The Four IT Operating Models of the AI Era: MIT CISR Names the Choice the CIO Now Owns

See also (wiki): wiki/it-operating-models.md, wiki/ai-maturity-models.md, wiki/agentic-ai-governance.md, wiki/board-ai-strategy.md

Executive Summary

• MIT CISR’s research briefing Enterprise IT Operating Models in the AI Era (Thorogood and Woerner, December 18, 2025; presented at the AsiaPac International Executive Forum on March 26, 2026) names four enterprise IT operating models for the AI era and reports performance data for each. The frame: choosing the right operating model is now a CEO/CIO strategic decision, not an org-chart housekeeping question.
• The framework rests on two axes: Innovation Velocity (how fast the industry’s business model is changing) and Enterprise IT Leadership, Modularity, and Reuse (how strongly the IT function architects shared platforms used across business units). The two axes produce four archetypes — Legacy Modernizer, Creative Sprinter, Adaptive Innovator, Efficient Builder — at 33% / 16% / 33% / 18% of the surveyed cohort.
• The performance gradient is sharp on the IT-leadership axis. Adaptive Innovators run 80% of operations on enterprise platforms and earn 71% of revenues from innovation. Legacy Modernizers run 53% / 37%. That is a 27-point platform gap and a 34-point innovation-revenue gap between the two diagonal corners.
• MIT CISR’s headline conclusion: “innovation is more effective with strong IT leadership, resulting in high modularity and reuse regardless of whether those enterprises operate in environments with low or high innovation velocity.” The choice that matters most is not centralization versus decentralization. It is the degree of IT leadership intensity.
• The framework triangulates with Forrester’s AI CIO (Apr 9, 2026) and McKinsey’s AI Transformation Manifesto (Apr 7, 2026). Three institutional voices in three months — analyst, consulting firm, academic — converged on the same thesis: as AI scales, the CIO’s accountability shifts from delivery to outcomes, and the operating model has to change to support it.
• Follow-on MIT CISR research (Thorogood, Woerner, Gill — University of Technology Sydney) reports a preliminary finding worth flagging now: organizations are seeing “incremental productivity gains from AI but no positive cash flows.” The operating model is the variable currently separating productivity-only wins from cash-flow-positive AI programs.

What MIT CISR Actually Built

Alan Thorogood and Stephanie Woerner conducted 39 executive interviews across 30 companies in 2024–2025, and cross-tabulated against MIT CISR’s 2025 Real-Time Business Survey (n=152). Transcript synthesis used a secure MIT instance of OpenAI’s GPT-5. From that interview corpus, two dimensions emerged that, when crossed, produce a 2×2 framework.

The two dimensions are not the centralized-vs-decentralized debate that has occupied IT operating-model literature for two decades. That debate is settled in MIT CISR’s frame: pure centralization stalls business-unit innovation, pure decentralization fragments the enterprise. The actual variable is whether the IT function leads with shared platforms, modules, and reuse — regardless of where decision rights sit on the org chart.

Horizontal axis: Enterprise IT Leadership, Modularity, and Reuse. The low end describes local decision-making, minimal technology reuse, limited cross-unit lessons sharing. The high end describes strong enterprise IT leadership, reusable platforms and modules, and a balance between local autonomy and enterprise capability.

Vertical axis: Innovation Velocity. The low end describes stable industries where multiyear planning and incremental improvement still work. The high end describes rapidly changing industries where business models shift inside a planning cycle.

The crossing produces four archetypes, each with its own profile of operations on enterprise platforms and revenues from innovation.

The Four Archetypes

Legacy Modernizer (33% of cohort) — low velocity, lower IT leadership intensity

• 53% operations on enterprise platforms, 37% revenues from innovation
• Mature enterprises in slower-changing industries. Business units achieve scale benefits independently. Local technology decision rights predominate; enterprise IT provides governance and shared tools rather than acting as the architect.
• Example case in the briefing: a multinational mining enterprise with diverse business-unit technologies.
• Standing problem: modernizing legacy systems while maintaining the discipline to build reusable platforms. Most Legacy Modernizers know they need more enterprise architecture and underinvest in it because the business case for reuse is harder to write than the business case for the next BU-specific system.

Creative Sprinter (16% of cohort) — high velocity, lower IT leadership intensity

• 70% operations on enterprise platforms, 49% revenues from innovation
• Business units drive innovation aligned to customer experience and product differentiation. Each unit operates in distinct regulatory environments inside a fast-changing industry. Enterprise IT acts as orchestrator, ensuring interoperability and security rather than dictating the platform stack.
• Business units typically own product and channel platforms.
• Example case: Bupa, the international health-insurance group, operates regional IT units (UK, Asia Pacific, Europe / Latin America) aligned to market needs while using common metrics and strategic alignment, enabling replication of successful experiments across regions.

Adaptive Innovator (33% of cohort) — high velocity, high IT leadership intensity

• 80% operations on enterprise platforms, 71% revenues from innovation
• Operates in fast-changing industries that require high innovation. Strong enterprise IT leadership and platform reuse are the enabling conditions, not constraints. Federalizes technology strategy through unified data platforms and enterprise-grade AI, while supporting local business-unit innovation through controlled frameworks.
• Example case: Heineken implements a three-component digital backbone — a digital core, forty enterprise business platforms integrated through the core, and an outer digital products layer for initiatives lasting months to several years. Heineken’s Digital and Technology unit empowers business functions to lead adoption and experimentation, then replicates success internationally. This is the only archetype in which the enterprise IT function is simultaneously the architect of standards and the enabler of business-led experimentation.

Efficient Builder (18% of cohort) — low velocity, high IT leadership intensity

• 65% operations on enterprise platforms, 54% revenues from innovation
• Operates in lower-clockspeed industries that standardize technology enterprise-wide for compliance, reliability, quality, and cost control. Enterprise IT owns the majority of platforms and data architecture, focusing on automation and risk reduction. Business units manage only the truly unique systems — anything common is owned by the center.
• Example case: Hunter Water (Australia) manages all non-operational technology through enterprise IT, with major platforms supporting billing, field service management, and asset management. Moving toward “two-in-a-box” platform management structures that pair business and technology leaders on each platform.

Why the Right-Hand Side of the Matrix Wins

The single most useful sentence in the briefing is Thorogood and Woerner’s framing of the dominant variable: innovation is more effective with strong IT leadership, resulting in high modularity and reuse regardless of whether those enterprises operate in environments with low or high innovation velocity.

The data supports it. The right-side archetypes — Adaptive Innovator and Efficient Builder — both report higher percentages of operations on enterprise platforms (80% and 65%) and higher percentages of revenues from innovation (71% and 54%) than their left-side counterparts. Within the right column, Adaptive Innovator outperforms Efficient Builder on innovation revenue (71% vs 54%), reflecting the velocity premium. Within the left column, Creative Sprinter beats Legacy Modernizer on innovation revenue (49% vs 37%) but neither matches the right column.

The honest read for executives: most companies are sitting in Legacy Modernizer or Creative Sprinter territory because that is where they ended up after a decade of “let business units choose their own tools” guidance. The shift to either Efficient Builder or Adaptive Innovator is real organizational work — building platforms that other units actually want to reuse, building governance that does not slow them down, and giving the IT leadership authority to enforce the standards.

This is the strategic decision MIT CISR puts on the table:

1. Determine the innovation velocity your industry actually requires, given AI-driven disruption risk.
2. Decide the strength of IT leadership principles that drive platform and module building, and reuse.
3. Recognize that the central choice is not centralization versus decentralization. It is the degree of IT leadership intensity.

Triangulation: Three Voices, One Operating-Model Shift

The MIT CISR briefing is the academic point in a three-voice convergence over the past four months that should be treated as more than coincidence.

Three different research methods, three different commercial positions, three different audiences — converging on the same prescription. The CIO of a 200–2,000-employee company who reads any single one of these can dismiss it as analyst-of-the-month. Reading all three together makes that harder.

Where MIT CISR adds the most: the other two sources describe what the CIO’s job becomes. MIT CISR describes what the operating model around that job has to look like for the new accountability to be supportable. Forrester says “govern outcomes.” MIT CISR says “you can govern outcomes from any of these four positions, but the right side of the matrix is where the cash flows.”

Where the Framework Stops Short — and What’s Coming

The December 2025 research briefing is a frame, not yet a deep treatment of how each archetype captures or fails to capture AI value specifically. Thorogood, Woerner, and Asif Gill (University of Technology Sydney) have a follow-on study under way — The Enterprise IT Operating Model’s Role in Capturing AI Value — explicitly designed to answer four questions:

1. How do IT operating models influence AI initiative intake and delivery processes?
2. How does AI governance vary across the four operating model types?
3. How do value capture, supplier ecosystem interaction, and risk management differ by model?
4. What is enterprise architecture’s role in scaling AI initiatives?

The preliminary finding from MIT CISR’s interview cohort is the line that should make every CIO uncomfortable: organizations report “incremental productivity gains from AI but no positive cash flows.” That is the same picture McKinsey reported in its November 2025 State of AI (only 6% of companies capture >5% EBIT impact, n=1,993) and BCG reported in AI at Work 2025 (only 5% of organizations capture substantial financial gains, n=10,635 workers). The operating model is the candidate explanation MIT CISR is now testing for why most companies are stuck on the productivity side of that gap.

For executives who want to read a verbatim Thorogood framing of the underlying business-unit / IT shift, his most-quoted recent line, given to CIO magazine: “IT is critical to every aspect of a business unit’s performance. Business units now incorporate technologies like AI throughout their operations. They take ownership of data and its acceptable uses, and are moving into low- and no-code software development, systems engineering, and integration, so enterprises must examine their IT organization to ensure they have the right skills both in IT and business units.”

Source Credibility

HIGH. MIT CISR is a non-vendor academic research center inside MIT Sloan; it has been publishing IT operating-model research for two decades and the 2025 framework explicitly builds on its prior work (The IT Operating Model of the Future: Four Archetypes, Thorogood, November 5, 2025). Sample sizes are modest but transparent: 39 interviews, 30 companies, plus the MIT CISR 2025 Real-Time Business Survey (n=152) for cross-tabulation. The use of GPT-5 for transcript synthesis is disclosed. The four cases named in the briefing (mining major, Bupa, Heineken, Hunter Water) are public companies whose IT structures can be cross-checked. Limitation: the four-archetype percentages (33% / 16% / 33% / 18%) sum to 100% but are based on a single research cohort; treat the percentages as directional rather than as a stable population estimate. The framework is best read as a strategic decision aid, not a benchmarking study with confidence intervals.

The session presentation referenced in this research file (March 26, 2026, AsiaPac International Executive Forum) is gated to MIT CISR members and forum attendees. The December 18, 2025 research briefing is the underlying publication of record and was used for all numerical claims and case content here.

Key Data Points

What This Means for Your Organization

For a CIO, COO, or CEO at a 200–2,000-employee American company, the practical use of this framework is not to argue about which archetype label fits best. It is to do three concrete things this quarter.

First, locate yourself honestly. Run the percentage calculation: what share of your operations actually run on enterprise-shared platforms, and what share is one-off business-unit infrastructure? What share of new revenue, in the past two years, comes from new products, services, or experiences (versus same-product growth)? If those two numbers are below 60% and 40% respectively, you are in Legacy Modernizer or Creative Sprinter territory regardless of how the org chart describes IT. The first move is naming that, in writing, to the executive team.

Second, decide the move you can actually make. Adaptive Innovator is the highest-performing archetype but it requires the strongest IT leadership and the most architectural capability — not every company has the leadership bench, the platform engineering talent, or the political mandate to get there in 24 months. Efficient Builder is the realistic destination for many lower-velocity industries (regulated utilities, manufacturing, regional financial services, healthcare delivery): standardize the data and platform layer, automate at the center, push only genuinely-distinctive systems out to BUs. The honest internal conversation is which of the two right-side archetypes is reachable, not whether to attempt one.

Third, change the IT-leadership language before changing the operating model. MIT CISR’s framing — “the degree of IT leadership intensity” rather than “centralized versus decentralized” — is a useful reframe to take into your next executive offsite. The fight in most companies is not federal versus unitary; it is whether the CIO has the standing to enforce reuse when a business unit wants its own thing. If the CEO is not yet willing to back that authority, the operating model cannot move regardless of how clean the architecture diagram looks.

If this raised questions about which archetype your organization is actually in, what the realistic 18-month move looks like, or how to structure the executive conversation about IT leadership intensity, I’d welcome the conversation at brandon@brandonsneider.com.

Sources

• MIT CISR, “Enterprise IT Operating Models in the AI Era” — Alan Thorogood and Stephanie L. Woerner, Research Briefing No. XXV-12, December 18, 2025. URL: https://cisr.mit.edu/publication/2025_1201_EntITOperatingModels_ThorogoodWoerner. Credibility: HIGH — non-vendor academic research center; modest but transparent sample (39 interviews, 30 companies, plus n=152 cross-tab); methodology including GPT-5 transcript synthesis disclosed; cases (mining major, Bupa, Heineken, Hunter Water) are public.
• MIT CISR, “Enterprise IT Operating Models in the AI Era — Session Presentation” — Alan Thorogood, March 26, 2026, AsiaPac International Executive Forum. URL: https://cisr.mit.edu/publication/2026_0326_EnterpriseITOperatingModelsAIEra_Thorogood. Member-only access; the session is the executive-forum delivery of the December 2025 briefing.
• MIT CISR, “The Enterprise IT Operating Model’s Role in Capturing AI Value” — Thorogood, Woerner, Asif Gill (University of Technology Sydney). Active follow-on research project. URL: https://cisr.mit.edu/content/enterprise-it-operating-models-role-capturing-ai-value. Preliminary finding: organizations report “incremental productivity gains from AI but no positive cash flows.”
• MIT CISR, “The IT Operating Model of the Future: Four Archetypes — Session Presentation” — Alan Thorogood, November 5, 2025, MIT CISR Annual Research Forum. URL: https://cisr.mit.edu/publication/2025_1105_FutureITOrg_Thorogood. The pre-AI-era articulation of the four-archetype frame, since extended in the December 2025 briefing.
• CIO.com, “How to plan for a new business technology operating model” — Article quoting Thorogood directly on business units’ increasing ownership of data and low-/no-code development. URL: https://www.cio.com/article/3961083/how-to-plan-for-a-new-business-technology-operating-model.html.
• EA Voices, “Enterprise IT Operating Models in the AI Era” — December 17, 2025 third-party summary of the briefing. URL: https://eavoices.com/2025/12/17/enterprise-it-operating-models-in-the-ai-era-2/.
• Triangulating sources (covered separately in corpus) — Forrester, The AI CIO Will Govern Outcomes At Scale (Mark Moccia, Apr 9, 2026); McKinsey/QuantumBlack, The AI Transformation Manifesto (Apr 7, 2026); BCG, AI at Work 2025 (n=10,635); McKinsey, State of AI (Nov 2025, n=1,993).

Brandon Sneider | brandon@brandonsneider.com April 2026