The Control Plane Is Becoming a Software Category

A connected machine can report its location. An autonomous machine can execute a task. Neither capability, by itself, establishes whether that machine is entitled to participate in a particular operation. As vehicles, equipment and software agents begin working across organizational boundaries, this distinction becomes foundational: intelligence enables action; authority makes action legitimate.

This is the emerging role of the control plane for the physical internet. It is not another screen for observing assets. It is the infrastructure that makes identity, permissions, software lifecycle and data exchange coherent across a changing network of participants.

Fleet management beyond telemetry: governing participation

In his September 2026 essay, Platform Science co-founder Jack Kennedy describes the control plane as the foundational layer that establishes and enforces the rules of a dynamic system. His framework spans identity, rights, provisioning, distribution, billing, release management, uptime and governance.^[1] The important insight is not the feature inventory. It is that these functions become mutually dependent when applications, operators and devices continuously enter and leave the system.

A fleet platform may already provide several of these capabilities. Companies such as Samsara, Motive and Geotab should not be reduced to passive dashboards. The distinction is architectural rather than competitive: monitoring explains what happened; a control plane establishes what is permitted to happen, and preserves that authority as circumstances change.

The asset is no longer the complete unit of software. The governed relationship between assets, people and applications is becoming equally important.

Two markets, a shared coordination problem

Commercial trucking and autonomous construction expose the problem from different directions. Trucking brings heterogeneous applications, drivers, customers and vehicle platforms together. Construction brings heterogeneous machines into a shared physical operation. Both require coordination that cannot be resolved inside a single asset.

On September 24, 2026, ASI announced a joint venture with SoftBank Group focused on autonomous construction, alongside a separate $225 million SoftBank investment in ASI. The announcement describes Mobius as an OEM-agnostic orchestration platform for mixed fleets including haul trucks, dozers, loaders and compactors.^[2] This is evidence of investment in cross-machine coordination—not proof that a universal control-plane market already exists.

Orchestration and governance are adjacent, not interchangeable. An orchestration engine assigns work and coordinates execution. A control plane determines which actors may receive assignments, which capabilities are available, and which policies constrain their use. The investment is significant because it illustrates the economic value of operating the fleet as a system rather than a collection of independently intelligent machines.

Connected vehicles need dynamic authority

Consider a leased vehicle used by a subcontractor, running third-party software and carrying a customer’s goods. Ownership, operation, software entitlement and data rights belong to different parties. No single login captures the relationship. A valid command must reconcile the actor’s identity, the asset’s capabilities, the contract, the operating context and the current policy.

The difficult cases occur during transitions: a lease expires mid-assignment; an operator’s credentials are revoked; a software update changes the capability set; a machine moves from one jobsite to another. The control plane must treat authority as versioned state, not a permanent attribute attached to a device.

NIST’s zero-trust architecture provides a useful foundation: access should not be implicitly trusted because of network location or asset ownership, and authentication and authorization are distinct functions.^[3] Industrial deployment adds harder requirements. Authorization must remain meaningful under intermittent connectivity, while evidence must show which policy was in force when an action was admitted.

The control plane cannot become a remote safety loop

The cloud metaphor has limits. A policy service can decide whether a robot may begin an assignment. It should not become the synchronous dependency for every braking decision. Physical systems require explicit separation between supervisory authority, task orchestration and deterministic local safety.

A credible architecture therefore needs bounded offline permissions, expiration rules and defined degraded modes. Some actions may continue under a valid local authorization; others must be refused or brought to a safe state. There is no universal answer. The permissible behavior depends on the hazard analysis and operational context.

Nor should “one control plane” imply one omnipotent database. Different organizations will retain different sources of authority. The opportunity is a federated arrangement in which delegated rights are intelligible across boundaries, conflicts are resolved explicitly and revocation has defined semantics. Centralized visibility is useful; unbounded centralized authority is not.

Where fleet operations and IoT platforms create economic value

The commercial breakthrough is not an additional subscription per machine. It is the reduction of friction around activities that currently require bespoke agreements and integrations: temporarily granting a contractor access, activating a certified application, sharing selected operational data, or accounting for equipment usage across organizations.

A construction operator could authorize a machine for a particular site and time window without surrendering the owner’s broader rights. A logistics customer could receive shipment-specific telemetry without gaining unrestricted access to the carrier’s fleet. A software provider could activate functionality for an assignment, then revoke it cleanly when the assignment ends. These are prospective transaction patterns, not claims about universal capabilities already deployed.

This is a market-formation problem: recognizing an economic category inside an architectural requirement. The buyer is purchasing accountable coordination across boundaries—not identity management, billing or device provisioning in isolation.

What would make the category real?

A name is insufficient. The category becomes credible when buyers can specify the job, compare providers and measure outcomes. Four tests matter:

• Portability: can a participant change hardware, applications or operators without rebuilding every relationship?
• Authority integrity: do delegated permissions, expiration and revocation behave predictably across organizational and connectivity boundaries?
• Operational evidence: can the system reconstruct who authorized an action, under which policy and with which constraints?
• Economic interoperability: can usage and entitlements be reconciled without confusing technical permission with contractual liability?

The durable advantage will belong less to whoever collects the most telemetry than to whoever makes participation dependable without making the network captive. Integration breadth matters, but so do neutrality, governance and the ability to preserve trust when participants disagree.

The network of things becomes governable

The Internet of Things connected assets. The next stage must make their relationships executable. Call it the intelligent network of things—iNOT—as a working concept, not an established standard: a network in which participants can discover capabilities, exchange bounded authority and coordinate economic activity.

In this sense, the physical internet is more than ubiquitous connectivity. It is the possibility that a machine can join a useful network without a bespoke organizational negotiation for every interaction. Achieving that will require common semantics, enforceable boundaries and credible evidence—not connectivity alone.

The control plane may become a consequential software category because it addresses a constraint that grows with every new connected participant. Autonomy expands what machines can do. The next infrastructure layer will expand what organizations can safely and legitimately do together.

The opportunity is not to put the physical world on another dashboard. It is to make its expanding network of capabilities governable.