AI is no longer merely a topic of experimentation for digital services companies (ESN) and information and communications technology (ICT) engineering and consulting firms.
According to the ninth edition of the Grand Angle ESN & ICT study, conducted by KPMG-Numeum, 2026 marks the start of a new cycle in which the question is no longer how to adopt technology, but how to rethink value creation, offerings, organization, and capabilities.
An Economic Model Under Strain
The consensus among leaders is clear: 58% expect a strong or disruptive impact from generative and agentic AI on their business model. The figure rises to 52% among large corporations and mid-market companies, and climbs to 68% among small and very small enterprises (52% foresee a strong impact, 16% a disruptive one).
The pressure mainly stems from demand. Clients primarily expect productivity gains from tools (75% of respondents), transformation of their business processes (58%), and, for about one in three, increased price pressure.
At the same time, 62% note a rise in the need for upfront consulting. The study concludes that value is shifting toward strategic advisory, architecture, governance, and ongoing support, at the expense of a production‑centric approach—the activity most exposed to productivity demands.
The competitive landscape is tightening as well.
The newly emerging, low-cost “AI-native” startups are cited as a risk by 80% of respondents, ahead of client internalization (53%), clients developing their own agents (48%), Big Tech and hyperscalers (40%), and SaaS vendors that integrate AI natively (33%).
Billing: Moving Beyond Time-Based Pricing
Billing on a fixed price or time-and-material basis remains predominant, but two out of three ESN & ICT players are already shifting their pricing models, the study finds.
Among the avenues cited: billing by unit of work or per project (59%), subscription or membership (42%), value- or outcome-based models (33%), monetization of assets or licensing (24%), and success-based billing (13%).
Conversely, 30% have not evolved their model at this stage.
Executives interviewed place the shift toward value creation at the top of the market moves deemed most likely in the near term (37% of them), ahead of platform and agent monetization, compression of daily rates, and then the growth of run.
The report quotes executives who say a project billed at a few thousand euros can generate value for the client estimated in the hundreds of thousands of euros, underlining the urgency to break away from revenue strictly tied to time spent.
Contracts are also evolving: 55% of large enterprises and mid-sized firms have updated their AI-related contracting practices, compared with 22% of SMEs.
The most affected clauses relate to security (35%), compliance, notably AI Act and GDPR (35%), and the use of generative AI on client data (32%). Liability clauses (11%) and intellectual property (7%) are less advanced.
Repositioning on the Value Chain
Regarding the levers to capture value and margin, respondents name upfront advisory, architecture, and governance (36%) as the top priority, followed by delivery productivity and price competitiveness (24%), and security, compliance, and responsible AI (17%).
Strategic consulting is also seen as the priority activity in the evolution of the value chain (40%), ahead of integration and build (28%), run (12%), platformization (12%), and software publication (5%).
AI is becoming a standalone offering, even though its revenue contribution remains limited: it currently accounts for an average of 7% of turnover, with an average target of 18% in three years.
Today, 19% of players derive more than 10% of their revenue from AI; 82% aim to reach this threshold within three years. The most widely offered services are AI strategy consulting (57%), data and AI (53%), AI-driven application modernization (47%), and agentic offerings (42%).
Momentum is increasing, especially in larger organizations: among large enterprises and mid-sized firms, 46% of AI offerings are in production and 19% scaled industrialized, compared with 20% and 6% among SMEs, where 44% of offerings are still at pilot stage.
The study notes that many clients can carry out an initial prototype themselves but still rely on ESN and ICT firms to industrialize it.
Differentiation, however, rests less on technology and more on people: 40% of respondents cite their consultants’ expertise as the primary lever, ahead of sovereign or secure offerings (20%), technology partnerships (14%), proprietary assets (11%), the ability to industrialize and run (10%), and competitive pricing and delivery timelines (5%).
Transforming the Internal: Real Gains, a Framework to Consolidate
Internally, the move to scale is underway.
For 86% of respondents, AI is primarily playing a role in production assistance rather than substitution. The structured hybrid model, combining teams, copilots, and agents, is the dominant adoption mode, cited by 34% of respondents, ahead of non-standardized individual tools (26%) and integration into certain professional workflows (25%). The most widespread uses involve pre-sales, development, and documentation.
The results are tangible: about 38% of ESN & ICT firms report productivity gains of more than 20% in delivery. Yet the impact on margins is not well measured: 27% of large enterprises and mid-sized firms and 42% of SMEs say they have not assessed it. On investment, 55% allocate between 1% and 3% of their turnover to this area.
Governance follows suit: 67% have a formal AI policy, 54% a security and compliance validation process, and 52% trace AI usages and costs. By contrast, 17% still have no formal framework. On client projects, 86% restrict AI usage to strict conditions.
The Battle for Talent
For the authors, the race to AI is as much a talent race as a technology race.
Over the next 18 months, 48% of players plan to hire primarily AI experts (data, LLM, agentic) and 42% hybrid profiles blending technology and business, ahead of seasoned professionals with 3 to 8 years of experience (39%). The most sought-after competencies are critical thinking and validation of AI-generated results (39%), followed by business framing and high‑level advisory (22%).
AI already weighs on recruiting: 62% report an impact, and 47% require AI proficiency from the start of a career.
For junior profiles, expectations vary: 47% anticipate a stable situation, 22% foresee a sharp decline, 9% expect an uptick, and 22% consider the model still evolving. The report warns of tensions in the entry of juniors and underscores the sector’s social responsibility to continue recruiting and training.
In recruitment itself, 79% use AI as an assistant, but no respondent is ready to entrust a recruitment decision entirely to AI.
Training remains an open field: only 20% of respondents have trained more than 80% of their workforce in generative AI, and 9% have trained less than 10%.
Training programs are more advanced in large organizations (62% report intermediate level among large enterprises and mid-sized firms) than in SMEs (43% basic level, 16% simply aware).
