Outcome-as-a-Service: From Selling Machines to Guaranteeing Results

On 3 November 2026, IIoT World published a report on a panel at the Industrial AI Summit 2026 devoted to outcome-as-a-service: equipment manufacturers (OEMs) no longer just sell machines, they guarantee a result. The speakers were Boris Scharinger (Siemens), Jan Pingel (Ingersoll Rand) and Karthikeyan Natarajan, CEO of Infinite Uptime, moderated by Rob Tiffany of IDC.

A caveat: the article is sponsored by Infinite Uptime, even though it is labeled "editorially independent", and it reports a discussion among panelists. The claims should therefore be read as the panelists' opinions, not as independent data.

Three Stages of the Model

The panel distinguishes three phases, which differ in who bears the risk:

• Product sale: the customer owns the machine and carries the risk.
• Availability guarantee: the supplier manages the risk of downtime.
• Output as a service: the supplier owns the plant, takes on all operational risk, and the customer pays for the output actually delivered.

Examples cited include: Rolls-Royce engines rented by the miles flown; Siemens winning a rail contract by guaranteeing the transport of one million passengers between Cologne and Düsseldorf every morning from 8 to 9; and Ingersoll Rand's "air over the fence", in which compressors are installed on land adjacent to the plant and owned by the supplier, who pays for the energy and sells compressed air.

When a Forecasting Error Becomes a Contractual Liability

For Siemens Energy gas turbines, the customer grants a 10-day maintenance window every two years and tolerates no unplanned outages in between. The financial exposure was such that Siemens' internal audit was tasked with verifying the prediction algorithms. The panel presented this as a signal: internal audit must be able to examine predictive models, because their errors become contractual liabilities.

The Impact on the Business Model

The transition is described as harder than the technology. A $100,000 machine that over 10 years generated roughly twice its price in service revenue now collects the full price in the first month; under an outcome-based model, the first year brings in about a third of the previous revenue. Startups can begin this way from day one, while established OEMs face a multi-year revenue "cliff" and often bring in financing companies, which take the largest share of the profit.

In Europe, turning capex into opex doesn't win tenders: what matters is shifting risk from the customer to the supplier.

According to the panel, the difference is made by transferring risks such as failures, maintenance scheduling and performance degradation.

The Advantage of Fleet Data

Pricing an outcome-based contract requires learning from hundreds of thousands of machines operating under different conditions. Ingersoll Rand generates 50 to 100 insights per equipment type, across thousands of predictions per week: probability of failure (by the hour, day, week or month), service needs such as filter changes, and capacity signals. If all machines are running at 95%, for example, the customer probably needs more equipment.

Independent AI vendors struggle more: customers often prohibit using their production data to train models for others. One panelist cited 20 projects in three years without a reusable product. OEMs with their own installed base don't need this permission and can share the interpretation rather than the raw vibration data, for example "bearing problem, 10-14 days before failure".

What This Means for Plants and Integrators

Practical guidance, to be assessed case by case:

• Expect broader requests for machine data access in service contracts, and define ownership, use and boundaries of that data up front.
• Ask for transparency on predictive models: accuracy, validation and traceability become contractual issues.
• Prefer sharing interpreted diagnoses over raw data when confidentiality is a constraint.
• For integrators and software consultants, the areas of value are accuracy, auditability and data governance.