Weekly review, week to 26 July 2026

Key structural shifts

The erasure of the concept of the "rear" across the entire depth of European Russia. The week formalized the transition from pinpoint long-range strikes to a systemic war of attrition against the economy. The maritime campaign of the Unmanned Systems Forces (USF), "MoLoChKa", has reached a strategic effect — from July 6 to 22, ~196 shadow fleet vessels were hit, shipping in the Sea of Azov fell by 83%, the Kerch ferry crossing lost 75% of its throughput, and Maersk suspended calls at Chornomorsk, redirecting cargo to Constanța. This is the first measurable operational result of economic coercion. At the same time, an "infrastructure war" against civilian logistics has taken shape: strikes on Wildberries/Ozon hubs (77% of the Russian online market, ~$4.3 billion in damage, ~13% of warehouses) hit the hyper-concentrated Moscow agglomeration, where ~50% of the country's warehouse capacity is concentrated source. The AFU's declared strategy is to bring range to 4000 km in 2026 and 5000–10000 in 2027 source, covering even Siberia, where "nothing has ever flown in" source.

The institutionalization of Ukraine as NATO's R&D laboratory. The German-Ukrainian alliance (2026 plan — 11.5 billion euros, the Brave Germany program with 50/50 co-financing) has entrenched the "Test in Ukraine" model as a mandatory standard of combat validation. Transfer has gone both ways: Magura → US shipyards (ReconCraft), Ukrainian UAVs → the Pentagon, engines with the Czech PBS Group, licenses for the L119 and Aster 30, four Rheinmetall plants and three Quantum Systems joint ventures. Ukraine has turned from a recipient into a supplier of combat experience, licenses and technologies.

A change in the AFU's military leadership in line with technological priorities. Syrskyi → Drapatyi (commander-in-chief), Hnatov → Skybiuk (General Staff). The officially declared priorities of the new command are scaling deep strike and middle strike, UGVs (unmanned ground robotic systems) and asymmetric solutions, and technological superiority. The reshuffle institutionalizes an already ongoing process as a deliberate course.

Technology dynamics and trends

The key qualitative transition of the week is from the "cheap FPV vs. expensive interceptor" race to a race of speed and autonomy, in which both sides arrived at the same engineering solutions almost simultaneously.

The jet-powered threat vector has become dominant. [TECH] The mass use of jet-powered "Gerans" (Geran-3/Shahed-238) with a speed of >300 km/h and a cruising altitude of 2000–3500 m has devalued machine-gun mobile fire groups, small-caliber anti-aircraft artillery and electric propeller-driven interceptors — they simply cannot catch the target source. The response has been a serial wave of high-speed turbojet interceptors: the Ukrainian SPYS (WIY/Quantum, up to 600 km/h, the first 50 delivered to the troops), P1-SUN JetKiller (Skyfall, 370–400 km/h, a dozen kills, serial production from August), the German-Ukrainian STRILA (415 km/h, a series of 15,000), the British Kestrel (Greenjets, impeller, 500–600 km/h) and the Swedish Kreuger. Symmetrically, "Russian Engineer" calls on Russia to create a 600–700 km/h turbojet interceptor based on the Yak-152/UTS-800 with a GSh-23 cannon source.

A new class of threat — the cheap drone cruise missile. [TECH] The combat use of the European Destinus RUTA Block 1 cruise missile has been documented for the first time: range 300+ km, warhead ~100–150 kg, T150 turbojet engine (thrust 1500 N, weight 17.5 kg), speed ~200 m/s (M≈0.59), RCS ~0.07 m². A critical engineering conclusion from the wreckage analysis: in GNSS-denied conditions the load-bearing navigation loop is not the INS (tens of km of error over 300 km) but a visual navigation system (VNS, from the Spanish UAV Navigation/Grupo Oesía); its vulnerabilities are cloud cover, snow and changed terrain source. The entire C-UAS arsenal developed against "putt-putts" is useless against the RUTA — "grown-up" cheap anti-aircraft weapons are needed.

An airborne interception echelon as a consensus solution. Both sides arrived almost simultaneously at the "airborne interceptor carrier": Russia ("Bars-Sarmat") uses the Yak-52 (a carrier of up to 6 interceptors on external hardpoints) and the An-2 as a flying launcher with control at 50–70 km; Ukraine uses the An-28/PZL M28. [TECH] The engineering logic: an air launch saves the interceptor's energy, "subtracting the fight against gravity" source. In parallel, a maritime vector has matured — USV "motherships" to move interception out to sea (debris falls into the water, coverage of coastal cities).

The key shift — from the "human in the loop" to autonomous target selection on a mass platform. Autonomous neural-network "Lancets" flying without a radio signal (invisible to ELINT and "Grafit", EW is useless) have been recorded, 20 UAVs at a time on the logistics routes of the Pokrovsk direction source. The Russian Backfire also has no communication channels, only a flight mission, a warhead of 9N235 fragmentation elements from the "Smerch", and a SineLink radio module as backup positioning against GPS spoofing source. Swarm autonomy of munitions has also been documented: [TECH] CWA (Collaborative Weapons Autonomy) on AGM-188 missiles demonstrated a time-coordinated approach of two munitions to a target (Time-on-Target) — this is no longer a concept.

The "measure–countermeasure" escalation in navigation and communications. [TECH] A classic example from Flash: stronger EW on locomotives forces the "Shahed Seeker" to lock onto the target with its camera at a longer range; stronger GPS suppression — the adversary builds up receiver protection with CRPA antennas source. The "Geran" has transformed from a "flying bomb" into a universal mesh platform: a 180° panoramic camera (Dahua), an Nvidia Jetson Orin with machine vision, optical homing on moving targets source. Tactically significant: the first control of "Molniyas" via a "Gerbera" → "take down the Gerbera and the Molniyas will fall on their own". In the field of control channels — [TECH] an engineering audit of frequency hopping: the open-source mLRS firmware can be reconstructed by ELINT means in tens of seconds (1.32 s per table cycle versus 19.2 s for ExpressLRS), giving the adversary +28 dB of jamming power — "closer to sticks than to military use".

HPM is moving out of the laboratories into serial production. [TECH] MORFIUS X-Rotor (Lockheed Martin) — a reusable airborne microwave interceptor that burns out 50+ drones per sortie with a single pulse regardless of signal stability source; Raytheon Coyote Block 3NK (19 of 19 drones with two interceptors), the electromagnetic Hermes (Auriga Space) on maglev. The strategic thesis of the week: HPM changes the economics itself (one pulse covers a group), while protecting a drone (shielding, redundancy) kills its main advantage — mass. A race of "shielded FPV vs. HPM" is beginning source.

Weak signals (Under the radar)

Russian drones with laser communications and mesh architecture. [TECH] An early signal on swarm architecture: operator → laser channel (non-emitting, unjammable) → high-altitude "Gerbera" relay → distribution of commands over mesh to strike drones. When one node is shot down, the network reroutes through its neighbors — this makes it possible to guide UAVs far beyond the line of contact and radically complicates both EW suppression and kinetic collapse of the network source.

Numerical modeling of UAV detector losses inside a car. [TECH] Electrodynamic simulations (Altair FEKO, method of moments) of detector signal reception on a pickup truck: an antenna on the dashboard forms a deep null in the rear hemisphere — a 6 dB loss ≈ a twofold reduction in range on a pursuit course; the main threatened sector is an elevation angle of 0–30° (low-altitude approach) source. A practical conclusion for the tactics of using mobile detectors that has so far escaped mass attention.

Embedded onboard AI on dirt-cheap hardware. An LLM (28.9M parameters, ~10 tokens/s) was run offline on an $8 ESP32-S3 source. So far it can only produce simple texts, but the fact of coherent generation on such hardware is an early marker for distributed onboard intelligence in cheap platforms.

Vector projection and implications

Confirmed dynamics (high confidence):

• The jet-powered threat vector will scale up, and propeller-driven interceptors and small-arms mobile fire groups will finally retreat into the close-in niche. Countermeasures required in the coming weeks: serial turbojet interceptors at 400–700 km/h with a fragmentation warhead and head-on guidance, an airborne/maritime launch echelon, cheap "grown-up" anti-aircraft weapons (rounds with proximity fuzes) against the RUTA/"Banderol" class.

• Economic coercion through logistics will continue to take effect faster than counterstrikes reduce production: the FP-1/2 are scaling ×5 (>100 thousand/year versus ~22 thousand in 2025), and the adversary's strikes on refineries and warehouses do not slow the output of Ukrainian drones. Both sides have acknowledged: the rear is no longer safe at any depth.

• HPM interception will become the main driver of drone re-architecture: shielding, distributed architecture, redundancy and analog backups. Whoever first puts compact microwave weapons into serial production will temporarily gain an advantage against swarms; the response is not "your own microwave" but a protected (and therefore more expensive) drone.

Analytical hypothesis (moderate confidence):

• Autonomous target selection on a mass neural-network platform (if the use of "Lancets" and the Backfire is confirmed at scale) devalues EW as the primary countermeasure — the focus of defense will shift to detection by optical/acoustic signature and kinetic interception. EW will retain its role against controlled mesh swarms, but not against fully autonomous munitions.

• The confrontation is shifting into near-Earth space: the maneuvers of "Kosmos-2609/2614" toward the Finnish ICEYE SAR satellites (approach to within 500 m) are an early sign of an orbital hunt for targeting sources. If this scales up, duplication of SAR reconnaissance sources and protection of Starlink/UASAT channels will be required.

• The cyber-cognitive vector as a separate domain: if InformNapalm's version that the "BARS Moskva" air defense crew was coerced into shooting down its own Su-57 is correct, this is a precedent for turning the adversary's air defense loop into a weapon against itself by hacking a training range. Not independently verified, but it sets a threat vector for critical information infrastructure (CII).

External analysis (Western sources)

• NATO is moving from monitoring to actively intercepting drones over its own territory. Romania shot down two UAVs within 24 hours (the second by an F-16 near the Danube Delta, ~10 km from the border), relying on a 2025 law — until recently the use of weapons by aircraft over alliance territory was unthinkable source. This is a direct continuation of the Black Sea dynamics that the Russian-Ukrainian loop describes from the inside.

• The depth of the air defense shortage is worse than the TG loop suggests. For the first time in 30+ years, the US is resuming procurement of the PAC-2 GEM-T (~$2 million) in order to save the scarce PAC-3 MSE (~$5.3 million) for priority targets source. This confirms the Ukrainian thesis that interceptor depletion makes ballistic missiles unpunishable.

• AI planning is entering real US exercises. The Army's COA-GPT platform (DEVCOM ARL) cuts operational planning from hours to seconds at Project Convergence Capstone 6 — a direct Western analog of the Ukrainian "Marichka" and the Russian combat AI loop, with the same principle of "augmenting, not replacing, the human" source.

• Kinetic C-UAS is being monetized. Ondas and RSE Ventures invested in FPF Defense (the Hammerhead interceptor for mass "Shahed" attacks, AI control, a cost per intercept of tens of thousands of dollars) source. The Western market is replicating the logic of cheap kinetic interception tested in the Ukrainian theater.

• Localizing UAV engine production as a national security priority. The US Army allocated ~$80 million for domestic production of electric motors for drones to reduce dependence on China source — the same vector of component-base sovereignty as the Ukrainian-Czech localization of turbines (Airlogix + PBS Group).

• The vulnerability of hyper-concentrated supply chains is a common problem. An explosion at a South Korean missile plant exposed a "single point of failure" (many missile components have a single certified supplier) source. It mirrors both the Ukrainian vulnerability (the Pavlohrad Chemical Plant as the sole producer of solid-propellant rocket motors) and the Russian logistical hyper-concentration that AFU strikes are hitting.