Most manufacturers still treat connected-product data as an internal operational byproduct. The EU Data Act—which started applying in September 2025—changed that, and I’m watching too many industrial firms respond as if it were a compliance filing rather than a structural shift.
The product just got bigger
Under the Data Act, makers of connected products and related services must give users and authorized third parties access to data the product generates. That means an HVAC chiller, a CNC machine, or a fleet vehicle now has two outputs: physical work and a continuous data stream with legal rights attached. Companies can no longer charge for the machine and keep the telemetry as an aftermarket moat.
The Data Act is not an IT regulation. It is a product regulation that turns operational telemetry into a deliverable.
This shifts digital transformation away from internal process improvement and toward external data interoperability. The foundation is no longer just ERP or CRM; it’s API contracts, semantic models, consent management, and audit trails that hold up across company boundaries.
The hidden operating model
Compliance is the easy part. The hard part is that most manufacturers have no owner for “machine data as a product.” Engineering owns the embedded software, IT owns the cloud platform, legal owns contracts, and nobody owns the data contract with a third-party maintenance provider or insurance company.
I see the next 18–24 months separating companies into two groups. One will build minimum-viable exports: raw telemetry dumps, poorly documented endpoints, and manual approval workflows that will drown support teams. The other will treat each data stream as a managed product with a schema owner, an SLA, and a published access policy—and then reuse that same infrastructure for their own AI and partner ecosystems.
The winning moves are less technical than organizational. Name a single senior owner for data-as-a-product with real P&L accountability. Inventory every connected product and its downstream data consumers. Then set data quality and availability targets the same way you set uptime targets for the machine itself.
This is no longer a migration or modernization story. It is the moment product, operations, and technology converge on one question: are you selling a machine, or are you operating a data platform?