Manufacturers should price AI by considering ‘outcome economy,’ expert argues
What you'll learn:
- AI pricing models may be flawed, as manufacturers risk losing money by charging for AI based on usage rather than the value it delivers.
- One expert argues that outcome-based pricing is the solution to align costs with customer values.
- Since manufacturing outcomes are easy to measure, firms that tie AI pricing to performance and operational improvements could gain a competitive edge in the emerging “outcome economy.”
Editor's note: This story was adapted from a contributed article written by pricing "evangelist" Stephan Liozu for Smart Industry's sister brand, IndustryWeek.
As manufacturers continue to implement AI agents, many are running into issues with pricing that could result in losing money over time instead of getting a proper return on investments in building and adopting the technology.
According to one expert, industrial leaders are implementing ineffective cost structures around their AI agents, resulting in a loss of revenue rather than a payoff on their investments.
Monetization misfires
About five years ago, industrial companies arrived at accepting they needed to become software companies, through adopting sensors, data lakes, digital twins, and more.
However, customers were largely unwilling to pay for subscriptions, resulting in loss in revenue and smaller software margins from data monetization, argued Stephan Liozu, an expert on pricing and value with experience in value-based pricing, pricing transformations, and pricing technology.
Liozu said these issues are made worse with the arrival of AI, and executives are walking into the same ambush with their eyes wide open.
Industrial manufacturers and C-suite members today are experiencing anxiety around long-term AI processing costs. As autonomous agents begin performing actual, long-horizon operational work, executives face a highly variable, unpredictable, and potentially volatile cost loop.
As organizations transition routine task execution from human personnel to autonomous software agents, managers must shift from measuring operational capacity by human full-time equivalents to calculating the runtime compute cost of model inference cycles.
The challenge for the industrial sector is that if you attempt to apply a standard, carpeted enterprise IT cloud framework to an uncarpeted plant floor or process network, you walk directly into the hyperscaler OpEx token trap.
Dangers of AI cost structures
What makes AI different—and more dangerous—is the cost structure, Liozu said. The cost of serving an AI feature is not fixed, not declining on a smooth curve and not fully under your control.
Inference cost swings with model choice, token length, context windows, GPU availability and how often your customer's autonomous agent decides to interrogate the system at 2 a.m. One enterprise buyer running an agentic workflow against your product can incinerate a quarter's gross margin in a weekend.
In response, the industry is using tactics like pushing consumption pricing onto the customer, metering the tokens, and charging per API call.
See also: Rockwell report: Days of ‘experimentation’ are over, DX is here to stay
However, those techniques are ineffective; consumption-based pricing for industrial AI is not a business model, but instead signals that a company does not know what their product is worth.
The ‘outcome economy’
Liozu said that the only defensible path is an “outcome economy,” in which pricing is tied to the measurable result your AI produces inside a customer's operation.
In industrial markets, outcomes are unusually concrete, and that is the single biggest structural advantage industrial companies have over pure software peers.
- These outcomes sit inside the customer's ERP and MES. When you price against them, three things change at once, Liozu said.
- The sales conversation stops being about software and starts being about profit and loss.
- The risk of AI non-performance shifts back to the party best equipped to manage it, which is you.
The compute-cost volatility that terrifies your CFO becomes an internal engineering problem, not a customer-facing invoice line.
Liozu claimed that three implications will follow:
First, most AI features being built today will not survive outcome-based scrutiny. If a company cannot point to the specific dollar your feature puts in the customer's pocket, it does not have a product.
Second, the people who should be setting AI prices are not in the product organization. They live in the operations, reliability and service teams who spend their days inside customer plants. They know what an hour of unplanned downtime actually costs because they were on the phone at midnight when it happened. If a former plant manager is not in your pricing meeting, you are going to price this wrong.
Third, the winners of the outcome economy will look less like software vendors and more like performance contractors.
See also: Stories of AI adoption: Wolfspeed uses ‘three-stage maturity model’ for agent deployment
Liozu named considerations like gain-share and contractual clauses where the supplier earns nothing in a month when the customer's key performance indicators did not move and earns a premium multiple when they do. He argued that is what sophisticated industrial buyers are quietly starting to ask for in requests for proposals—and what they will demand openly within 18 months.
About the Author
Sarah Mattalian
Staff Writer
Sarah Mattalian is a Chicago-based journalist writing for Smart Industry and Automation World, two brands of Endeavor Business Media, covering industry trends and manufacturing technology. In 2025, she graduated with a master's degree in journalism from Northwestern University's Medill School of Journalism, specializing in health, environment and science reporting. She does freelance work as well, covering public health and the environment in Chicagoland and in the Midwest. Her work has appeared in Inside Climate News, Inside Washington Publishers, NBC4 in Washington, D.C., The Durango Herald and North Jersey Daily News. She has a translation certificate in Spanish.