On June 16, McKinsey released an analysis of the "2026 Global Technology Agenda" in China. The original survey covered hundreds of CIOs worldwide, with the core conclusion summarized in one sentence: The mission of top CIOs has shifted from "managing IT" to "leveraging AI and data to drive business growth".
I went through this report three times. What made me want to write something is that it reveals a very counterintuitive phenomenon: many companies are increasing their AI budgets while CIOs are becoming increasingly marginalized. AI projects are being decided by CEOs/CFOs/CTOs, while CIOs have become mere "order takers."
This phenomenon is disheartening. However, McKinsey's assessment is that it is temporary. Once the deep application of AI agents and data is rolled out, the CIO who "understands the business and can implement technology" will instead become the most scarce talent.
This is not chicken soup for the soul; it is the window of opportunity that this AI cycle offers to CIOs.
McKinsey's Judgment: Three New Core Roles of the CIO
The report breaks down the new mission of top CIOs into three tasks:
- Strategic deep integration: AI is no longer a sub-project of IT, but the main line of corporate strategy
- Technical speed improvement: The iteration of models, agents, and data infrastructure must keep pace with business rhythm
- Intelligent large-scale deployment: from one scenario, ten scenarios, to the entire company
Translation: The former CIO was "in charge of computer rooms, repairing computers, and deploying systems." The current CIO is "turning AI into a new production tool and data into new production materials, making the entire company revolve around these two things."
The role has changed, so the assessment criteria must change too. You can no longer prove yourself using operations metrics like "99.99% system uptime." You have to speak in terms of "how much new revenue AI has brought to the company, how much cost it has saved, and how many new business scenarios it has enabled."
This matter sounds vague, but there is a set of data in McKinsey's research that quite illustrates the point.
The flow of IT budget has changed
In 2026, AI-related spending will surpass traditional hardware procurement for the first time, becoming the single largest item in corporate IT budgets.
In other words, the area where enterprises spend the most on IT has shifted from "buying servers" to "buying AI capabilities."
The direction of managing money has changed, and the logic of managing money has also changed—if CIOs still allocate budgets according to old methods, they can no longer control this money.
Data Assetization: The "Foundation" More Important Than AI
There is a paragraph in the report that I read over and over again:
"The CIOs of leading enterprises are embedding AI and data into daily operations, building intelligent-driven organizational forms."
There are two key points: AI + data, embedded in daily operations.
The reason many companies' AI projects can't move forward is not because the models aren't strong enough, but because the data isn't ready. Customer data is in CRM, product data is in ERP, financial data is in SAP, and operational data is in various Excel files—when AI tries to access it, either it can't retrieve it, or the retrieved versions don't match.
So McKinsey proposed the concept of "data assetization," which is quite straightforward: Data should not be scattered across various systems as a "cost," but should be managed as an "asset."
What is assetization? Three things:
- Owned: Each piece of data has a clear owner and responsible person, it cannot be "everyone's".
- There are standards: data format, naming, and metrics are unified across the entire company; the same "customer" cannot refer to three different things in three systems.
- Available: AI Agent can directly invoke it, not just leave it "sleeping" in the data warehouse.
If these three things are not done well, an AI Agent is just a "smart fool"—no matter how good its brain is, if it can't reach the data, it can't get the job done.
Conversely, companies that excel in these three areas see significantly higher returns from AI projects. McKinsey data shows: Companies with mature data assetization achieve AI investment returns 1.8 times higher than their peers.
AI Agent: From "Tool" to "Digital Employee"
The report also mentioned a trend called "agent scaling".
How to understand this term? Let me explain using a few examples I've encountered.
A manufacturing client adopted AI Agent last year, starting with three scenarios:
- Exception handling for purchase orders (previously 3 hours of manual review, now 8 minutes with Agent)
- Predictive maintenance for equipment faults (Agent automatically pushes repair work orders to the site)
- Monthly financial closing reconciliation (Agent automatically matches bank statements with ERP data)
This is called "toolization" — AI is helping employees do specific tasks.
What this company wants to do this year is "scale up":
- Expand three scenes to thirty scenes
- Enable collaboration between Agents (e.g., a procurement anomaly triggers equipment maintenance, with automatic linkage between Agents)
- Connect AI output to approval, assessment, and KPI
This is what McKinsey calls "agent scaling" — AI is no longer just a tool, but a "digital employee" in the company, with positions, responsibilities, and performance evaluations.
| Stage | Features | Typical actions |
|---|---|---|
| Tooling | AI replaces employees for single tasks | Contract review, data extraction, document generation |
| Streamline | AI embedded in business workflows, undertaking multiple steps | From inquiry to quotation to order placement, end-to-end |
| Scale | AI agents can collaborate across departments | Procurement-Production-Finance Full Chain Linkage |
| Organized | AI has positions, assessments, and KPIs | Digital employees and human employees manage together on the same platform |
I compiled this table myself to understand it in reference to the content of McKinsey's report.
Most Chinese enterprises are still at the tooling stage, and those that have advanced to the process stage are already considered leading. Those that have truly reached the "organizational" stage—meaning they treat AI Agents as formal employees to manage—currently account for less than 10% globally.
What should the CIO do?
McKinsey's report didn't state it explicitly, but the research implied several judgments. I translate them into action recommendations for CIOs:
1. Stop being a "order taker"
The initiator of an AI project should be the CIO, not the business department. The business department raises requirements, and the CIO evaluates technical feasibility, ROI, and data readiness before making the final decision. This is a key shift for the CIO from being "directed" to "directing."
2. Govern data first, then deploy AI agents
This is the most hardcore conclusion in McKinsey's data: Data governance maturity directly determines the ROI of AI Agents.
The order cannot be reversed. First, spend 6-12 months solidifying data assetization, then deploy AI Agents; otherwise, the Agent will just be an empty shell.
3. Don’t try to cover too many scenarios; first, thoroughly address the "high-frequency, low-risk" ones.
McKinsey's research contains a counterintuitive data point: Companies with more than 10 scenarios actually see an increase in AI project failure rates.
The reason is simple — as the number of scenarios increases, governance cannot keep up, data cannot keep up, and assessment cannot keep up.
A smart CIO selects 3-5 high-frequency, low-risk scenarios that can be measured with clear ROI, thoroughly implements them first, and then replicates.
What McKinsey didn't say but I must add:
This approach has a prerequisite—the company CEO must truly regard the CIO as a strategic partner.
If the CEO still treats the CIO as "the computer repair person," then the CIO's transformation is like a clever woman who cannot cook without rice. In McKinsey's research, the common factor among top CIOs who succeed is that the CEO is willing to listen to the CIO talk about AI at strategy meetings, rather than only listening to the CFO talk about money and the CTO talk about architecture.
Practical advice for business managers
If you are the CEO or CIO of a traditional enterprise with annual revenue exceeding 1 billion, what is the practical significance of this report? My assessment is three things:
First, the budget structure needs to be adjusted. This year, AI-related spending should account for at least 30% of the IT budget, and it is recommended to reach 50% by 2027. This is not a "pilot budget" but a "mainline budget."
Second, the organization must move. Set up a cross-departmental data governance committee, with the CIO reporting directly to the CEO (or at least at the C-level), and do not let the CIO communicate through a CTO.
Third, performance evaluation must change. Add a KPI for CIOs: "Number of scenarios where AI agents are operational + ROI," making AI implementation a hard metric, not a "soft suggestion."
Finally, let me speak the plain truth.
The terms "AI agents" and "data assetization" sound very "conceptual." But what makes this McKinsey report most worth reading repeatedly is that it ties the fate of the CIO to the fate of enterprise growth—
In the past, the CIO was a "cost center," and IT was a department that spent money;
If current CIOs can seize the opportunity presented by this wave of AI agents, they can transform into a "growth center", making IT a profit-generating department.
Whether this transformation can be completed is not determined by the CIO alone, but by the entire corporate governance structure.
But for CIOs personally, the window of opportunity in this AI cycle has truly opened. Whether you can seize it depends on your actions over the next 18 months.
