSpace, launched 23 Jan 2024) and Okulus-02 probably as another Chinese radar satellite (Dongpo-16 most likely). He considers it more likely that Novy Kosmos launched no satellite of its own in April; a Novy Kosmos demonstrator among the six is a "remote possibility". [SINGLE-SOURCE (Hendrickx), well argued from published orbital elements] https://www.thespacereview.com/article/5335/1

WHAT KIND OF SATELLITES MIGHT THEY BE? (every item is a hypothesis)

Nothing official exists. The candidates people put forward:

• Radar-imaging satellites. Anatoly Zak, as quoted by russianforces.org, suggested the group might be radar-imaging satellites developed by NPO Mashinostroyeniya (Reutov), possibly the "Almaz-X" / "Diamond-X" concept advertised around 2023: five satellites of 350 to 400 kg each for a 500 to 550 km sun-synchronous orbit, launched by Soyuz/Fregat. That is a good numerical fit for five satellites at about 550 km. Some public catalogues list Kosmos 2612 as an Earth-observation satellite of about 500 kg, which is almost certainly a catalogue guess based on that suggestion rather than new information. [SINGLE-SOURCE (Zak); the ~500 kg figure is UNVERIFIED]
• Inspector satellites of the «Нивелир» (Nivelir) lineage. Jonathan McDowell has called Kosmos 2612 "an inspector satellite", and Czech space scientist Michal Vaclavik identified it as part of the Nivelir programme of inspection satellites. Bart Hendrickx has documented Nivelir-class satellites shadowing US reconnaissance satellites since 2017. [SINGLE-SOURCE via America 2.0's summary of X posts and The Space Review; I did not read the originals]
• Something else entirely. Electronic-intelligence, communications-relay, or an experimental swarm testbed. Nothing rules these out. [SPECULATION]

An important logical point. A radar-imaging mission and an inspection mission are not exclusive. A group of small satellites flying in one plane could be a legitimate imaging constellation whose operators also practise proximity operations on a convenient real target. The proximity behaviour is documented; the mission label is not.

THE SIX SATELLITES ONE BY ONE

On the numbering. Russia has not released the Kosmos numbers. They come from the US Space Force catalogue, which as of 2 May listed each payload as "COSMOS 261x PAYLOAD". Several analyst sources note the number-to-object mapping is presumed (the RussianSpaceWeb table says "presumed", and the McDowell catalogue marks each with a question mark per America 2.0). This matters because the identity of "the one that passed at 2.5 km" depends on it. [DOCUMENTED that the labels are presumed]

Orbit table at insertion (US catalogue values, as of early May, from RussianSpaceWeb)

Other launch objects, for completeness: 2026-083A (68753) Soyuz upper-stage rocket body; 083C (68755) and 083E (68757) debris; 083D (68756) the Volga stage. NORAD 68760 and 68761 belong to another launch. [DOCUMENTED]

Kosmos 2609 (2026-083B, NORAD 68754): the odd one out

• Plane: launched into the different 98.25 degree family, not the 96.95 degree one. [DOCUMENTED]
• May manoeuvres: according to Integrity ISR (22 May) and Skyrocket's summary, the plane changes of mid-May were by Kosmos 2610 to 2614, i.e. the five satellites that began near 96.95 degrees. That would make 2609 the sixth satellite that did not manoeuvre into ICEYE-X36's plane. [DOCUMENTED by two sources agreeing]
• The conflict: CSIS (Swope and Li, 15 Oct) says five of the six made the plane changes and that Cosmos 2609 got closest to ICEYE-X36, at 2.5 km, on 26 May. If 2609 was in the 98.25 degree family and did not change plane, a 2.5 km pass by it would be a coincidental crossing rather than a rendezvous, which does not fit CSIS's description, so either CSIS mis-assigned the number (the numbering is presumed), or Integrity's list of which five manoeuvred is off by one. I cannot resolve this from public information. [UNVERIFIED which satellite made the 2.5 km pass]
• What it did later: not documented in any source I found. [UNKNOWN]
• Why it might be different (SPECULATION): a different plane could mean a different job, a spare, or a satellite whose role is not about ICEYE at all, for example a conventional payload released together with the "tugged" group.

Kosmos 2610 (2026-083F, NORAD 68758)

• Launched in the 96.95 degree family. [DOCUMENTED]
• Part of the May plane-change group; Integrity says it, with 2611 to 2614, came to about 97.8 degrees between 14 and 20 May. [DOCUMENTED]
• In Integrity Issue 152 (23 Aug), the five were stacked about 0.6 to 5.2 km above ICEYE-X36 and drifting through its plane. By October it has ~112 to 114 m/s of estimated propellant expenditure since launch and has not manoeuvred since July. [SINGLE-SOURCE: Integrity ISR, using CelesTrak data; the delta-v is their estimate]
• Role unknown; behaviour is "group-typical". [UNKNOWN]

Kosmos 2611 (2026-083G, NORAD 68759)

• Launched in the 96.95 degree family; part of the May group; ~112 to 114 m/s spent; no manoeuvre since July. [SINGLE-SOURCE, Integrity ISR]
• Forward-looking item: Integrity says its plane and that of Kosmos 2613 are converging on that of ICEYE-X37 (a sister ICEYE satellite), with plane crossings around 6 and 15 November. Integrity suggests watching these two for the same "climb, let nodal precession do the work, descend" sequence that Kosmos 2612 used. [SINGLE-SOURCE; forecast, not fact]

Kosmos 2612 (2026-083H, NORAD 68762): the active one

• Launched in the 96.95 degree family; part of the May group; in August it was the closest of the stack, only about +0.6 km above ICEYE-X36. [SINGLE-SOURCE, Integrity ISR]
• Since launch it has spent about 175 m/s of delta-v by Integrity's estimate (as of 12 Oct, before the 15 and 16 Oct burns), about 60 m/s more than any sibling. [SINGLE-SOURCE; ESTIMATE by Integrity]
• Its manoeuvres are detailed in the timeline: 30 Sep climb, 7 Oct trim, 12 Oct descent, 15 and 16 Oct closing burns, small burn(s) around 18 Oct. [DOCUMENTED, multiple sources]
• Possible name: "inspector satellite" (McDowell), Nivelir (Vaclavik). [SINGLE-SOURCE / UNVERIFIED]
• What we can tell from its behaviour (ESTIMATE): it is manoeuvrable, precisely navigable (it can place itself within a few km of another satellite from 475 km away in about three days), and has enough propellant for further close-in work. It has no publicly known payload.

Kosmos 2613 (2026-083J, NORAD 68763)

• Launched in the 96.95 degree family; part of the May group; ~112 to 114 m/s; no manoeuvre since July. [SINGLE-SOURCE]
• Plane converging with ICEYE-X37's, crossing about 15 November (see 2611). [SINGLE-SOURCE; forecast]

Kosmos 2614 (2026-083K, NORAD 68764)

• Launched in the 96.95 degree family. Integrity noted in the May-June reporting that 2614 was still in the process of its inclination change after the other four, so it may have been the last of the five to finish. [DOCUMENTED per RussianSpaceWeb quoting Integrity]
• America 2.0 reports that on its data, 2614 will cross near ICEYE-X36 around 21 October, but on a slightly different orbit, which would be a fast pass rather than a rendezvous. [SINGLE-SOURCE; forecast]
• Otherwise as 2610. [UNKNOWN]

What the May manoeuvres actually were

In May, between about 14 and 20 May, the five satellites each raised their inclination from about 97.0 degrees to about 97.8 degrees, a change of roughly 0.8 degrees, at a cost Integrity puts at 105 to 106 m/s each. [DOCUMENTED] A quick check, my own: the delta-v for a pure plane change is about 2 x v x sin(change/2). At 550 km the orbital speed v is about 7.6 km/s. For 0.8 degrees: 2 x 7.6 km/s x sin(0.4 degrees) = 2 x 7.6 x 0.00698 = 0.106 km/s = 106 m/s. That matches the reported figure, so the reported change is physically consistent, which is encouraging for the data quality. [ESTIMATE]

Integrity also noted that the satellites had been launched with their right ascension of the ascending node (the longitude where the orbit crosses the equator northward, i.e. the "rotation" of the orbital plane around the Earth's axis) just 0.5 degrees from ICEYE-X36's. Matching inclination as well would put them 0.5 to 22 km from ICEYE-X36 at various times. [DOCUMENTED, via RussianSpaceWeb quoting Integrity]

After the plane changes, nothing was done to keep the planes matched. This is the oddest part of the May story. If the aim were a permanent co-orbital relationship, you would expect them to hold the match. Instead, sitting at slightly different altitudes above ICEYE-X36 (0.6 to 5.2 km), their planes precessed at slightly different rates and drifted apart through the summer. That is why many observers, including McDowell and Langbroek initially, were doubtful that anything was being targeted. Seen in hindsight, it looks like the plane change was preparation for the later pass, or a rehearsal. [SINGLE-SOURCE via Integrity/CSIS for the facts; the interpretation is SPECULATION]

The 2.5 km pass. CSIS says Kosmos 2609 came within 2.5 km on 26 May. As discussed above, the identity is disputed. Several other satellites came "near" ICEYE-X36 in May as well. COMSPOC and LeoLabs reportedly published a timeline of the May approaches. [DOCUMENTED that a ~2.5 km pass occurred; UNVERIFIED which satellite]

TIMELINE

The date of the last ICEYE-X36 manoeuvre matters for one reason: it tells us ICEYE can manoeuvre, but has not been seen doing so since June. The satellite is only about 90 kg, and its remaining propellant and agility are unknown. [SINGLE-SOURCE]

ORBITAL MECHANICS IN PLAIN LANGUAGE

This section explains why the pattern of manoeuvres looks the way it does. The same physics applies to every satellite.

Orbits as hoops

Picture an orbit as a rigid hoop, tilted in space, with the satellite sliding around it. Three numbers fix the hoop's orientation: its inclination (the tilt from the equator), its RAAN (how the hoop is rotated around the polar axis), and its shape and size (here nearly circular at about 550 km). A satellite on the hoop moves at about 7.6 km/s and completes it in about 95 minutes. [DOCUMENTED textbook physics]

What "coplanar" means

Two satellites are coplanar if their hoops lie in the same flat plane, which needs the same inclination and the same RAAN. Two satellites with the same hoop and same speed never get closer or further; they just chase each other around. Two satellites in different planes only meet at the two points where their hoops cross each other, a fast flyby at several km/s, which is almost useless for sustained observation or contact. That is why plane-matching matters so much in a proximity operation.

Why changing plane is so expensive

To rotate a hoop you must push the satellite sideways at the point where the hoops cross, and the satellite is already travelling at 7.6 km/s. The fuel needed is about 2 x v x sin(angle/2). A 1 degree plane change at 550 km costs about 2 x 7.6 x 0.0087 = 0.13 km/s = 133 m/s. For comparison, a small satellite might carry a few hundred m/s in total. A 0.8 degree change (about 106 m/s) therefore eats a large fraction of a small satellite's budget. This is why CSIS calls such moves "unusual" and why the May manoeuvres were noticed. [ESTIMATE, standard formula]

The Earth does part of the work for free

The Earth is slightly fat around the equator. That bulge makes every orbit's plane slowly precess (turn about the polar axis). For sun-synchronous orbits this is arranged to be about 0.986 degrees per day, to keep pace with the Sun. The rate depends slightly on altitude and inclination. A satellite a little higher precesses a little slower. So if you want to move your plane relative to someone else's, you can either burn a lot of fuel directly or change your altitude and wait, letting the Earth's bulge do the rotation. [DOCUMENTED physics; the application is from Integrity]

This is exactly what the 30 September climb did. Integrity's account: by 30 Sep Kosmos 2612's plane had drifted about 0.24 degrees from ICEYE-X36's (about 28 km of cross-track separation at the widest point). A direct sideways burn to fix this would have cost about 31 m/s. Instead, 2612 climbed 46.5 km (about 25 m/s), which slowed its precession so that the planes walked back into alignment at about 0.02 degrees per day, then on 12 Oct it dropped back down (27.9 m/s) at exactly the moment the planes lined up. Total about 57 m/s versus 31 m/s for a direct fix, but the extra cost bought something the direct fix would not: a new position along the orbit too. [SINGLE-SOURCE (Integrity) for the account; the arithmetic is consistent with ESTIMATE below]

My check of those numbers. Changing a circular orbit's altitude by da costs roughly (v/2) x (da / a). With v = 7.6 km/s, a = 6,920 km and da = 46.5 km this is 3.8 x 0.0067 = 0.0255 km/s = 25 m/s, which matches. And a higher satellite laps a lower one: the along-track drift per orbit is about 3 x pi x da, so at da = 46.5 km this is about 440 km per orbit, or about 3.6 degrees per orbit; with about 15 orbits a day that is about 54 degrees per day, which matches Integrity's "about 55 degrees per day". [ESTIMATE; confirms the reported figures are physically coherent]

Relative motion: why "same altitude" matters

Two satellites on the same circular orbit stay the same distance apart indefinitely. If one is slightly lower, it travels slightly faster and gradually catches up with the other from behind; if slightly higher, it falls behind. The rule of thumb: along-track drift of about 9.4 km per orbit for each 1 km of altitude difference, or roughly 140 km per day. This is why a slightly lower satellite trailing by hundreds of kilometres will close the gap in days, and why it takes only a tiny burn to stop closing at the right moment. [ESTIMATE]

A caution about published rates. Integrity's 12 Oct summary says 2612 was about 1 km lower than ICEYE-X36 and, without further burns, would still be about 90 km behind on 8 Nov, starting from 475 km on 13 Oct. By my rule of thumb a 1 km altitude difference would close 475 km in about 3.4 days, not 26 days, so either the "1 km" refers to a different measure (semi-major axis versus altitude) or the real difference was nearer 0.1 km. I flag the mismatch rather than resolve it. [ESTIMATE; unresolved]

Natural-motion circumnavigation

When two satellites are on nearly identical orbits, one appears to circle slowly around the other without any burns. If it has exactly the same orbital period as its neighbour but a slightly different eccentricity, it traces a little ellipse around it, which takes one orbit (about 95 minutes) to complete. This gives a free "walk-around" view from every angle. Integrity notes 2612 and ICEYE-X36 both have eccentricity 0.002 and arguments of perigee differing by about 4 degrees (needing only about 0.25 m/s to match), so Russia "can hold any standoff range it chooses, including natural-motion circumnavigation". [SINGLE-SOURCE; the principle is DOCUMENTED textbook physics]

What a "close approach" is

Because everything moves at 7.6 km/s, "close" is relative. A conjunction between unrelated satellites means passing at several km/s, which is dangerous because a collision at that speed releases debris. A rendezvous and proximity operation (RPO) is slow and controlled: relative speed of a few m/s (McDowell's "20 km/h" is about 5.6 m/s). The fact that the relative speed was close to zero is what distinguishes deliberate rendezvous from a coincidental pass. [DOCUMENTED]

How close is too close? A University of Mississippi space-law professor quoted by Axios said there is no strict legal limit on how close satellites may get, but countries generally do not approach within about six miles (10 km), because at closer distances the risk of collision from either moving increases significantly. [SINGLE-SOURCE; one expert opinion]

ICEYE-X36 AND WHY RADAR SATELLITES MATTER

The satellite. ICEYE-X36 (COSPAR 2024-043F, NORAD 59103) is a small synthetic aperture radar (SAR) satellite of roughly 90 kg, launched in March 2024. It is registered to ICEYE US, the American subsidiary of ICEYE, a company founded in Finland. Its orbit is around 541 to 556 km at about 97.8 degrees (sources differ slightly: 541 km on one tracker, 555 to 556 km in Integrity's 12 Oct report, and CSIS and others give about 543 km; these differ because of different altitude definitions and different dates). [DOCUMENTED for identity; altitude conflict noted]

What SAR is, plainly. A radar satellite sends microwave pulses toward the ground and records the echoes. As the satellite moves along its orbit, it records many echoes of the same patch of ground from slightly different positions; computers combine them so the small antenna behaves like a very long one (the "synthetic aperture"), producing sharp images. Because radar supplies its own illumination and microwaves pass through cloud, it images day or night and through clouds. That is a huge advantage for military use in the Ukrainian climate, where optical satellites often cannot see the ground. [DOCUMENTED textbook physics]

Why it matters to Ukraine. In January 2026 ICEYE announced an agreement with "a customer within the Ministry of Defence of Ukraine" to expand cooperation on space-based intelligence using its constellation (as quoted by Axios). Radar images can show vehicles, ships, airfields, damage to infrastructure and changes in terrain. A Turkish defence site (Times of Defence) says this imagery supports "standoff precision targeting"; I treat that as UNVERIFIED commentary. A Ukrainian official said it was unclear whether this specific satellite supports Ukraine. [DOCUMENTED that ICEYE has a Ukrainian MoD customer; UNVERIFIED for X36 specifically]

A constellation, not one point of failure. ICEYE-X36 shares its orbital plane with X37 and X38, and Ukraine has access to the wider ICEYE constellation. Disabling X36 alone would not blind Ukraine, which suggests any hostile action against it would be at least as much about signalling as about military effect. [SINGLE-SOURCE via COMSPOC/America 2.0; the inference is SPECULATION]

Context: Russia's earlier warning. In October 2022 a Russian foreign ministry official, Konstantin Vorontsov, told a UN committee that commercial satellites supporting Western militaries could be "legitimate targets" for retaliation. Integrity ISR cites this as "intent signaled"; CSIS notes Russia has said it might consider commercial satellites supporting Ukraine to be legitimate military targets. [DOCUMENTED that the statement was made (via two secondary sources)]

WHAT IS SPECIAL ABOUT THIS GROUP

Every earlier Russian proximity-operations case (below) involved one satellite, or one parent and its released child. This cluster is different in several ways. All of the following are observations, with the interpretation labelled.

• Six satellites at once. Earlier cases were single launches or pairs. [DOCUMENTED]
• A shared, expensive plane change to match a non-Russian target. Five satellites each spent about 106 m/s to do so in May. Earlier Russian shadowing of US satellites involved one satellite choosing a matching plane. [DOCUMENTED]
• A "stack" then a "single pursuer". After the May events, the group dispersed in altitude and plane; then one satellite alone was brought back. That looks more like a pool of candidates, from which one is selected than a fixed pairing. [SPECULATION]
• No attempt to hold the match in the summer. Unlike a deliberate stalking effort, the satellites drifted. Hindsight reads this as "park in the area, wait for the right moment". [SPECULATION]
• Possible swarm or distributed role. If the six were designed to work as a set, one could imagine different units for different purposes (an imaging unit, a close inspector, a "relay"), or simply redundancy: if one fails, another can take over. An alternative reading is that they are six copies of the same bus with the same job and Russia is simply testing the same behaviour on several. [SPECULATION; no source states a swarm role]
• Fuel budgets. By Integrity's estimate, 2612 has spent about 175 m/s, versus 112 to 114 m/s for four siblings. All have spent more than a typical small satellite would for ordinary station-keeping. A satellite designed to burn this much fuel on manoeuvres presumably has a propulsion system sized for it. [ESTIMATE, based on Integrity's estimates]
• Next candidates. 2611 and 2613 are drifting toward the plane of ICEYE-X37, with crossings on about 6 and 15 November. If either repeats the climb-wait-descend pattern, it would confirm that the group is a programme with a repeatable technique rather than a one-off. [SINGLE-SOURCE; forecast]

RUSSIA'S STATED POSITION ON COMMERCIAL SATELLITES

What was said, and when. The best-known statement is by Konstantin Vorontsov, Deputy Director of the Russian Foreign Ministry's Department for Non-Proliferation and Arms Control and deputy head of Russia's delegation. The unofficial English text posted on the UN meetings site is headed as delivered at the First Committee's thematic discussion on outer space on 26 October 2022 (New York). Reuters, RFE/RL and others reported it on 27 October 2022, which is why many secondary sources give that date. The key passage, from the posted text: "We would like to specifically stress an extremely dangerous trend that goes beyond the harmless use of outer space technologies and has become apparent during the latest developments in Ukraine. Namely, the use by the United States and its allies of civilian, including commercial, infrastructure elements in outer space for military purposes. Apparently, these States do not realise that such actions in fact constitute indirect participation in military conflicts. Quasi-civilian infrastructure may become a legitimate target for retaliation." The text adds that the use of civilian satellites in this way is "questionable under the Outer Space Treaty" and "must be strongly condemned". [DOCUMENTED: UN statement text]
Wording differences. Reuters and RFE/RL render the key phrase as "a legitimate target for a retaliatory strike", a different translation of the same sentence. The text above is the unofficial UN-site translation. [DOCUMENTED]
An earlier version. Vorontsov used almost the same words at the UN open-ended working group on reducing space threats in Geneva on 12 September 2022 (Ars Technica and Space.com quote it; Ars suggested a reference to Starlink, but no company was named). [DOCUMENTED via press quoting the UN-site text]
Repeated in 2023. Vladimir Ermakov, head of the Foreign Ministry's non-proliferation and arms control department, repeated the point in an interview with Russian journalists reported by TASS and then Newsweek in October 2023, again without naming any company. [DOCUMENTED via Newsweek's report of TASS; I did not read the TASS original]
What I could not verify. I found no verified statements by Sergei Ryabkov or Maria Zakharova on this specific point, and none that names ICEYE. Starlink was widely assumed to be the target of the 2022 remarks, but no text names it. [UNVERIFIED if attributed to Ryabkov or Zakharova]
Meduza. Meduza's June 2026 piece headlines this as "'Legitimate targets,' Moscow said in 2022" and cites Vorontsov at the UN. It is a reference to the same statement. [DOCUMENTED]

Where the "nearly four years" passage comes from. The passage that ends "Nearly four years after making these statements, Russia is showing signs of carrying out this threat" is from the Integrity ISR analysis "Is Russia Maneuvering to Threaten an ICEYE Satellite?" (Greg Gillinger, 22 May 2026), in its Editor's Comments. It was reprinted in Integrity's ISR University newsletter (issue 146, 1 June 2026, and later issues). It is not CSIS text. Note that Integrity dates the speech 27 October 2022 (the reporting date; the delivery date was 26 October) and quotes Vorontsov accurately in the "quasi-civilian infrastructure may become a legitimate target for retaliation" form. [DOCUMENTED as to source]
Balance. A general threat against commercial space systems that support Ukraine is not evidence of intent against ICEYE-X36 specifically, and it is conditional ("may become"). Russia has not said anything about these six satellites. The statement shows a declared legal and political position; it does not show what the satellites are for. [Editorial caution]

Sources: UN statement text https://estatements.unmeetings.org/estatements/11.0010/20221026/5yPwCsESxyBr/N5pGP22K6MRm_en.pdf ; Russian Mission to the UN https://russiaun.ru/en/news/261022_v ; Reuters, 27 Oct 2022 https://www.reuters.com/world/russia-says-wests-commercial-satellites-could-be-targets-2022-10-27/ ; RFE/RL https://www.rferl.org/a/russia-western-commercial-satellites-legitimate-targets/32102888.html ; Ars Technica, Sept 2022 https://arstechnica.com/tech-policy/2022/09/russian-diplomat-suggests-attacks-on-satellites-in-possible-reference-to-starlink/ ; Space.com https://www.space.com/russia-private-satellites-legitimate-target-wartime-united-nations ; Newsweek on Ermakov https://www.newsweek.com/russia-shoot-western-satellites-ukraine-war-1835049 ; Integrity ISR https://integrityisr.com/is-russia-maneuvering-to-threaten-an-iceye-satellite/

THE INSPECTOR-SATELLITE HYPOTHESIS: ONE POSSIBILITY, LABELLED

The hypothesis [SPECULATION]. The six satellites are inspector or proximity-operations satellites, like Russia's earlier Nivelir-programme missions, and 2612 is practising or carrying out close approach to a commercial radar satellite.
What supports it. The costly plane change (about 105 to 106 m/s, Integrity ISR) is unusual for ordinary imaging satellites; 2612 then matched altitude and drifted to within a few km, or closer, of ICEYE-X36 (Langbroek, CSIS). Precedent [DOCUMENTED, Hendrickx, The Space Review]: Nivelir parent satellites (14F150) flew in the planes of US reconnaissance satellites (Kosmos 2542 with USA 245 in 2019 to 2020; Kosmos 2558 with USA 326 in 2022; Kosmos 2576 with USA 314 in 2024), and the Kosmos 2543 subsatellite released a fast object in July 2020 that the US called a likely weapon test.
Caveats. (1) Nivelir parents stayed tens of km from their targets, and the close approach and projectile release came from a small subsatellite, not from a satellite approaching a target directly; the 2026 satellites have no known subsatellite. (2) Those missions launched one or a few satellites; here six were launched together, five into one orbital family, which fits a constellation as well as an inspection mission. (3) The Volga stage and Soyuz-2.1b combination is unusual for Russian military launches, and Hendrickx said it may indicate a new type of satellite. (4) No index or contractor has been tied to these satellites, so Nivelir is a comparison, not an identification. (5) Hendrickx, McDowell and Langbroek each expressed doubts in May that the plane match was deliberate; the October rendezvous changes that picture but does not identify the purpose. (6) A commercial SAR satellite's design is public, which makes pure inspection a less obvious motive (Hendrickx, via Meduza).
Other possibilities: radar imaging constellation (Zak), jamming or electronic action, physical interference, or an exercise or demonstration. Observation so far fits several of these. [SPECULATION]

ANALYSIS: WHY THE FUEL BURNS ARGUE AGAINST A PURE IMAGING ROLE

Analysis, not reporting. Everything in this subsection is [SPECULATION] built on the documented manoeuvres.
What an imaging satellite wants. A radar or optical imager earns its keep by staying in a stable, predictable orbit with a steady view of the ground, and by saving its propellant for keeping that orbit over years. The ~105 to 106 m/s plane change in May (Integrity ISR) is a large fraction of what a small satellite typically carries, and it buys nothing for ground imaging: the orbit's inclination was already suitable. Close approaches to another satellite also cost fuel and take pointing time away from the ground. Spending that fuel shortens mission life, which is why plane changes in low Earth orbit are rare.
Hendrickx's point. Bart Hendrickx told Meduza that if you wanted to study ICEYE or affect it, it is hard to see why you would need five satellites, and that chance remained likely at that time (June). That argues against a single-target plan for the whole group. It also fits another reading: the group was built for a purpose other than ICEYE, and only some members are being used against it. [SINGLE-SOURCE, dated]
What a manoeuvrable inspector, decoy or weapon-type role would want. Propellant margin, precise navigation and willingness to burn fuel for position are the core requirements of an inspector, a close-in observer or a co-orbital weapon. The behaviour of 2612 (matching plane, altitude and eccentricity of its target, closing in steps) looks more like that role than like imaging. Analogy: a photographer stands still; a bodyguard or a pickpocket follows someone.
The six may not be one type. Kosmos 2609 was released separately into a different plane and has behaved differently, and the five others share an insertion family. Different satellites could carry different payloads, for example imagers, a decoy and an inspector. Nothing public confirms that they are identical.
2612 may be the special one. Integrity ISR estimates 2612 has spent about 175 m/s of delta-v since launch (as of 12 Oct, before its 15 and 16 Oct burns), compared with about 112 to 114 m/s for its siblings. The extra ~60 m/s is exactly the later series of burns that put it alone beside ICEYE-X36. That could mean it is a differently equipped satellite, or simply that it was chosen to carry out the approach. The data cannot separate those. [ESTIMATE (Integrity); interpretation SPECULATION]
Limits of the argument. Satellites can be built with large propellant tanks for any role, including imaging with agile orbit adjustment. A military imager might also practise proximity operations as a secondary task. The behaviour makes a pure imaging role less likely, not impossible, and does not tell us whether the purpose is inspection, interference or demonstration.

HISTORY: RUSSIAN CLUSTERS, SHADOWS AND "NESTING DOLLS"

Context from earlier Russian activity. Items from my background knowledge, rather than sources I re-read for this page, are marked [BACKGROUND]; please verify before publication.

Kosmos 2499 (2014) and the «Нивелир» / «Буревестник» lineage

Kosmos 2499 was launched on 23 May 2014 on a Rokot (as a co-passenger; earlier text here said December 2014, which was wrong) and carried out rendezvous and proximity operations with its Briz-KM upper stage. Hendrickx's table lists Kosmos 2491 (25 Dec 2013), 2499 (23 May 2014) and 2504 (31 Mar 2015) with the presumed military index 14F153. Bart Hendrickx and others have described it as a small manoeuvrable satellite associated with the «Буревестник» (Burevestnik) experimental work and linked it to the programme of inspector satellites now called «Нивелир» (Nivelir). Other Nivelir-type satellites (the Kosmos 2491, 2504, 2519, 2521, 2523 family) followed. [BACKGROUND; Hendrickx's reporting. Not to be confused with the Burevestnik nuclear cruise missile; see the separate Burevestnik article, which concerns the missile and not these satellites.]

Do any index numbers or names point at Kosmos 2609 to 2614? As of 18 Oct 2026, no public source gives a GRAU/industry index, programme name, designer or contractor for these six satellites. Russia disclosed only "spacecraft in the interests of the Ministry of Defence" (launch statement and success confirmation, 17 Apr 2026). Gunter's Space Page lists the operator, contractors and type as blank, and describes them as "small military satellites of unknown purpose". [DOCUMENTED that nothing is public]
For comparison, the Nivelir indices are known from Hendrickx's reporting: the Nivelir project is 14K167/14K245; parent satellites are 14F150 (NPO Lavochkin, Karat-200 bus, optical telescope, flown on Soyuz-2.1v/Volga except Kosmos 2576 on a 2.1b); the subsatellites are presumably 14F162 (TsNIIKhM); Kosmos 2491/2499/2504 and 2536 are 14F153 (possibly Burevestnik-related); Kosmos 2535 is 14F157. Prime contractor TsNIIKhM, chief designer Vladimir Verkhoturov. The earlier Nivelir parents were each placed in the plane of a particular US reconnaissance satellite and stayed tens of km away; they are 14F150-type satellites of unlike mass and number. Nothing links these indices to the 2026 cluster. [DOCUMENTED for the lineage; the link to 2609-2614 is UNVERIFIED]
Differences from Nivelir: six satellites in one launch, a Soyuz-2.1b with Volga (Hendrickx told Meduza the combination is unusual and may mean a new type), and several satellites sharing one target plane. Hendrickx also told Meduza it is likely the satellites ended up in the ICEYE plane by chance, asking why anyone would need five satellites to study one. McDowell and Langbroek called the alignment coincidence plausible in May; later events (the 2612 rendezvous) post-date those remarks. [SINGLE-SOURCE (Meduza, 9 Jun 2026), dated]
Candidate identities, as of now: NPO Mashinostroyeniya radar satellites ("Diamond-X", Zak) [SINGLE-SOURCE / SPECULATION]; Nivelir-type or a new inspector type [SPECULATION; attributions to McDowell/Vaclavik are second-hand]; Okulus [ruled out by Hendrickx, see above]. Meduza\'s phrase "hunter satellites" and its "inspector" pattern comment are journalism, not identification. Sources: https://meduza.io/en/cards/four-russian-military-satellites-came-within-13-kilometers-of-a-radar-satellite-operated-by-a-finnish-company-that-supplies-ukraine-with-battlefield-imagery-legitimate-targets-moscow-said-in-2022 ; https://www.thespacereview.com/article/4979/1 ; https://space.skyrocket.de/doc_sdat/kosmos-2609.htm

Kosmos 2519 / 2521 / 2523 (2017): the "nesting doll"

In 2017 Kosmos 2519 released a small satellite (2521), which in turn released a smaller object (2523) that moved away at speed. A senior US official said it "exhibited characteristics of a weapon system". CSIS repeats the "nesting doll" phrase. [DOCUMENTED via CSIS; the numbering is BACKGROUND]

Kosmos 2542 and 2543 (launched 2019; events 2020)

Kosmos 2542 was launched in November 2019 and shadowed the US reconnaissance satellite USA 245 in early 2020, at distances of a few hundred km down to tens of km; the US Space Command commander publicly called this "unusual and disturbing" and described the behaviour as "dangerous". Kosmos 2543 was released by 2542 in 2020 and in July 2020 released an object which travelled at up to 186 m/s relative to the parent. The US and UK condemned this as a test of a space-based weapon system, a characterisation Russia rejected. [DOCUMENTED re 186 m/s via CSIS; the shadowing of USA 245 is BACKGROUND]

Kosmos 2558 (2022)

Launched August 2022 into the plane of a US government satellite (USA 326), it was described later by the US Space Command as "evident" that it was a "space weapon" per CSIS. [DOCUMENTED that CSIS says the commander said so in 2026; the target identity is BACKGROUND]

Kosmos 2576, 2588, 2589/2590

Kosmos 2576 (May 2024) was deployed into the same orbit as a US government satellite; the US told the UN Security Council it was "likely a counterspace weapon" (as quoted by America 2.0, citing Secure World Foundation). Kosmos 2588 (2025) pursued the US satellite USA 338; Langbroek called it "the fourth time in five years" a Russian satellite had parked in the orbit of a US reconnaissance satellite, and none ended in an attack. Kosmos 2589 and its sub-satellite 2590 (launched June 2025 on Angara-5/Briz-M into a high elliptical orbit) have spent months in rendezvous activity at geostationary altitude, and in late September 2026 2589 parked beside the Russian Blagovest communications satellite Kosmos 2526 (Integrity ISR). [DOCUMENTED via the cited secondary sources]

Luch / Olymp-K

Luch (Olymp-K) satellites in geostationary orbit, launched 2014 and 2023, have repeatedly drifted next to commercial and military communications satellites of other nations; they are widely described as signals-intelligence or inspection satellites. See also the Luch MKSR article, which notes that Olymp-K borrowed the Luch name but is a different programme from the relay satellites. [BACKGROUND; DOCUMENTED on the site page for the naming point]

The ground-launched and nuclear threads (not the same thing, but they colour how the West reads this)

Nudol direct-ascent ASAT. In November 2021 Russia destroyed its own old satellite Kosmos 1408 at about 480 km with a ground-launched missile (the Nudol system), creating more than 1,500 pieces of trackable debris; nearly two-thirds of the tracked pieces fell out of orbit within a year per SpaceNews as quoted by America 2.0, the rest may take over a decade. [DOCUMENTED]

Kosmos 2553 and the nuclear ASAT concern. In February 2024 US officials described a Russian development of a nuclear anti-satellite capability connected with Kosmos 2553, launched in 2022 into a high orbit around 2,000 km. The White House said it was "not an active capability that's been deployed", and the Secure World Foundation review says the Russian nuclear ASAT "has not been deployed nor tested". It is a separate, indiscriminate-weapon concern from Cosmos 2612, which is at a low altitude where a nuclear blast would also harm Russian satellites. [DOCUMENTED via cited secondary sources; Kosmos 2553 is not an EKS satellite]

Where this cluster differs from all of these. Per CSIS: "no public information indicates [Russian satellites] have done so near non-Russian satellites" in this manner, i.e. a rendezvous with a satellite not belonging to a government, and not Russian. McDowell's "first LEO rendezvous of two active satellites where the owner of one did not give approval" is a narrow but defensible claim. The commonly stated "first of its kind threat" is only fair in that narrow sense: Russian satellites have shadowed US government satellites before (for several years), but not a commercial satellite supporting a country at war with Russia. [SINGLE-SOURCE / DOCUMENTED]

POSSIBLE PURPOSES (every one SPECULATION, with indicators)

Nothing below is known. Each hypothesis lists what would be consistent with it and what would tell it apart from the others.

1. Inspection / close-up intelligence collection

• Idea: use cameras, RF receivers or radar sensors to learn what the target is (bus design, antenna, orientation, how it is operated, whether it manoeuvres). [SPECULATION]
• Consistent with: a gentle standoff approach, natural-motion circumnavigation, lighting-aware timing. Integrity noted the Cosmos 2589 case had poor lighting, hinting that "lighting" is a clue for optical inspection.
• What would tell: the satellite holding station at a few hundred metres to a few km in good sunlight, circling, then leaving; no release; no change in ICEYE-X36 behaviour.

2. Signals collection (listening to the radar's downlink and uplink)

• Idea: a satellite close to ICEYE-X36 could pick up its radar emissions and downlinks, learning its modes, pulse patterns and timing, which is valuable to any force that wishes to avoid being imaged. [SPECULATION]
• Consistent with: a long loiter in the same plane; no need to be very close.
• What would tell: hard to distinguish from the outside; a long stay without a close pass would favour this.

3. Intimidation, harassment and fuel-burning

• Idea: force ICEYE to waste its small propellant supply on avoidance manoeuvres, shortening its life, and send a political message. [SPECULATION]
• Consistent with: a close, persistent shadow; repeated returns; ICEYE manoeuvres that it would not otherwise make.
• What would tell: an ICEYE avoidance burn (not seen as of the sources I found; last detectable manoeuvre June).

4. Radio-frequency jamming or interference

• Idea: broadcast noise or false signals on the satellite's command or downlink frequencies, or into its antennas, to degrade imaging, downlinking or control. Closeness raises the power received, so a small jammer can have a big effect. [SPECULATION]
• Consistent with: ICEYE outages with no orbital change; reports from Ukraine of missing imagery.
• What would tell: an ICEYE or Ukrainian statement, outages timed with passes. Attribution would be difficult. Langbroek's comment that "attribution is very difficult in space" applies directly.

5. Kinetic or physical action (collision, projectile, grapple, or "nesting doll" release)

• Idea: ram, release fast fragments or a small projectile (as Kosmos 2543 released an object at up to 186 m/s), or grab the satellite. [SPECULATION; the analogies are DOCUMENTED but not applied to this case in any source]
• Consistent with: getting to within metres at low relative speed, then a new catalogued object, or ICEYE-X36 breaking up, tumbling, or changing orbit.
• What would tell: a new object in the catalogue from the April launch; a debris cloud; a sudden change in ICEYE's orbit. Langbroek's 18 Oct note says no sign of disturbance as of the elements he used. Debris at about 543 km would last years and would threaten Russia's own satellites and its planned Rassvet constellation (see Rassvet LEO), a built-in disincentive. A claim by Russian commentators that a break-up would cause "Kessler syndrome" is, in the America 2.0 summary, considered unlikely at this scale because it needs a decades-long chain of collisions.

6. Demonstration, signalling and deterrence-testing

• Idea: show the West that Russia can place a satellite alongside a commercial satellite supporting Ukraine, without doing anything harmful, to test the response. CSIS frames the approach within "gray zone" activities and suggests Moscow "may be willing to take that risk, believing the West will not respond forcefully". [SPECULATION; CSIS's analysis]
• Consistent with: a pass, then departure (the pattern reported on 18 Oct).
• What would tell: the satellite leaves; Russian officials comment; the same pattern against other satellites (X37 in November).

7. Test or training: a rehearsal for a later operation

• Idea: any new RPO capability needs practice on real, uncooperative targets, such as practising approaches, navigation, and relative-motion control. [SPECULATION]
• Consistent with: repeated approaches by different units; the group's staggered behaviour.
• What would tell: more events involving different group members and targets.

Important: these are not exclusive. A single approach can serve several purposes: inspection and signalling and rehearsal. And the "no harm" outcome so far is consistent with all of them, which is why analysts say they cannot tell. [SPECULATION]

LEGAL AND POLICY CONTEXT

• Outer Space Treaty (1967). Bans nuclear weapons and weapons of mass destruction in orbit, and says states bear international responsibility for national space activities, including those of non-governmental entities, and keep jurisdiction over objects they register. ICEYE-X36 is registered in the US, so the US is responsible for it; Russia's satellite is registered to Russia. It does not ban conventional proximity operations. McDowell does not think the approach "quite violates the Outer Space Treaty". [DOCUMENTED via the cited sources' summary of the treaty]
• "Due regard" and harmful interference. The treaty requires parties to act with "due regard" for others and avoid "harmful interference". Whether a close shadow is harmful interference is exactly the unresolved legal question. [DOCUMENTED text; application is open]
• Commercial satellites as targets. International law of armed conflict treats objects that make an effective contribution to military action as potentially lawful targets even if civilian-owned (dual-use). Russia has said exactly this about commercial satellites supporting Ukraine. Many Western lawyers dispute how far that goes and warn about effects on neutral users. [BACKGROUND; DOCUMENTED that Russia made the statement]
• NATO. NATO has said attacks "to, from, or within space" could lead to Article 5 being invoked, decided case by case. ICEYE-X36 is a US-registered commercial satellite of a company founded in Finland (a NATO member since 2023). Whether an attack on it would count is untested. [SINGLE-SOURCE via America 2.0]
• US posture. In September 2026 US Air Force Secretary Troy Meink said the US "now has on-orbit space control weapons capable of defending the joint force". Moscow responded by renewing a call for a treaty banning weapons in space. Russia and China have proposed such a treaty since 2008; many states reject it as unverifiable, and it would not cover ground-launched missiles. [SINGLE-SOURCE via Washington Times and America 2.0]
• No "rules of the road". No binding rule sets a minimum distance for proximity operations or requires consent. This is the reason the event is unusual: it is arguably within the letter of existing rules and outside their spirit. [SPECULATION/analysis]

WHAT TO WATCH FOR

• Whether 2612 returns after the 18 Oct drift away, especially near 24 October. [forecast]
• New catalogued objects from 2026-083 (any letter beyond K, or new NORAD numbers). A release would show up this way.
• An ICEYE manoeuvre or statement, or any word from Ukraine, US Space Command, the Pentagon, Finland or NATO.
• Imagery gaps from ICEYE customers, which would hint at interference.
• The siblings: 2614's fast crossing near ICEYE-X36 about 21 October; 2611 and 2613 toward ICEYE-X37 about 6 and 15 November; any new climb-and-descend sequence.
• Fuel: additional burns by 2612 in the next days; independent delta-v accounting (Integrity uses CelesTrak data).
• Better data: radar-tracker (LeoLabs) or government statements that resolve the "few hundred metres versus few km" discrepancy.
• Russian official comment, which so far (as of the sources I found) has not addressed the satellites.

OPEN QUESTIONS

• What is the mission of each satellite, and are they identical?
• Why is 2609 in a different plane, and did it take part in the May manoeuvres? Which satellite made the 2.5 km pass?
• Why were the planes not held after May?
• Why was ICEYE-X36 chosen, rather than, say, X37 or another sensor satellite? Was it a convenient target, or specifically the one used by Ukraine?
• Is any of the six a Novy Kosmos demonstrator? (Hendrickx says the named Okulus-01/-02 are Chinese-built and not on this launch; a separate home-built demonstrator among the six is a remote possibility only.)
• What did the 18 Oct small burn achieve, and where is 2612 now?
• Does the cluster carry released sub-satellites or other deployable items?
• Will ICEYE-X36 keep working and will its users say so?

SOURCE CONFLICTS (summary)

• Which satellites made the May plane change: five of six, including 2609 (CSIS) versus 2610 to 2614 only (Integrity, Skyrocket). Integrity's May report speaks of four, with the fifth still in progress.
• Who passed at 2.5 km on 26 May: Cosmos 2609 (CSIS). Inconsistent with 2609 not changing plane.
• Altitude of the pair: about 543 km (CSIS and others), 541 km (orbitalradar), 555 to 556 km (Integrity, 12 Oct). Different dates and definitions.
• Closest distance: a few km (McDowell, 16 Oct), a few hundred metres or less (Langbroek, 18 Oct), 26 m (orbitalradar prediction), "450 feet" or "85 feet" (Washington Times, quoting predictions).
• Current status: orbitalradar predicts a 26 m pass on 19 Oct; America 2.0 reports a drift to 12 km on 18 Oct. Later data win.
• Number of objects launched: 6 payloads (US catalogue) versus 8 satellites (RussianSpaceWeb headline) versus 9 to 10 objects (flightatlas).
• Inclination figures: 96.95 (launch), about 97.0 to 97.8 (May), 97.85 to 97.90 (Oct). Consistent once the May and October changes are accounted for.
• "First flight" of Soyuz-2.1b/Volga: flightatlas says first and only; I could not confirm.

SUGGESTED SUB-ARTICLES

• Rendezvous and proximity operations explained: relative motion, natural-motion circumnavigation, safe ellipses, and why 5 m/s is a "hold".
• Nivelir / Burevestnik satellite lineage: Nivelir is now covered in its own article (Kosmos 2491, 2499, 2504, 2519/2521/2523, 2542/2543, 2558, 2576, 2588); a separate Burevestnik (satellite interceptor) article is still open.
• Kosmos 2589 / 2590 and GEO proximity operations, and Luch/Olymp-K.
• ICEYE and commercial SAR for war: how SAR works, revisit rates, and what Ukraine uses commercial imagery for.
• Counterspace weapons taxonomy: kinetic, non-kinetic, cyber, RF, directed energy; and what each looks like from the outside.
• Space law for conflict: Outer Space Treaty, due regard, dual-use.
• Volga and Soyuz-2.1b multi-payload launches: space tug mechanics.
• Novy Kosmos / Okulus and the Russian SAR microsatellite sector (NPO Mashinostroyeniya).

SOURCES AND HOW THEY WERE USED

• CSIS Aerospace Security Project, Clayton Swope and Chantal Li, "Could Russia Be Poised to Use an On-Orbit Weapon Against a Commercial Satellite?", 15 October 2026. https://www.csis.org/analysis/could-russia-be-poised-use-orbit-weapon-against-commercial-satellite (read directly).
• Integrity ISR, "The integrity flash", Issue 155, 12/14 October 2026 (Cosmos 2612 manoeuvre analysis, delta-v estimates). https://isruniversity.com/2026/10/12/issue-155/ (read directly; contains the figures labelled SINGLE-SOURCE above).
• Marco Langbroek, SatTrackCam Leiden blog, "[UPDATED] KOSMOS 2612 and ICEYE-X36: a worrisome cat-and-mouse-game in Space", 17 and 18 October 2026. https://sattrackcam.blogspot.com/2026/10/kosmos-2612-and-iceye-x36-worrisome-cat.html (read directly).
• The Hindu, Vasudevan Mukunth, 18 October 2026. https://www.thehindu.com/sci-tech/science/russian-military-satellite-stalks-us-radar-satellite-in-orbit/article71599219.ece (search extract read).
• Axios, Thomas Novelly, 17 October 2026 (LeoLabs, Ukrainian official, Space Force May remark, legal commentary). https://www.axios.com/2026/10/17/russia-ukraine-satellite-space (read directly).
• The Washington Times, Ben Wolfgang, 16 October 2026 (orbitalradar prediction, Meink remark; some wording questionable as noted). https://www.washingtontimes.com/news/2026/oct/16/russia-use-space-weapon-us-finnish-radar-satellite/
• Anatoly Zak, RussianSpaceWeb, "Soyuz rocket launches a military payload" (launch, catalogue tables, May reporting; part of the page not readable). https://www.russianspaceweb.com/spacecraft-military-soyuz2-volga-2026-0417.html
• russianforces.org (Pavel Podvig), "Launch of multiple military satellites from Plesetsk", April 2026.
• Gunter's Space Page / Skyrocket, "Kosmos 2609, ..., 2614" (summary, which names 2610 to 2614 as the manoeuvring group).
• America 2.0, "Space War Watch" (aggregator; many X posts by McDowell, Langbroek, Vaclavik quoted second-hand; data computed from CelesTrak) https://spacewar.america2.news/ , read 18 Oct 2026 about 22:12 UTC version.
• orbitalradar.com conjunction page 59103-68762 (predicted values; unreliable for the current state).
• Times of Defence (18 Oct 2026) (restates Integrity; contains unverified commentary).
• Satcat.com listing was not separately read for this page.
• Background items (marked [BACKGROUND]): Hendrickx on Nivelir; Kosmos 2499, 2519/2521/2523, 2542/2543, 2558, 2576, 2588 summaries; Kosmos 1408 test; Kosmos 2553 concern. Verify against primary reporting (Hendrickx in The Space Review, US Space Command statements, Secure World Foundation's Global Counterspace report) before publication.
• Earlier Project Orlan articles on the pages linked below.

RELATED ARTICLES

• Nivelir (14K167 / 14K245) inspector satellites (the documented Russian parent-and-subsatellite programme; no link to this cluster is documented)
• EKS Kupol / Tundra (the Russian missile-warning satellites; unrelated mission, similar tracking-based sleuthing)
• Russian command, control and communications
• A-135 / A-235 Nudol (ground-launched ASAT lineage)
• Russian Space Surveillance System (Okno / Nurek / Sanglok)
• Luch MKSR
• Rassvet LEO
• GLONASS
• AESA radar