Who Owns That? The Operating Model Failure at the Heart of Enterprise AI in 2026

Ask your leadership team who owns your enterprise AI agenda. You will hear: everyone. Which means, in practice, no one.

That answer is no longer a growing pain. It is now the primary reason most organisations are failing to generate returns from AI investment. PwC's 2026 Global CEO Survey, drawing on 4,454 CEOs across 95 countries, found that only 12% of organisations have achieved both cost reductions and revenue gains from AI. Fifty-six per cent report neither. McKinsey's State of AI 2025, surveying 1,993 respondents across 105 nations, found that 88% of organisations use AI in at least one function. Only 39% attribute any EBIT impact to those deployments.

The gap between those figures is not a technology gap. It is an operating model gap. Deloitte's 2026 State of AI in the Enterprise, drawn from 3,235 executives across 24 countries, found that just 34% are using AI to deeply transform their business rather than incrementally automate within existing structures. Enterprise AI is not failing because the models are weak. It is failing because no one has decided who owns what, and no one is measured on the outcome.

Enterprise AI is not failing because the models are weak. It is failing because no one has decided who owns what, and no one is measured on the outcome.

Three distinct ownership models are emerging in 2026. Each produces a different result.

Decentralised: Each function owns its AI agenda independently. The CFO funds finance AI. The CMO funds marketing AI. The CHRO funds HR AI. The outcome is high experimentation volume and low enterprise-level scale. Functions build in silos, data does not move, governance does not exist, and the board cannot read the aggregate return. This describes many large enterprises today.

Centralised: A Chief AI Officer or equivalent owns the enterprise AI agenda. The outcome is better governance and slower adoption. Business functions wait for the AI office to approve and build. The organisation is better protected and less productive. The CAIO becomes a bottleneck rather than an accelerator.

Federated: Functional leaders own outcomes. A central AI office owns standards, infrastructure, and governance. This is the model producing the highest likelihood of enterprise value. The CFO is accountable for AI ROI in finance. The CMO is accountable for AI ROI in marketing. But they are building on shared infrastructure, shared data governance, and shared measurement frameworks owned centrally.

PwC's 2026 AI Performance Study, surveying 1,217 senior executives across 25 sectors, found that 74% of AI's economic value is captured by just 20% of organisations. The primary differentiator between those organisations and the rest is not the sophistication of their models. It is the clarity of their operating model. The conclusion from the research is consistent: central governance, local accountability.

The CFO should not own Marketing AI. The CHRO should not own Supply Chain AI. The Chief AI Officer should not own every use case. When everyone owns their part properly, somebody finally owns the whole. This piece maps how each major function currently sits within that framework, and what the gap looks like in practice.

Finance was the most conservative function two years ago. It has reversed fastest in terms of executive intent. A Salesforce study found that 70% of CFOs described themselves as conservative on AI in 2020. By 2025, that figure had fallen to 4%.

The execution reality is narrower. CFO Connect's State of AI in Finance 2026 found adoption in the finance function has reached 56%, doubling since 2023, yet still the lowest of any corporate function. Bain's CFO Survey 2026 confirmed that only 15 to 25% of CFOs have fully scaled AI in their departments. Forty-five per cent remain in limited pilot mode. Where AI is working in finance, the evidence is specific: accounts payable automation has reduced invoice processing costs by approximately 80%, from 13 to 20 dollars per invoice to 2 to 3 dollars. Gartner's April 2026 research predicts CFOs implementing strategic AI deployment will unlock 10 additional margin points by 2029.

Marketing has more AI activity than any other corporate function and the widest gap between ambition and outcome of any function in the enterprise.

Gartner's 2026 CMO Spend Survey, covering 401 CMOs in North America, the UK, and Europe, found CMOs allocating an average of 15.3% of marketing budgets to AI initiatives. Seventy per cent say becoming an AI leader is a critical goal for 2026. Only 30% report having mature or fully developed AI readiness capabilities. BCG's 2026 survey of 300 global CMOs found 96% stating AI is driving end-to-end transformation of their function. Only a third have actually done the work. Forty-two per cent use GenAI to assist humans with discrete tasks only. Just 8% run campaigns where multiple agents operate autonomously. McKinsey research published in April 2026 found that organisations implementing agentic workflows in marketing can expect 10 to 30% revenue growth from hyperpersonalised campaigns, and estimates those systems will accelerate campaign creation and execution by 10 to 15 times. The distance between the 8% running those workflows and the 42% using AI to write copy is enormous.

McKinsey consistently identifies marketing and sales as the function with the clearest revenue uplift from AI, yet most CRO conversations still treat AI as a productivity tool rather than a revenue architecture decision.

The pattern McKinsey identified in 2025 is now in early deployment: agentic systems that handle pipeline analysis, next-best-action recommendations, forecast modelling, and outreach personalisation at a scale no human sales team can match. McKinsey's November 2025 research found that in any given business function, no more than 10% of organisations are scaling AI agents. In sales, that 10% is already measurably different in pipeline velocity, forecast accuracy, and conversion rates from those still running AI as a copilot for individual reps.

The CRO question for 2026 is not whether to deploy AI in sales. It is whether the CRO has the architecture literacy to govern what is being built, the data infrastructure to make revenue intelligence reliable, and the change management capability to shift a sales organisation from individual output to team-plus-agent output.

Legal is the function with the steepest adoption trajectory of the past 18 months. FTI Consulting and Relativity's General Counsel Report, published March 2026 and drawing on 224 GCs and CLOs across 12 countries, found 87% of general counsel now report using AI within their teams. This compares to 44% in 2025 and 20% in 2023. The ACC's 2026 CLO Survey found 47% of CLOs reporting their CEOs now expect them to develop technology and AI proficiency alongside legal expertise.

The risk is specific. Shadow AI in legal carries a professional liability dimension absent from other functions. Courts have sanctioned lawyers for AI-generated hallucinations. When non-legal teams draft contracts or interpret policy using consumer-grade AI tools without legal oversight, the exposure sits with the general counsel regardless of where the decision was made. The EU AI Act's high-risk obligations covering employment-related AI apply from August 2026.

SHRM's State of AI in HR 2026, drawing on 1,908 HR professionals surveyed in December 2025, found only 39% of organisations have adopted AI within the HR function itself. The IMD's 2026 analysis of CHRO priorities found 91% ranking AI and digitisation as their top concern, ahead of governance, engagement, and talent combined.

HR's challenge is structural in a way no other function faces. The function responsible for managing the workforce implications of AI is the function with the least internal AI maturity from which to lead that process. BCG's AI at Work 2026 report found 72% of all respondents say skill expectations have shifted. Only 36% feel they have received adequate upskilling. Deloitte's 2026 report identified the AI skills gap as the single largest barrier to enterprise AI integration. The EU AI Act's August 2026 compliance deadline covers recruitment screening, candidate evaluation, performance monitoring, and promotion decisions. Non-compliance carries fines of up to 35 million euros or 7% of global annual turnover.

The CIO has become the de facto owner of enterprise AI infrastructure in most large organisations. That is not the same as owning enterprise AI strategy, and the distinction is generating real confusion.

Salesforce's CIO Study 2025, drawing on 200 CIOs across 24 countries, found full AI implementation rose 282% worldwide from 2024 (11%) to 2025 (42%). Ninety-six per cent of CIOs say their organisations either use or plan to use agentic AI within two years. Eighty-one per cent say AI agents have increased the need to work more closely with other functions including HR, Finance, and Sales. Fewer than half are currently doing so. Only 23% are completely confident they are investing in AI with built-in data governance. McKinsey's 2025 research identified IT as the function with the highest level of scaled AI agent deployment, ahead of every other business function. That concentration of AI maturity inside IT has created a new tension: CIO owns the infrastructure, the Chief AI Officer owns the strategy, and the business owns the outcomes. That triangle generates enormous organisational confusion and, in many cases, three separate roadmaps pulling in different directions.

Operations and supply chain is the function with the longest AI history and the most uneven scaling story. Deloitte's 2026 report confirms adoption is most advanced in manufacturing, logistics, and defence, where autonomous systems are already embedded in production workflows. PwC's 2026 Digital Trends in Operations Survey found 89% of operations leaders citing at least one reason why technology investments have not delivered expected results. Integration complexity tops the list, followed by data quality and user adoption. Only 51% of organisations establish a clean, structured data foundation before scaling digital initiatives.

Procurement deserves specific attention. It is simultaneously one of the largest buyers of enterprise AI technology and one of the least governed points of entry. Most organisations now have AI governance policies. Most of those same organisations are allowing hundreds of AI tools to enter the business through procurement channels without meaningful oversight. BCG's AI at Work 2025 research found 54% of employees say they would use unauthorized AI tools if their organisation does not provide what they need. Komprise's 2025 IT survey found 90% of enterprises concerned about shadow AI from a privacy and security standpoint, and nearly 80% reporting they had already experienced negative AI-related data incidents. The governance paradox is precise: the function buying the AI is often not the function governing it.

This is the counterintuitive conclusion the research points to. Assigning enterprise AI to a single owner does not solve the problem. It relocates it.

The enterprises generating disproportionate returns from AI in 2026 have not concentrated ownership. They have distributed it deliberately and structured accountability precisely.

CEOs own transformation. That is not a metaphor. The organisations where the CEO has named AI as a strategic priority with defined outcome targets are the ones where functional leaders take the agenda seriously. Functional leaders own outcomes. The CFO is accountable for AI ROI in finance. The CMO is accountable in marketing. The CRO in sales. The CHRO in people operations. The CLO in legal and compliance. CIO owns infrastructure, data governance, and integration architecture. CAIO owns standards, acceleration, and cross-functional coordination. When everyone owns their part properly, somebody finally owns the whole.

The hire that solves the enterprise AI problem does not look like any hire that came before it. It is a CFO who was built for an AI-native finance function. A CHRO who can redesign workforce architecture under hard regulatory pressure. A CLO who can govern shadow AI risk enterprise-wide. A CMO who can connect AI investment to revenue outcomes with rigour. A CIO who thinks in operating models as much as in systems architecture. A CAIO with a P&L, not a PowerPoint deck.

At Sentiro Partners, this is the hire we are most frequently engaged on. If your organisation is trying to decide who should own enterprise AI, the answer may not be a new title. It may be a new leadership specification.

Sentiro Partners helps organisations define and appoint the leaders required for that shift.