The productivity gains unlocked by AI are rattling a longtime pillar of the digital services sector: time-based billing.
The shift toward outcome-based contracts is no longer a marketing option but an economic necessity. This was the finding at the first ESN & ICT Forum organized by Numeum in June of last year, where more than 60 sector leaders gathered to debate new ways of consuming digital services in the cloud and AI era.
“For years, ESNs sold time, profiles, ‘man-days.’ But today, clients don’t buy time… they buy outcomes. And that is where AI changes everything,” summarizes Charles Mauclair, president of Numeum’s ESN/ICT College.
Concretely, this trajectory distinguishes three consumption logics: purchasing licenses and infrastructure (ownership), consuming cloud capacity and managed services (usage), and finally buying measurable results such as productivity gains and operational capabilities (value).
Management models priced by the TJM (Daily Rate) — still dominant — are increasingly challenged by commitments built around performance or gain-sharing.
AI, the leading engine of growth… and of tension
The study “Grand Angle ESN & ICT 2025,” published by Numeum and KPMG in October 2025, illustrates the scale of the phenomenon.
Generative AI tops the market opportunities identified by 81% of ESN and ICT firms surveyed, ahead of digital transformation (58%) and cybersecurity (56%). In practice, 72% of sector players already use AI in their delivery processes, and 67% in their administrative functions.
Measured productivity gains are substantial: 12.5% in 2025 for publishers, with a projected rise to 17% in 2026. For ESNs, the trajectory is even more pronounced, with gains expected to move from 15% to 22.3% between 2025 and 2027.
But this boost in productivity is partly turning against the sector’s traditional economic model. “AI raises team productivity, but it pressures the volume-based business model built on human labor,” notes Numeum in its conjuncture analysis.
The federation cites a figure that alarms ESN leaders: 22% of CIOs already believe that agentic AI could curb certain software expenditures, potentially triggering a ripple effect on provider revenues.
“The economic equation is under pressure,” admits Charles Mauclair. “ESNs must reshape their offer to stay relevant: move from selling time to creating augmented value.”
Three transformation trajectories are emerging: a pivot toward hybrid models that mix human expertise with AI agents, the development of new high-value activities (design and orchestration of agents, governance, integration with business processes), and a reshaping of contracts around outcomes achieved rather than time billed.
Contracts shifting from the TJM to the outcome-based approach
This contractual reshaping is already very tangible in negotiations and tenders.
Three mechanisms recur regularly. First, productivity commitments accompanied by penalties or bonuses. Second, pricing indexed to the savings achieved and the additional revenues generated. Finally, gain-sharing arrangements where the provider and the client share the savings produced by AI.
“Clients no longer buy time… they buy results,” emphasizes Charles Mauclair. ESNs that continue selling time risk a sustained squeeze on their margins.
The French market offers several illustrations of this shift.
BNP Paribas renewed for three years its framework agreement with Mistral AI, extending generative and agentic AI use cases to critical group processes. The Caisse des Dépôts signed, with Sopra Steria, Computacenter and Mistral AI, a framework agreement worth €140 million over four years to deploy 40,000 licenses of generative AI for about 100,000 users. At TP (formerly Teleperformance), around 7% of revenue now comes from “revenue-as-a-service” contracts tied to outcomes such as revenue generation, customer retention, or savings achieved.
These examples show that major French accounts have already steered their providers toward a value-based logic rather than mere effort.
Sell augmented capabilities, not profiles
For Numeum, the transformation also entails a redesign of the offerings themselves. In the Grand Angle study, 63% of respondents report creating new AI-based offerings, 58% note faster delivery cycles, and 54% say they can respond more quickly to tenders thanks to AI.
“It’s no longer about selling profiles, but selling operational capabilities augmented by AI,” sums up Mauclair.
This transformation will not happen without effort. Numeum highlights three priority challenges.
First, training. An EY–Syntec Conseil study cited by the federation shows that 66% of companies that succeed with department-wide AI adoption have invested in a structured training program, compared with less than 20% among those that fail.
Next, governance, with a focus on data security and regulatory compliance (AI Act, DORA, GDPR). Finally, sovereignty, with a preference for French or European AI solutions.
“AI is not a threat to ESNs; it is an opportunity to reinvent themselves,” Mauclair concludes. Provided that reinvention touches offers, skills, and contracts.
