Executive Summary
A recent MIT Sloan study highlights that 65% of enterprise AI initiatives stall due to rigid, legacy IT operating models that cannot support continuous model deployment and edge-data pipelines. The findings urge CIOs and Chief AI Officers to adopt agile 'Data-to-Value' operational frameworks to ensure sustainable AI integration.
Executive Summary
Enterprise artificial intelligence has matured from experimental pilots to core business functions, yet the underlying IT infrastructure has not kept pace. Currently, 65% of enterprise AI initiatives stall before reaching production. The bottleneck is not a lack of technological capability, but the rigid, sequential nature of legacy IT operating models. To realize sustainable AI return on investment, organizations must structurally evolve from static IT provisioning to dynamic, cross-functional “Data-to-Value” pipelines.
What Has Changed Recently
Recent market signals indicate a definitive breaking point for traditional IT service management. Gartner projects that by 2028, 75% of Fortune 500 companies will abandon traditional ITIL frameworks in favor of AI-native operating models. Simultaneously, data from MIT Sloan reveals the heavy toll of inaction, confirming the high stall rate of AI initiatives due to legacy constraints. Furthermore, while the emergence of autonomous refactoring agents is beginning to lower the technical barrier to modernizing legacy code, these tools address only the symptoms of outdated infrastructure, not the structural operating model itself.
The Core Strategic Challenge
The fundamental challenge is that you cannot run an AI-first enterprise on an operating model designed for the 2010s. Traditional IT processes were built for stability, sequential deployment, and siloed governance. AI, however, requires continuous model deployment, edge-data pipelines, and dynamic resource allocation. Forcing generative AI through legacy pipes creates immense friction, resulting in massive capital waste. This is a systemic legacy issue, not a failure of IT leadership. Bolting new AI capabilities onto fragmented, siloed data architectures actively destroys value and traps organizations in perpetual pilot phases.
Three Strategic Pillars
Unified Governance Through CIO and CAIO Co-Leadership The historical divide between business innovation and IT execution is fatal to AI scaling. Stronger organizations mandate unprecedented co-leadership between the Chief Information Officer and the Chief AI Officer. Together, they dismantle legacy silos and co-create governance structures that balance rapid model deployment with enterprise-grade risk management.
Architecting ‘Data-to-Value’ Pipelines Static IT provisioning must be replaced by dynamic, cross-functional workflows. What matters is the velocity and reliability with which data moves from source to model to business outcome. Leading enterprises are restructuring their operating models to support edge-computing pipelines and unified data architectures, eliminating the fragmented workflows that typically stall AI integration.
Shifting from Sequential to Continuous Operations Traditional IT frameworks rely on sequential approvals and rigid release cycles. AI models, however, degrade over time and require continuous training, monitoring, and deployment. The new operating model must natively support continuous integration and deployment specifically tailored for machine learning, ensuring that models remain accurate and aligned with shifting business realities.
The Forward View
Moving forward, executives must view AI readiness as a structural evolution, not merely a technology upgrade. Leaders should monitor the integration of AI-native IT processes within their own organizations, prioritizing the dismantling of silos over the rapid procurement of new AI tools. Do not overreact to the promise of autonomous refactoring agents or view them as a substitute for fundamental operating model redesign; they are enablers, not solutions. The immediate next step is to audit your current IT operating model against your AI ambitions. If your deployment pipelines are still governed by traditional, sequential processes, your AI ROI will remain trapped in infrastructure friction.
Topics & Focus Areas
About Mauro Nunes
I write about the realities behind enterprise AI adoption: where strategic intent runs ahead of operating readiness, where governance becomes a business advantage, and where leaders need clearer thinking, not louder promises. My perspective is shaped by director-level work in digital transformation, enterprise platforms, data, and AI-first modernization across multi-country environments. That experience informs how I think about adoption, governance, execution, and scale.