Anduril Lattice Operates Without Cross-Authority Mesh Substrate
by Nick Clark | Published April 25, 2026
Anduril's Lattice operating system has emerged as the most visible commercial command-and-control fabric in Western defense, integrating Sentry Towers, Ghost UAS, Roadrunner and Bolt interceptors, Pulsar electronic-warfare nodes, Anvil counter-UAS effectors, and selected partner platforms into a unified mission-autonomy surface. The architectural element above Lattice — cross-authority mesh substrate in which sensing, identification, and engagement rules ship with the data object rather than living server-side inside a single vendor's mission-autonomy service — is the layer governed spatial mesh provides. This article examines the structural distinction between platform-centric mission autonomy and a credentialed-data-object substrate, and why coalition operations under DoD CJADC2, AUKUS Pillar II, and equivalent allied programs require the latter as a non-negotiable composition layer.
Vendor and Product Reality
Anduril Industries operates Lattice OS as a sensor-fusion, track-management, and effector-tasking platform across a vertically integrated portfolio. Sentry Towers contribute persistent EO/IR and radar tracks; Ghost-series sUAS provide loiter and ISR; Pulsar contributes RF detection and electronic attack; Roadrunner and Bolt close the kinetic loop on aerial threats; Anvil and the broader counter-UAS family handle low-altitude engagement. The platform exposes Lattice for Mission Autonomy as the higher-order layer that ingests these tracks, fuses them into a common operating picture, and dispatches engagement directives to the appropriate effector. Lattice has been fielded against operational programs including Replicator, the Air Force's CCA-adjacent autonomy work, U.S. Customs and Border Protection deployments, and allied procurements; the platform is CMOSS-compliant where required and has been adapted to ABMS and JADC2 reference architectures.
The technical execution at platform scale is mature and, on its own terms, justified by the operational record. What Lattice provides is a Lattice-internal control plane. Mission rules, identification logic, engagement criteria, and cross-platform handoff sequences are evaluated inside Lattice services running on Lattice-controlled compute, against a data model curated by Anduril. Partner contributions enter through SDK adapters that translate external sensor and effector telemetry into Lattice's internal representation; once translated, they are governed by Lattice-side policy. This architecture is operationally coherent within the Anduril perimeter and produces the cross-platform integration story the company markets. It is also where the architectural gap originates.
Architectural Gap
Coalition warfighting under contested-logistics, contested-comms conditions does not have a single mission-autonomy authority. A combined task force operating under CENTCOM with U.K., Australian, Japanese, and Korean contributions, augmented by industry-supplied autonomy from multiple primes, has at minimum six independent rule-issuing authorities — national, service, theater, coalition, vendor, and unit — each of which must retain the ability to bind, revoke, or condition the use of its sensors and effectors in real time. Lattice as deployed today expresses this multi-authority reality through partner-integration agreements and per-deployment policy configuration on the Lattice server side; the rules do not travel with the data object across authority boundaries.
The structural concerns this produces are well known to defense procurement. First, vendor lock-in: once a coalition's sensor and effector inventory is normalized into Lattice's internal representation, migration cost away from Lattice is paid in track continuity, mission-rehearsal data, and operator training. Second, coalition data-fabric capture: partner-nation contributions, once ingested, are governed by an authority chain that ultimately terminates at a U.S. commercial entity, which is a sovereignty problem several allied programs have explicitly flagged. Third, architectural friction at the platform boundary: any non-Anduril autonomy stack — whether a competing prime's mission system, a national lab's experimental controller, or a coalition partner's sovereign autonomy service — must either operate as a Lattice client or run an entirely parallel C2 thread, with no middle path for credentialed peer-to-peer composition. Fourth, the rules-don't-ship-with-the-data problem: a track produced by a U.K. sensor, identified under U.K. ROE, and offered for engagement to a U.S. effector cannot carry its identification credential and engagement constraints with it as a structural property of the data object; those constraints must be re-evaluated server-side under whatever policy the receiving Lattice instance is configured with.
What the Spatial-Mesh Primitive Provides
Governed spatial mesh inverts the locus of authority. Each contributing party — vendor, service, nation, unit — operates its own mesh under its own credentialing infrastructure. A track, an identification claim, an engagement authorization, or a no-strike constraint is expressed as a credentialed data object whose admissibility rules are bound to the object cryptographically, not stored separately on a server that happens to be processing it at the moment. When a U.K.-origin track is offered to a U.S. effector, the receiving party verifies the credential chain, evaluates the bound constraints against its own admissibility policy, and either composes the action or declines it; in either case, no party has surrendered authority to a single vendor's server-side rule engine.
Federation between meshes is declared, not implicit. Two parties wishing to share tracks, hand off engagement authority, or compose a multi-platform effect publish a federation agreement that specifies which object classes cross the boundary, under what credential requirements, and with what revocation semantics. The mesh substrate enforces the federation; revocation is structural, not advisory — when a national authority withdraws consent, downstream composers lose the ability to use the affected objects whether or not their local Lattice instance has refreshed its policy. This is the architectural element that makes coalition autonomy survivable under partner-withdrawal, contested-comms, and adversarial-deception conditions that platform-centric mission autonomy cannot structurally address.
Composition Pathway With Lattice
The primitive composes with Lattice additively rather than as a replacement. Lattice continues to serve as Anduril's internal sensor-fusion and effector-tasking platform; its tracks, identifications, and engagement directives are emitted as credentialed mesh objects rather than as platform-internal messages. From the perspective of an external party — a coalition partner's sovereign autonomy service, a competing prime's mission system, a DoD-operated reference C2 — the Anduril contribution appears as one credentialed mesh participant among many, with its objects carrying Anduril's credentials and the constraints negotiated through the relevant federation agreement.
Concretely, a Sentry Tower track passes into Lattice as it does today, undergoes Lattice's fusion and identification, and is then published into the mesh as an object credentialed by Anduril and conditioned by the operating authority's ROE. A Roadrunner engagement directive is emitted as a mesh object whose authorization chain includes the issuing operator, the cognizant ROE authority, and the platform. Non-Anduril participants — a partner-nation Aegis cell, a competing counter-UAS vendor, an experimental autonomy from a service lab — contribute their own credentialed objects into the same mesh, federate with the Anduril contribution under declared agreements, and compose engagements peer-to-peer. Cross-platform handoff between, for example, an Anduril Ghost track and a non-Anduril interceptor proceeds through credentialed mesh composition rather than requiring the non-Anduril effector to operate as a Lattice client.
The integration is engineering-tractable. The mesh substrate sits beside Lattice's existing publish surface; existing Lattice integrations continue to function unchanged for parties that prefer platform-internal operation; the cross-authority composition layer is additive for parties that require it. Operator workflows are unchanged. The CMOSS, ABMS, and JADC2 reference-architecture commitments Anduril has already made are preserved.
Commercial and Licensing Trajectory
Defense procurement is moving structurally toward non-vendor-locked autonomy. The 2025–2027 program window — Replicator-2, CCA Increment 2, AUKUS Pillar II autonomy lines, and the European Defence Fund's autonomy calls — increasingly conditions awards on demonstrable coalition interoperability and sovereign-authority preservation. A platform that can credibly claim to operate as a credentialed mesh participant rather than requiring coalition partners to enter its perimeter is structurally better positioned for these awards than one that cannot.
Anduril's competitive position benefits from adopting the substrate layer rather than resisting it. Adoption preserves Lattice's role as the dominant Western mission-autonomy platform while removing the procurement objection that the platform's success depends on coalition data-fabric capture. The licensing pathway is conventional: a field-of-use license to Anduril for the mesh substrate within Lattice's control plane, with the underlying primitive remaining available for federation with non-Anduril participants under the patent's broader terms. The alternative — continuing to operate Lattice as a platform-centric authority while coalition procurement increasingly demands the substrate layer — concedes the architectural high ground to whichever competitor adopts it first.
The adoption case is reinforced by the structural realities of contested-comms operations the Department has been candid about in 2025–2026 doctrine updates. A platform-internal mission-autonomy stack reverts under denial to whatever rules it was launched with, mediated by whatever fragmentary updates can be pushed across degraded links. A credentialed mesh substrate degrades differently and more gracefully: each participant retains the credentialed objects it has cached, the federation agreements it has loaded, and the revocation tokens it has accepted, and composes locally against that substrate without dependence on a reachable central service. For the Pacific theater geometry specifically — where comms denial is a planning assumption rather than an edge case — the substrate layer is not merely a procurement convenience but an operational requirement. Anduril shipping that layer as a structural property of Lattice transforms the platform's positioning from "the integrated autonomy platform" into "the integrated autonomy platform that survives the contested-comms case the Department is planning for," which is the framing the next round of program awards will reward.