Girish Joshi, SVP of technology at Collabera, has led Fortune 500 digital shifts for 25+ years and now drives AI and agentic transformation.
Many expected AI to follow the same playbook as the cloud. Enterprises would adopt, integrators would implement, and services growth would follow.
Three years into the generative AI era, that script has yet to play out. Every boardroom has an AI agenda, yet most services companies are still struggling to convert enthusiasm into predictable, scalable revenue.
I don’t think this means AI is failing, but that the traditional IT services business model is. Underneath that failure sits a structural change that I believe will define the industry’s next decade: For the first time in four decades, services firms have a credible path to growing revenue without growing headcount.
The Coupled Growth Era
Since the ERP and client-server era, the services business model has rested on a single assumption: To grow revenue, grow people. Revenue and headcount rose together, and that coupling built some of the world’s largest services companies. AI is the first technology wave that breaks the assumption instead of feeding it.
Demand explains part of it. Deloitte’s 2026 State of AI survey found that while 74% of organizations hope to grow revenue through AI, only 20% are doing so. It is tempting to read that as hesitation, but it’s better read as discipline.
CIOs have spent decades cleaning up after rushed technology decisions and are responsible for keeping banking systems, airline reservations, healthcare platforms and supply chains running continuously, so AI must clear a higher bar than technical capability. Governance, security, regulatory compliance, data privacy and measurable outcomes must be resolved before pilots become programs.
Nor will AI-generated code by itself accelerate enterprise transformation. Modernization is constrained less by the speed of writing software and more by architecture decisions, integration complexity, testing rigor and operational resilience. Enterprise adoption lags innovation for rational reasons, so services demand will arrive more slowly and selectively than the hype suggests.
The Real Scoreboard: Decoupling
Meanwhile, many firms point to AI bookings as proof of progress. Bookings are the wrong scoreboard. Many reported bookings are broader transformation programs with AI as one component, still sold by the hour and delivered by headcount. The economics underneath have not changed.
While many boards still measure AI success in bookings, they should be measuring whether revenue has finally decoupled from headcount. For four decades, those two lines moved in lockstep. The firms that grow revenue while headcount stays flat will have proven their model has genuinely changed. Investors should ask whether AI is increasing operating leverage rather than simply inflating bookings. Decoupling is the single number that separates firms that changed how they make money from firms that changed how they talk.
Transforming The Pyramid Into A Diamond
Decoupling has a human consequence that few leaders are discussing. Nearly every services and staffing business rests on a talent pyramid: a wide base of junior professionals billed at a multiple of cost, supporting a narrow top of architects and domain experts.
AI automates the base first. Routine development, testing and support are exactly the work clients expect AI to absorb, which is the reason they ask why a 10-person team cannot be five. Staffing and staff augmentation firms, whose entire model is billable headcount multiplied by hours and rates, are likely to feel this compression fastest, because AI reduces both the people an engagement requires and how long it lasts.
The margin damage is the visible problem. The hidden one is that the base of the pyramid is also the industry’s apprenticeship system. Today’s architects were yesterday’s junior developers. If AI absorbs entry-level work, firms are not just losing margin; they are cutting off the supply line of their future senior talent.
The response is to redesign the pyramid into a diamond: fewer entry roles, redefined around AI supervision and quality ownership, with deliberate investment in accelerating people toward judgment-heavy work AI cannot yet do. No firm I am aware of has fully solved this, and the ones that do will control the scarcest asset of the next decade.
The Three-Engine Growth Strategy
Managing this transition requires what I call the three-engine growth strategy. Importantly, no engine is safe by default.
1. The Cash Engine
This includes managed services, cybersecurity, engineering services and application management—the recurring businesses that fund AI investment. But this engine is under pressure because these are also the areas where clients will demand AI-driven efficiency gains first.
The response is not to defend existing margins indefinitely, but to redesign the model around AI-driven cost curves: Automate delivery, share productivity gains contractually and become the provider clients trust to operate the next generation of technology environments.
For firms with strong balance sheets, this engine can also be acquired, as industry transactions increasingly favor predictable recurring revenue over simple scale. For example, Accenture’s $4.18 billion purchase of Dragos, runZero and NetRise added over $200 million in recurring revenue, while ServiceNow’s $7.75 billion Armis deal added about $340 million.
2. The Commercial Model Engine
Outcome-based pricing is likely to become an important model as enterprise AI matures, a case Cognizant’s CEO has made compellingly in Fortune. The challenge is timing: Firms cannot underwrite business outcomes at scale while most buyers are still experimenting. Build the capability now so it is ready when adoption reaches scale.
3. The AI Portfolio Engine
Make fewer, deeper bets around specific industry workflows where AI can deliver measurable business outcomes. Run them as accountable businesses, not innovation experiments, and measure success by recurring AI revenue and customer impact rather than certifications earned or prototypes delivered.
Writing The Better Playbook
None of this is skepticism about AI. Gartner forecasts worldwide AI spending will reach $2.59 trillion in 2026, and its analysts call this the inflection year for enterprise AI spending. The window to reposition is before commercialization scales, not after.
For decades, technology services firms competed by scaling people faster than competitors. The AI era changes the question. The firms that I believe will be successful are the ones that learn to scale value faster than people. That is the shift investors should watch, boards should measure and CEOs should build for. AI is not replacing the services industry. It is replacing the assumptions on which the services industry was built.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
