Strategy report · Digital services and artificial intelligence
After the Billable Day
From time sold to value produced: why AI forces digital services providers to rebuild their economics.
Axel Tombereau, Managing Partner, Odyssey · October 2026 · 28 pages
Download the report (PDF)
2/3
of a provider’s revenue exposed to AI
Capgemini and Sopra Steria portfolios, among the few that split revenue by service line
−20%
of contract value at renewal
HCLTech: “a contract worth $100m may now be worth $80m”
< 1%
EBITDA margin at three years on time and materials
Odyssey simulation, $60m provider at 9% margin
11%
EBITDA margin with layered revenue
Odyssey simulation, same provider at three years
The essentials
AI makes the billable day deflationary
Margin now depends on layered revenue and a cost base managed per unit produced
Recovering demand and healthy order books hide the essential point: as long as billing rests on time spent, every productivity gain from AI reduces revenue. Neither volume nor daily rates will protect margin — only a rebuilt pricing unit and cost base will. Two decision-makers are involved: the chief executive, who owns the model, and the shareholder, who funds the transition.
For chief executives
01
AI-driven demand growth masks structural deflation: productivity gains flow to the client
AI is now judged in the executive committee by its effect on the P&L, no longer by IT alone on technical or functional grounds. The client buys a measurable outcome and uses the same tools as its provider to verify it. While billing rests on time, every gain cuts the value of the next contract: “a contract worth $100 million may now be worth $80 million,” observes HCLTech. Absent a gain-sharing clause, that gain goes to the client.
02
Layered revenue combines a fixed-fee base, an outcome-linked fee, and asset licensing, on instrumented operations
Roughly two-thirds of revenue sits in service lines that AI reshapes or replaces, based on the published portfolios of Capgemini and Sopra Steria, among the few providers that split revenue by service line. The build demand that offsets this erosion is temporary, so durable revenue will come from running those AI systems. The base covers fixed costs, the outcome-linked fee captures productivity, licensing monetizes assets independently of headcount; costs are managed per unit produced, inference included. The model enters at each contract renewal, through a new pricing unit, a gain-sharing clause, and a measurement protocol. In our simulation, a provider at 9% EBITDA margin falls below 1% within three years on time and materials, and reaches 11% with layered revenue.
03
Four groups of entrants are moving in: incumbents keep their position only by leading the alliances they operate within
Model vendors, consulting firms, AI-native entrants, and software vendors are converging on services, as Accenture's acquisition of Faculty for about £740 million (≈ $1.0bn) illustrates. That convergence exposes how fragile the incumbents' advantages have become: mobilizing engineers and operating at scale lose their value once agents do the work. Two advantages remain — access to client decision-makers and knowledge of systems in production — which make providers the natural operators at the center of the value chain. Holding that position requires contractual alliances spanning what the client now expects, from AI strategy through to financial measurement.
For shareholders
04
Leaving time and materials takes a multi-year commitment: temporarily weaker ratios are the price of a durably higher margin
For two to three years, reported revenue declines and working capital rises, since outcome-linked revenue is collected after costs are incurred and some development is capitalized. That trough is an investment, not a loss. Once the transition is complete, EBITDA margin exceeds its pre-AI level, reaching 11% against 9% in our simulation, while it falls below 1% if time and materials is maintained. The difficulty is timing: the trough coincides with exit preparation and outlasts the horizon on which most executives are incentivized. Shareholders have three options: (1) extend the holding period; (2) recapitalize with a patient partner, as Blackstone did at Sia Partners for up to €250 million (≈ $270m); (3) sell ahead of the trough, leaving the value of the transition to the buyer.
05
Capital now prices the execution corpus, an asset most providers have neither contracted for nor booked
Thrive Holdings, backed by OpenAI, raised about $2 billion in August 2026 to acquire digital services firms among other targets. Beyond teams and order books, these investors are buying: (1) processes, how the work is actually done, exceptions included; (2) business data the provider handles but rarely owns; and (3) execution traces — what was produced, what was reworked, and why — the material an agent needs to work in a real environment. Ownership must sit with the provider, not the client: an asset whose ownership is neither defined in the contract nor carried on the balance sheet has no value in a transaction.
06
Leadership in 2030-2032 will rest on six measures that no dashboard tracks today
All six follow from management decisions and can be tracked by shareholders.
Measure 1
Share of revenue independent of time spent
Measure 2
Share of revenue covered by a signed measurement protocol
Measure 3
Fully loaded cost per engagement, inference included
Measure 4
Assets owned, capitalized, and licensed
Measure 5
Human rework rate on agent output below 30%
Measure 6
Neutrality toward model vendors
None requires an acquisition or industrial investment, so all can be started immediately, provided finance and financial control (FP&A) are equipped to measure them. They also give shareholders a scorecard for their holdings, the same one a buyer will apply.
Window to act
By 2030-2032, a provider’s standing will depend less on its size than on how quickly it shifts to this value-creating model, that is, on how fast it changes its pricing unit and cost base. The opportunity arises at each contract renewal, while client AI build work still funds the transition.
Open distribution
Read the report
Twenty-eight pages, freely available. The report may be shared within your executive committee, your board or your investment team.
Download the report (PDF)
Axel Tombereau, Managing Partner of Odyssey, has more than ten years of experience in digital services, notably at Alten, Segula Technologies, and Atos. He advises the sector's executives and shareholders.
Odyssey works with executive committees and their shareholders through two practices. AI StratOps builds a tailored AI strategy and translates it into an operational roadmap. AI FinOps measures its return, fully loaded costs included, for the executive committee and the board. Odyssey sells no execution services: we work alongside providers, never in competition with them.