For decades, the answer was a machine.
Capital markets, customers, and AI are now asking a harder question at once, and most OEM leadership teams don't yet have an answer. This paper offers a map to the quest.
Anubhav DwivediMachine Science
Glen AllmendingerHarbor Research
Enterprise value is defined by what the OEM sells, what risk it carries, and how it is paid. It moves from transactional "sell-and-forget" to guaranteed performance and outcomes.
In most OEMs, commercial and technical capabilities are out of sync with each other.
Full outcome ownership requires balance-sheet capacity, actuarial-grade data, and direct customer access that most OEMs do not have yet.
| Ownership model | Binding constraint | What we see |
|---|---|---|
| Family-owned machinery company | Capital & specialist talent | Deep customer intimacy and fast decisions. Tends to move more slowly on digital, but builds a sturdier transition if it captures context and aligns service workflows before promising outcome guarantees. |
| PE-backed industrial platform | Hold-period economics | Capital and performance discipline, under investor pressure to launch services fast. Without decision-quality data linking machine events to outcomes, those offerings tend to disappoint. |
| Captive equipment unit | Fragmented decision rights | Access to broader corporate resources, constrained by a siloed view of the customer and decision rights split across the parent organization. |
Only 15–20% of engineering-focused use cases align with today's agentic AI capabilities. Design, simulation and PLM work is deterministic and safety-critical, with little tolerance for autonomous action: it calls for machine learning and optimization, not agents making decisions. The remaining 80–85% is better served that way, predicting remaining useful life or forecasting uptime and maintenance.
Aftermarket is different: judgment-heavy, high-frequency, tolerant of human-in-loop correction, and directly touching revenue through parts, uptime and service contracts. At L3–L4 maturity, more than half of aftermarket use cases are a good fit for agentic AI — it's where outcome-based pricing naturally lives.
A digital twin reading every sensor every minute did not earn a whole-machine uptime guarantee. It replaced periodic manual inspection with a priced, data-backed call on exactly when one part fails.
The shift converted improved service standards into a contracted, guaranteed-uptime outcome with a fee attached and a penalty for missing it.
A 100% farm-power-output commitment was only affordable because a 100x field-service ratio cut maintenance cost 90% first. The guarantee followed the productivity gain.
"The commercial side of OEM service — parts planning, contract entitlements, dispatch, technician coordination — has been treated as a low-tech afterthought; not engineering."
Independent research by The Machine Science Company and Harbor Research.
Outcome-based pricing means charging for the result equipment delivers (uptime, throughput, output) rather than the equipment itself. The customer pays for verified performance under a contracted guarantee, instead of a fixed hardware price.
A maturity model is a staged framework (in this paper, five levels) that maps how an OEM's business model, risk ownership, and pricing evolve from transactional hardware sales toward outcome-as-a-service, so a leadership team can locate where it stands today.
Not quite. EaaS usually means the customer pays a subscription for access to equipment, similar to a lease, while outcome-based pricing ties payment specifically to a measured result (uptime, output, or performance), regardless of how the equipment itself is financed.