The F-35 remains the lynchpin of allied fifth generation air operations, but its tactical value is constrained when mission data cannot flow freely to the unmanned systems and legacy platforms that will do the bulk of distributed sensing and attritable effects. The technical problem is not mystery. The F-35 uses a purpose built, high‑bandwidth, directional waveform known as MADL to protect stealthy communications in a high threat environment. That waveform was designed for secure, close tactical networking among F‑35s and related low observable platforms, not for seamless exchange with other waveforms like Link 16 or proprietary links on uncrewed systems.

Two practical patterns have already emerged as fixes and should be hardened into doctrine and acquisition requirements. First, use airborne and attritable translator nodes that bridge waveforms at the edge. Tests such as the Department of the Air Force gatewayONE experiments with an XQ‑58A Valkyrie showed the concept in action: an unmanned, attritable platform carrying a gateway payload can translate between MADL and other datalinks and push fifth generation sensor tracks to other shooters. That December 2020 campaign achieved partial success while also highlighting reliability and integration shortfalls that still need attention.

Second, use hardened Open Systems Gateways in higher endurance airborne hosts. Lockheed Martin’s Project Hydra demonstrated a U‑2 equipped with an Open Systems Gateway integrating MADL, the F‑22’s IFDL, TTNT, and Link 16 to share 5th generation sensor data down to ground nodes and other platforms. Those demonstrations point to an operational model in which a persistent gateway aircraft or a network of gateways provides waveform translation and multilevel cross domain mediation while preserving low probability of intercept characteristics where required.

Architectural fixes that should be prioritized

1) Mandate OMS/UCI and MOSA interfaces for teaming with uncrewed platforms. The Open Mission Systems approach and the Universal Command and Control Interface provide a machine‑readable, mission‑level message set and an architectural pattern that makes translator and mediator software practical to write, certify, and maintain. Embedding OMS/UCI capability where mission logic and sensor products are emitted avoids brittle point integrations and reduces the time to onboard new swarm or loyal wingman capabilities. The DAF and broader MOSA workstreams already advocate these standards.

2) Treat translators as first class, modular mission packages. Translate, don’t reengineer. Gateways should expose a small, certified set of translated messages rather than attempt to mirror every waveform feature. That reduces certification surface and enables rehosting of translator logic on multiple platforms from high altitude gateways to attritable drones. Gateway software should follow modular open systems approach patterns so the same translator can be swapped between a U‑2, an XQ‑58, and a shipboard node. Practical demos have shown this is feasible but the industry must agree a common message catalog and a hardened runtime for the gateway to meet security and latency needs.

3) Push compute and swarm control to the edge with secure, standardized APIs. Swarm coordination requires low latency decision loops that do not survive a round trip to a distant C2 cloud. The F‑35 cannot and should not become the centralized swarm manager. Instead the jet should be a tasking node that issues intent level commands and constraints. Edge managers on the swarm and local gateways should handle choreography, collision avoidance, and local sensor fusion. That pattern reduces bandwidth pressure on MADL and keeps the F‑35 pilot focused on campaign level decisions. AFRL and Skyborg lineage work on autonomy demonstrates the feasibility of moving more autonomy onto attritable airframes while preserving human oversight.

4) Harden cross domain solutions and multi level security. Translating classified MADL content for use by lower classification consumers or coalition partners requires rigorous cross domain guards, policy engines, and auditable sanitization steps. The move from lab demos to operations will fail if gateway translations create uncontrolled spill of sensitive data. Design gateways with hardware security modules, role based filtering, and audited message transforms so that coalition sharing and uncrewed control do not create new security liabilities. Guidance and implementation approaches for MOSA and cross domain handling already exist in DAF acquisition guidance and should be invoked for every gateway contract.

5) Standardize transport and message catalog for swarm intents. The technical community should converge on a minimal set of intent messages for tasking swarms from crewed platforms: detect and track handoff, search sector assignment, jamming corridor assignment, weapons release authority request, and casualty avoidance constraint. Those messages should be expressible using OMS/UCI or a compatible schema, and be lightweight enough to fit over TTNT, MIDS M‑IDS/Link‑16, and other tactical waveforms when gateways are unavailable. A common message catalog reduces translator ambiguity and enables more robust federation across vendors.

Operational and programmatic steps to accelerate adoption

1) Fund gateway fielding as an operational capability not as a one off experiment. Experimentation in ABMS and JADC2 contexts demonstrated the utility of gateways but funding lines must shift from test events to operationalized squadrons of gateway payloads. That means buying translator mission packages, training operators, and codifying TTPs. The Joint All Domain Command and Control concept recognizes that ABMS, Project Convergence, and Project Overmatch are complementary efforts that will need translator nodes to function coherently.

2) Require OMS/UCI compliance on collaborative combat aircraft and CCAs. The Air Force’s Skyborg and Collaborative Combat Aircraft experiments show that autonomy stacks and attritable airframes are viable hosts for gateway and swarm management functions. Requiring OMS/UCI on those programs will make them immediately useful to existing F‑35 operators via standardized mission messages.

3) Create a certification sandbox for translator software. Certification of translation functions is the choke point. A dedicated sandbox with representative waveforms, a message catalog, and standardized interface tests will let industry iterate quickly and give program offices objective criteria for fielding. The sandbox should include adversary waveform emulation to validate that gateway mediation preserves LPI/LPD properties required by stealthy communications when appropriate. Project Hydra and gatewayONE proved the concept; industrializing it requires repeatable test harnesses.

Risks and unavoidable constraints

Reliance on gateways creates new attack surfaces. Gateways will become high value targets and will require protection by dispersal, redundancy, and low probability of intercept design choices. Translators will also impose latency and potential fidelity loss. The value proposition is not perfect fidelity but operationally relevant information shared fast enough to change outcomes. Finally, software and certification cycles are slower than the tempo at which autonomy and swarm tactics are evolving. That gap must be closed with MOSA based modularity and by giving program offices authority to approve and field translator increments rather than full monolithic releases.

Conclusion

The underlying technologies needed to make the F‑35 work with drone swarms exist today: directional, stealthy waveforms for platform preservation, translator gateways for cross‑waveform mediatation, open mission interfaces to remove brittle integrations, and edge autonomy to scale swarm behaviors. Demonstrations from gatewayONE and Project Hydra proved the operational concept. Work in Skyborg and AFRL autonomy shows AI agents can manage attritable jets in tactical contexts. The remaining work is engineering and program discipline: standardize OMS/UCI message catalogs, certify modular translators in sandboxes, fund gateway fielding as operational capability, and adopt edge based swarm managers that accept intent APIs rather than raw sensor dumps. Get those elements right and the F‑35 becomes less of a solitary node and more of an orchestrator in a distributed, survivable kill web.