Weekly review, week to 19 July 2026

Key structural shifts

Autonomous terminal AI guidance has moved from R&D into serial production. This is the key axis of the week, pulling scattered signals into a single line. Starting with an analysis of the pairing of the VYRIY FPV platform + the TFL-1 optical terminal guidance module (The Fourth Law) at a distance of ~110 km, the trend materialized into a mass weapon by the end of the week: the arrival in Ukraine of the first batch out of 50,000 Shrike UAVs with Auterion Strike kits (the Skynode S module) and, most importantly, an analysis of the European AI kamikaze HX-2 (Helsing), which, as Russian rear-area personnel admit, has been encountered in "practically all rear areas" source. Symmetrically, terminal guidance has been fitted to the Ukrainian FP-1/FP-2 and the Russian "Geran-4 Seeker". This nullifies all defenses built on the logic of "we'll jam the control channel".

Satellite communications on the line of contact have become the main target of attack — and remain invulnerable to EW. A two-sided fork has formed: on the one hand, an escalation of "weapons against communications" (Chinese HPM systems against Starlink, Iran's declaration of Starlink infrastructure a legitimate military target, passive direction-finding of terminals); on the other, Starlink V5 (rev5_pez_prod2) dropping the GNSS receiver, which makes terminal operation independent of navigation suppression by EW source. Civilian satellite internet no longer exists in this war for either side.

Maritime drone warfare has become a strategic factor relieving pressure on the front. The USF operation "MoLoChKa" hit 172 vessels of Russia's shadow fleet in 13 days, collapsing traffic in the Sea of Azov by 78% (from 132 to 29 vessels according to SAR imagery). The critical operational effect: Russia is forced to pull up to 200 crews of the "Rubikon" center and the 51st Air Defense Division off the line of contact to protect shipping source — that is, the maritime campaign directly draws the best Russian anti-drone component away from the front line.

Technology dynamics and trends

The rising vector: AI terminal guidance and swarms. The qualitative transition of the week — from experiment to serial production. Auterion has already operated 22 drones under unified coordinated control (swarm control), with the responsibility formula "the soldier selects the target, the software automates execution after the decision". [TECH] The HX-2 represents the reference architecture of a serial autonomous kamikaze: an X-wing layout + quadcopter propellers, weight 12 kg, speed up to 220 km/h, range up to 100 km beyond line of sight, edge AI with an EO/IR gimbal, autonomous search-reidentification-attack upon loss of communications (claimed "EW-immune"), swarming via the Altra platform, production of >1000 units/month at the "Resilience Factory", price ~$30,000 source. Telling is the honest figure for real-world accuracy — ~36% in combat versus the claimed 100% in tests: autonomy works, but not perfectly yet.

The scaling vector: jet-powered "Gerans" are breaking air defense. [TECH] The direct consequence in figures: according to Odesa monitoring, the interception rate fell from 90.6% to 83% with the transition to jet-powered drones flying at 300–500 km/h. The "Geran-3" is a tactical variant with electro-optics; the "Geran-4" is effectively a new design with different aerodynamics, almost entirely in the "Seeker" variant with machine vision, a jam-resistant command-and-telemetry channel, relaying through relay drones without "Rassvet" satellite terminals, and an optical channel against GPS spoofing source. It is the "Seekers" that have become the main tool for hunting ships in ports. The Russian attempt at an uncrewed interceptor based on the "Geran" with R-60 missiles is degrading: [TECH] the R-60, with an uncooled seeker, a ±12° field and production ended 40 years ago, gives "near-zero" effectiveness; the R-73 (±45°/±60°) slightly improves the odds, but the carrier is slow and lacks search systems source.

The interception vector: C-UAS is entering serial production on the defender's economics. The "Shahed vs. interceptor" race has accelerated on both sides. Ukraine has codified the "Tsyklop" (two modifications, wingspan <1 m, ~1.5 h endurance, V2 — >100 kills, the first "Shahed" over the Kursk region) and brought the "Kybchyk"/Kestrel to market (up to 420 km/h, ceiling 6800 m, [TECH] a laser proximity sensor for standoff detonation of the warhead). Russia responds with the AI interceptors "Khant" (a neural network predicting maneuvers) and [TECH] "Viverna-Kogot" (YOLO detector, Hailo AI accelerator of up to 26 TOPS on a Raspberry Pi 5, onboard processing). The economics have shifted in favor of the defense: the Russian "Perun" combat laser has a cost of use of ~205 rubles per 10-second shot and a range of up to 5 km; [TECH] the "Tsitadel" anti-aircraft gun system with programmable detonation of 30-mm rounds spends ~3 rounds per drone. At the same time, at the deputy prime minister level (Manturov's meeting on interceptor drones), Russia acknowledges a systemic lag in mass interceptors.

The GPS-free navigation vector. It has taken shape as a separate line of bypassing EW. [TECH] The myth of "EW protection" via L5 has been debunked: GPS L5 / Galileo E5a / BeiDou B2a sit on the same frequency of 1176.45 MHz — a single wideband jammer jams all three; without CRPA and deep INS integration, it is an ordinary civilian receiver source. Working alternatives: SineLink/SNS with multilateration over a network of beacon-relays (beacons are already being found on pine trees in Russia's border areas), inertial-optical odometry on the Hornet DE-2, and an updated MIT Pushbroom Stereo (navigation by a single "slice" 10 m ahead without a satellite).

The carrier vector: USV → UGV. The world's first landing of an armed UGV (a robot of the Rys/Roboneers family, 7.62 mm machine gun, AI target detection) from a USV on the Kinburn Spit, simultaneously with the first combat use by the US of Corsair strike USVs (Saronic) against Iran — both events in one cycle. [TECH] Corsair: 7.3 m, warhead 453 kg, range >1850 km, control via Starlink. This turns the USV from a kamikaze/scout into a multi-platform carrier.

Weak signals (Under the radar)

Airborne EW based on the "Baba Yaga" against the DJI Mavic — a quiet systemic solution. [TECH] A directional patch antenna (2.4 or 5.8 GHz, dual-polarized), Chinese jamming units and large fanless heatsinks (to save weight) are mounted on the heavy Ukrainian hexacopter. The effect is qualitative: detection of Russian "Mavics" at 8–10 km and suppression 3–10 times farther than from the ground — within a month one unit lost a significant number of "Mavics" source. Moving EW into the air radically changes the geometry of suppression, and Russian channels are already calling for the approach to be copied urgently, but "the question of carriers is open".

The Phantom Twist stealth drone (Northwestern). Rotating the body at up to 25 rev/s makes the vehicle visually blurred and ~10 times less noticeable — a direct answer to ubiquitous machine vision and AI recognition. Given that optical detection is becoming the main C-UAS channel against drones without a radio link, this is an early prototype of a whole class of "anti-detection" form factors.

The idea of a maritime "short-range air defense". [TECH] A proposal to deploy USVs carrying interceptor drones in the coastal zone to intercept "Shaheds" over the sea on the approach to Black Sea ports source. Russia already has elements of naval anti-aircraft defense, Ukraine does not; closing this niche is a logical continuation of the "USV carrier → striker" pairing.

A sovereign edge-AI autopilot. The Russian student OS "Dronkam" on Orange Pi/RK3588S with an onboard LLM, autonomous GPS-free navigation (SLAM, ArUco) and voice programming of missions on Qwen2.5-Coder — an early signal of a domestic edge-AI autopilot emerging, so far unnoticed.

Vector projection and implications

Confirmed dynamics (high confidence):

• Terminal EW is losing value; demand for kinetics and optics is growing. Against a drone that needs neither a satellite nor an operator link (HX-2, TFL-1, "Seekers"), jamming is useless. The required countermeasures are shifting to early detection + an autonomous interceptor drone, optical/thermal guidance of interceptors and cheap "last line" kinetics (anti-drone cartridges, shotguns, anti-aircraft gun systems). Russian sources themselves note: the only answer to the HX-2 is early detection and an autonomous interceptor source.

• The "swarm vs. anti-swarm" race will accelerate within weeks. With the arrival of 50,000 Shrike/Auterion drones, swarm attack is entering its combat phase; KTsPN is already calling for countermeasures to be sought "proactively". Expect scaling of lasers ("Perun", HEL Rheinmetall+MBDA, the American 500-kW JLWS), electromagnetic catapults for mass launch of interceptors (Hermes) and programmable close-in munitions.

• Satellite communications will remain resistant to EW; the fight will shift to SIGINT direction-finding and kinetics against orbit. Starlink V5 without GNSS removes the dependence on navigation suppression; the realistic response is not "burning the orbit" (arithmetically doubtful because of the inverse-square law) but passive signature direction-finding of terminals on the Ku uplink (14–14.5 GHz) with TDOA/FDOA localization followed by a kinetic strike. The practical countermeasure for both sides is emission control discipline (EMCON), which has gone from "desirable" to mandatory.

Analytical hypothesis (moderate confidence):

• C-UAS will become a mandatory element of any ground platform. Ukraine's experience is already reshaping Western doctrines: in the Sandhills project the US Army is integrating turret-mounted shotgun systems (Edda) directly onto autonomous breaching vehicles after losing Lancer UGVs to FPVs; Britain is withdrawing tanks from Estonia in favor of light anti-tank forces with drones. The vector — building autonomous C-UAS into UGVs, USVs and armored vehicles as a standard module.

• The "domestic rear" is becoming a systemic target, which will trigger a race to protect civilian infrastructure. The change in the target profile of strikes on Russia (marketplace logistics hubs, filling stations, combined heat and power plants, power lines, communication nodes), with a declared list of targets "for months" and Russia's promised mirror strike on Ukrainian energy in winter, foreshadows mass demand for cheap stationary C-UAS based on machine vision, AEW aerostats (a solution to the radio horizon for ground radars, which are blind below 50 m) and physical barriers ("Pautina", gabions).

• Fedorov's resignation is a real risk factor for the Ukrainian tech loop. The open glee of Russian Z-channels serves as an indirect indicator of which areas (restricting Starlink for Russia, drones, AI, deep strike) inflicted the most damage. A testable hypothesis for the coming weeks is a possible pause/reassembly of the AFU's "digital wing"; at the same time, the creation of a miltech-czar post (Beskrestnov, "Flash", as presidential adviser on defense technologies) may soften the effect.

External analysis (Western sources)

• Ukraine's experience is directly rewriting NATO doctrine on armor and C-UAS. Britain is withdrawing tanks from Estonia, replacing them with light anti-tank forces with UAVs following the Hedgehog 2025 exercise, where Ukrainian operators demonstrated the effectiveness of drones source; in the Sandhills 2.0 project the US Army is putting Edda turret-mounted shotgun modules (defeating ~18 cm drones at 10–100 m, acoustic and radar sensors) directly onto autonomous breaching vehicles source. The Telegram loop discusses this as a "line-of-contact specific" — the West is already institutionalizing it.

• Autonomous AI-enabled EW closes the "detect-decide-act" loop without an operator. L3Harris and Shield AI conducted the first flight test of the DiSCO ecosystem with Hivemind software on the Green Wolf UAV: the system itself detected and classified EM threats and rerouted other drones through safe corridors in real time source. This complements the Ukrainian airborne EW theme with a move toward autonomous spectrum management by a swarm.

• Airdropping USVs — a new quality of maritime robotics. The Royal Air Force, in a world first, parachuted the autonomous K3 SCOUT boat from an A400M into a stormy sea (Project Beehive) source. This is the next step beyond the Black Sea "USV carrier" logic: rapid delivery of naval drones from the air to hard-to-reach areas.

• The Western response to swarms is navigation suppression, not just kinetics. IAI unveiled the non-lethal HYPNOSIS system for disrupting drone swarms by distorting navigation signals source — but, as Starlink V5 and GPS-free optical navigation show, this line of countermeasures is being devalued by exactly the same trends that the Russian-Ukrainian loop records.

• Western analysis confirms the effect of the maritime campaign on the front. Kyiv Independent, citing the USF commander, notes: protecting the shadow fleet is forcing Russia to transfer resources from combat sectors source — independent confirmation of the thesis of the maritime campaign as a strategic trade-off in Ukraine's favor.

• The crisis of Western artillery metallurgy as a driver of "dronization". The failure of 155-mm production in the US (the GD-OTS plant has not produced a single usable blank; instead of 100,000 shells/month — 36,000) explains why mass cheap drones are becoming a substitute for artillery — this structural context is barely reflected in the Telegram loop.