July 30, 2026, (Inside AI) — Capgemini CEO Aiman Ezzat declared a "multi-year modernization supercycle" on Thursday, arguing that companies must overhaul decades-old technology systems before they can deploy artificial intelligence at scale. The statement came as the IT consultancy raised its 2026 revenue growth target after stronger bookings.
Speaking to analysts, Ezzat identified legacy systems, fragmented data, and complex technology estates as the biggest barriers to AI adoption, not access to models. His remarks frame a coming wave of enterprise spending on data platforms, applications, and core infrastructure.
"Every organization today wants to become agentic," Ezzat said, referring to AI systems that perform multi-step tasks. "But before they can become agentic, they must become AI-ready, and most are not."
This readiness gap creates a paradox: generative AI can produce answers, but it often fails to execute business processes consistently when underlying systems remain disconnected. Ezzat noted that years of accumulated technical debt have left data scattered across incompatible systems, making it difficult for AI tools to access reliable information.
"AI is not only creating demand for new business capability; it's also accelerating the modernization of the technology foundation on which those capabilities depend," Ezzat said.
The modernization push reflects a shift in corporate spending patterns. While companies remain willing to invest in AI, Ezzat observed that budgets are becoming more targeted, with clients prioritizing large-scale transformation programs over standalone experiments and pilot projects.
Technical Debt Blocks Agentic AI
Capgemini's diagnosis aligns with broader industry research. A 2025 McKinsey study found that 70% of enterprises cite legacy system integration as the top obstacle to scaling AI. The consultancy's own AI readiness report shows that only 13% of organizations have the data foundations needed for advanced AI.
Agentic AI, which requires systems to reason, plan, and act across multiple applications, demands a unified data layer that most firms lack. Without modernization, even the most advanced models remain isolated from the operational core.
Ezzat's supercycle narrative echoes past infrastructure booms, such as the Y2K remediation and cloud migration waves. However, the AI-driven cycle could be larger because it touches every layer of the stack, from mainframe data stores to modern cloud APIs.
Spending Shifts From Pilots to Platforms
The move away from AI experiments mirrors a maturation in enterprise strategy. Early generative AI projects often delivered impressive demos but failed to scale, as they relied on brittle point-to-point integrations. Now, firms are investing in robust data fabrics and API-first architectures that can support multiple AI use cases.
This shift benefits large systems integrators like Capgemini, which can bundle strategy, implementation, and managed services. The company's raised revenue guidance suggests that clients are committing to multi-year engagements rather than one-off proofs of concept.
Competing viewpoints caution that modernization alone does not guarantee AI success. Some analysts argue that organizational readiness, including talent and governance, is equally critical. A recent MIT Sloan paper found that companies with strong data cultures but moderate technical debt often outperform those with pristine infrastructure but weak AI literacy.
Ezzat acknowledged that spending is becoming more targeted, but he stressed that the modernization cycle is a prerequisite for any serious AI ambition. The comments, reported by Leo Marchandon in Gdansk and edited by Matt Scuffham, signal that the AI hype is giving way to a harder, more expensive phase of enterprise transformation.
