The strait is closed and the numbers prove it
Before February 28, 2026, roughly 3,000 commercial vessels transited the Strait of Hormuz each month. That volume carried 20% of global petroleum and 20% of global LNG. As of June 2026, transit has dropped to approximately 5% of pre-conflict levels. The strait is functionally closed.
The closure is a dual blockade. Iran is mining and interdicting to prevent coalition military logistics. The US and allied navies are blockading Iranian oil exports. On June 6, US forces downed Iranian drones near the strait and struck radar installations at Sirik and Qeshm Island. Iran responded with ballistic missiles targeting US bases in Kuwait and Bahrain. An extended conditional ceasefire is nominally in place. Commercial shipping remains halted.
DDIL in its most concrete form
Naval vessels operating in and around the strait face active jamming, GPS denial, communications intercept, and kinetic threats to relay assets. SATCOM is contested. Undersea cables in the Persian Gulf are at risk. Commercial cellular infrastructure is unreliable during active hostilities.
Any AI system requiring cloud reachback does not function here. Any sensor pipeline that assumes a reliable uplink breaks when the adversary decides to break it. The DoD AI Strategy memo directed AI that works 'on-board, in real time, and often without connectivity.' Hormuz is the operating environment that directive was written for.
Maritime edge nodes need a different architecture
A destroyer in the Persian Gulf, a patrol craft in the Gulf of Oman, a maritime patrol aircraft over the Arabian Sea, an embarked marine unit on a commercial vessel: each is a disconnected edge node. Each needs local AI inference for ISR processing, threat classification, and decision support. Each needs mesh to share a common picture with nearby units when SATCOM is unavailable. Each needs store-and-forward to preserve data until a backhaul path opens.
You do not fix this by scaling down a cloud platform. You build something that treats connectivity as a bonus. Local models run on whatever hardware is available. The mesh shares priority data across surviving links. Evidence stays on the device with full provenance. Sync happens when connectivity comes back. EdgeLance builds for dismounted operations and contested spectrum the same way.
Classification-aware data handling under fire
Maritime operations in a contested strait generate data across multiple classification levels simultaneously. Unclassified AIS tracks, CUI sensor feeds, SECRET intelligence products, and TS/SCI targeting data all exist on the same vessel. A system that cannot enforce data boundaries between tiers either leaks classified material or prevents operators from sharing anything.
EdgeLance enforces classification-aware data tiers at the node level. Each tier has its own storage boundary, sync policy, and wipe timer. When a link opens, the system knows which data can transit based on classification and link security posture. Priority alerts move first. Evidence syncs next. Bulk data waits for a secure high-bandwidth path.
Kinetic denial on top of electronic denial
The Chinese EW threat in the South China Sea is primarily spectrum denial: jamming GPS, disrupting SATCOM, degrading comms. Hormuz adds kinetic denial on top. Radar sites are destroyed. Drones shot down. Missiles hitting bases. The connectivity problem is not just jamming. It is physical destruction of the infrastructure links depend on.
Spectrum denial means the system works without reliable RF. Kinetic denial means the system works when relay nodes, base stations, and satellites may be physically destroyed. The store-and-forward window is not hours. It may be days or weeks. EdgeLance's mission continuity architecture handles both because it makes no assumptions about when connectivity returns. The node operates independently. The mesh reforms around whatever survives.
The planning assumption has changed
Before February 2026, a 95% traffic reduction at Hormuz was a wargame scenario. Now it is a measured fact. Any defense AI vendor whose architecture assumes persistent cloud connectivity has to explain how their system works in the strait right now, not in a PowerPoint scenario but in the actual environment where US forces are in combat.
EdgeLance was designed for this environment before it materialized. Local AI, mesh networking, store-and-forward, classification-aware boundaries, fleet management that works without a network. These are not features added in response to Hormuz. They are the core architecture, built because contested and denied environments are where the joint force actually fights.