The AI-enabled enterprise

The enterprise of 2030 will look structurally different from today's enterprise—not because AI mandates a new org chart, but because it makes the cost of the existing one visible. Decision quality becomes measurable. Redundant coordination layers are exposed. Structural components that justified themselves through information management face a harder question: what unique value do they create when intelligence is no longer scarce?

AI is not primarily a technology shift. It is a value visibility shift. When decision quality is measurable and capability gaps surface in data rather than anecdote, the operating model design is no longer insulated from scrutiny. BCG's 2024 research found that 74 per cent of companies struggle to achieve and scale value from AI despite significant investment.¹ Only four per cent have achieved the capabilities that consistently generate material value.² McKinsey finds that just 17 per cent of organisations report that five per cent or more of their EBIT comes from generative AI.³ AI does not force organisations to change everything. It forces them to justify everything.

Accelerant or inflection point?

There are two coherent positions on what AI means for enterprise architecture, and the distance between them is not academic.

The first treats AI as an accelerant. Productivity improves, decision-making is enhanced, and AI is deployed as a layer on top of the existing structure. Functions continue as outcome owners. The corporate centre coordinates. The operating model remains broadly intact, with AI embedded as a capability enhancement. This is the path of least structural resistance—and the path most organisations are currently following.

The second treats AI as a structural inflection point. As intelligence becomes shared infrastructure rather than local tooling, the case for existing structures comes under pressure. The location of intelligence shifts. Functional boundaries blur where AI closes the information gaps that originally justified them. The role of the corporate centre moves from coordination to governance. Business functions must justify their structural weight in terms of the capability they generate, not merely the processes they own.

The data resolves this debate empirically. BCG's research on the gap between AI leaders and followers identifies a key differentiator: high-performing organisations are three times more likely to redesign workflows in depth than to automate existing processes.⁴ The accelerant path—layering AI onto existing architecture—is the primary reason 74 per cent of organisations are failing to realise material value. Structural adjustment is not a consequence of AI maturity. It is a precondition for it.

The structural value test

The framework for evaluating what AI changes in each part of the enterprise is straightforward. Every structural component—the corporate centre, each business function, each layer of management, each shared service model—must answer three questions:

•First: what unique value does this component create? Not what it does, or what it owns, or how long it has existed—but what value would be lost if it were restructured or removed.
•Second: is that value amplified or diminished under AI? Some roles become more valuable when intelligence is abundant—cross-functional integration, risk ownership, capability development and commercial judgement in ambiguous situations. Others become less valuable when the reporting and information-management functions that justified them are automated.
•Third: is the component foundational, evolutionary, or in structural decline? This classification is not a verdict. It is a signal about where investment and leadership attention should concentrate—and where resources and attention should be withdrawn.

AI does not dictate the answers. It removes the ambiguity around them.

The corporate centre

The traditional corporate centre earns its structural weight through capital allocation, risk oversight and portfolio coordination. These remain genuine value-creating activities in an AI-enabled enterprise. The question is not whether the centre continues to perform them. It is whether it has evolved to ensure that AI and data capabilities work coherently across the enterprise.

A high-value corporate centre in the AI enterprise owns the standards that govern how AI is deployed—not as a regulator, but as an architect. It governs the AI portfolio by shaping its design, not just approving individual projects, ensuring that deployments across functions and geographies are coherent rather than competitive, that model risk is understood at the enterprise level rather than buried in operational units, and that the data standards on which AI depends are defined as enterprise assets rather than locally managed variables.

JPMorgan Chase offers a clear institutional illustration. In restructuring its AI governance, CEO Jamie Dimon moved AI and data out of the traditional technology hierarchy, placing them directly under him and the company president.⁵ The signal was architectural: AI governance is not a technology function. It is a corporate centre function because it determines how the enterprise deploys strategically consequential infrastructure.

A corporate centre that aggregates reporting, hosts committees and arbitrates disputes reactively is performing coordination. A corporate centre that owns standards, governs AI portfolio logic and manages model risk at the enterprise level is performing intelligence stewardship. The distinction will increasingly determine whether the centre is a source of enterprise advantage or enterprise friction.

Business functions

Business functions—finance, legal, HR, procurement, risk and technology—have traditionally earned their structural weight through expertise concentration, process ownership and professional standard-setting. The structural question AI poses is more precise: do functions accelerate enterprise capability, or manage internal processes?

AI closes the information gaps that originally justified functional structures. When AI can synthesise contract risk across thousands of documents in minutes, the legal function's value is not contract review—it is judgement, standard-setting and capability development. When AI can model scenario-based financial forecasts across hundreds of variables, the finance function's value is not modelling, but interpreting what the numbers mean and framing commercial decisions. The work that matters shifts from execution to judgement. The structural question is whether functions shift with it.

High-value AI-era functions share a common characteristic: they build reusable human–AI capabilities that flow into the product, customer, and regional structures that own commercial outcomes. They convert operational learning into enterprise standards. They develop hybrid talent capable of integrating AI into professional judgement. They are net exporters of capability rather than net consumers.

The failure pattern, as the data confirms, is structural rather than technological. Functions that do not evolve from process owners to capability builders are the clearest and most costly example of why enterprise structure—not AI capability—is the binding constraint.

Hierarchy and management layers

Management layers have historically earned their structural weight through three functions: aggregating information upward, supervising execution and coordinating across units. All three are exposed to AI in ways most organisations have yet to confront directly.

Management layers most at risk are those whose primary role is aggregating and relaying information.

Information aggregation—compiling reports, synthesising team outputs, translating operational data into management formats—is the function most directly exposed. When AI can aggregate and present operational data in real time, roles whose primary contribution is information relay lose the justification for the limited information access they once provided. Gartner predicts that by 2026, 20 per cent of organisations will use AI to flatten their structures, eliminating more than half of current middle-management positions.⁶

Amazon's 2025 directive offers a concrete signal. CEO Andy Jassy required each organisation to increase its ratio of individual contributors to managers by at least 15 per cent, enabled by AI-driven task automation.⁷ The implication is clear: where AI absorbs routine management execution, the case for certain management layers weakens—and enterprises must respond deliberately rather than allow erosion by attrition.

AI does not remove hierarchy. It exposes which parts of the hierarchy depend on controlling information rather than creating value.

Shared intelligence infrastructure

The most consistently underestimated requirement of AI at scale is the infrastructure that makes scale possible. Most enterprises treat this as an IT matter. It is not.

Enterprise AI requires explicit data stewardship: clear ownership of the data on which AI systems depend, with accountability for quality and governance sitting at the executive level rather than solely within technology teams. It requires model lifecycle governance, platform ownership, defined risk velocity thresholds, and integration standards that ensure local deployment does not create structural fragmentation.

This is not a technology stack. It is the enterprise operating model intelligence spine—the layer that determines whether AI creates compounding value or compounding fragmentation.

The evidence is consistent. Gartner reports that 85 per cent of AI projects fail due to poor data quality—issues that remain manageable in pilots but become structural in production.⁸ Fragmented data governance is not a technical inconvenience. It is the primary reason AI investment fails to generate enterprise value.

Global Business Services

The structural choice GBS represents is among the highest-leverage decisions facing organisations in this transition. It has three credible futures: as a transaction processor (declining strategic ground), as an automation hub (incremental evolution), or as an enterprise intelligence engine (central to shared intelligence infrastructure).

The data suggests this choice is already being made, often implicitly. The Hackett Group's 2025 research found that 42 per cent of GBS organisations piloted generative AI in 2024, and 63 per cent of early adopters report measurable gains in productivity, cost savings and service quality.⁹ Deloitte's 2025 Global Business Services Survey found that 55 per cent of organisations with strong GBS leadership achieved savings above 20 per cent.¹⁰

The question is not whether GBS survives AI. It is whether it becomes central to the shared intelligence infrastructure that enables AI to scale—or peripheral to it. That is a strategic choice, not an operational one, and it belongs at board level and at the heart of enterprise design.

What withers away

The structural consequences of AI are not evenly distributed. Some components become more valuable. Others become harder to justify. It is worth being explicit about the latter, because the instinct to protect existing arrangements is strong and the cost of doing so is now measurable.

•Federated, ungoverned data ownership withers away. Where business units maintain separate data architectures and resist enterprise governance, the shared intelligence infrastructure cannot form.
•Localised AI tooling without integration logic withers away. Functions that deploy AI independently create tool proliferation and data fragmentation that resist integration.
•Business functions that prioritise territorial control over capability acceleration wither away. Functions that measure contribution in process ownership rather than capability export become structural friction rather than value.
•Management roles built primarily around information relay wither away. This is an economic consequence of AI making information relay available at near-zero cost.

Board-level questions

AI is now a board-level topic in most large organisations. In fewer cases, it is discussed with the structural precision that matters. The questions that matter are not about adoption rates or investment levels. They are about structural consequences:

•Where does enterprise intelligence sit in our architecture? Is it an enterprise asset, governed at the centre and deployed consistently?
•Is the corporate centre stewarding intelligence or merely coordinating functions?
•Are business functions accelerating capability or protecting domain?
•Which management layers would struggle to justify their structural weight if decision quality were transparent?
•Are we scaling capability or scaling coherence?

AI is the most rigorous performance audit most organisations have ever applied—not to their people, but to their structures. Functions and layers long insulated by complexity will not remain so for long.

Organisations that use AI to examine and refine their operating model will compound advantage. Those that embed AI within legacy logic—as an accelerant rather than an inflection point, as a tool rather than a structural signal—will amplify the friction they hoped to reduce.

In an enterprise where intelligence is abundant and transparent, structure must earn its place.

PwC's 2024 research found that AI-proficient workers command a 56 per cent wage premium.¹¹ PwC's 2026 CEO Survey found that 87 per cent of CEOs deploying AI expect new workforce skills will be required.¹²

References

BCG (2024). AI Adoption in 2024: 74% of companies struggle to achieve and scale value from AI. www.bcg.com/press/24october2024-aiadoption-in-2024-74-of-companies-struggle-to-achieve-and-scale-value
BCG (2025). Are You Generating Value from AI? The Widening Gap. Only 4% of companies have achieved cutting-edge AI capabilities that consistently generate significant value. www.bcg.com/publications/2025/are-you-generating-value-from-ai-the-widening-gap
McKinsey & Company (2024). A generative AI reset: Rewiring to turn potential into value. Only 17% of organisations report that 5% or more of EBIT comes from generative AI. www.mckinsey.com/capabilities/tech-and-ai/our-insights/a-generative-ai-reset-rewiring-to-turn-potential-into-value-in-2024
BCG (2025). Targets Over Tools: The Mandate for AI Transformation. High performers are three times more likely to redesign workflows in depth rather than automate existing processes. www.bcg.com/publications/2025/targets-over-tools-the-mandate-for-ai-transformation
Harvard Business School (2024). JPMorgan Chase: Leadership in the Age of Generative AI. www.hbs.edu/faculty/Pages/item.aspx?num=67230
Gartner (2024). Gartner predicts that through 2026, 20% of organisations will use AI to flatten their organisational structure, eliminating more than half of current middle management positions. www.gartner.com/en/newsroom
Amazon (2025). Andy Jassy directive requiring each organisation to increase the ratio of individual contributors to managers by at least 15% by end of Q1 2025. www.aboutamazon.com/news/company-news/2024-letter-to-shareholders
Gartner (2024). 85% of AI projects fail due to poor data quality, with issues masked in pilot environments becoming visible in production. www.gartner.com/en/newsroom
Hackett Group (2025). Report Reveals GBS AI Adoption Accelerating, With 63% Seeing Early Gains. 42% of GBS organisations piloted GenAI in 2024. www.thehackettgroup.com/the-hackett-group-report-reveals-gbs-ai-adoption-accelerating-with-63-seeing-early-gains/
Deloitte (2025). 2025 Global Business Services Survey. 55% of organisations with strong GBS leadership achieved over 20% average savings; 58% have begun or plan to begin their GenAI journey. www2.deloitte.com/us/en/pages/operations/articles/2025-global-business-services-survey.html
PwC (2024). 2024 Global AI Jobs Barometer. Sectors most exposed to AI experience 4.8× higher labour productivity growth; AI-proficient employees command a 56% wage premium. www.pwc.com/gx/en/news-room/press-releases/2024/pwc-2024-global-ai-jobs-barometer.html
PwC (2026). 29th Annual Global CEO Survey: Leading through uncertainty in the age of AI. 87% of CEOs who have deployed AI expect it will require new skills from their workforce. www.pwc.com/gx/en/ceo-survey/2026/pwc-ceo-survey-2026.pdf

← Back to Perspectives

Enterprise Operating Model Design