AI is not an IT initiative. It's a CEO mandate.

The pattern repeats almost word for word. A CEO decides "we need to do something about AI." The question gets handed to IT, to the CIO, to the CTO, or to a senior AI hire that has not been made yet. Six months later there is a roadmap, a pilot, and a steering committee. Very little has changed in how the business operates. The initiative dies the death of every other IT-sponsored change programme: slow, under-funded relative to its real upside, and starved of the authority across functions that redesigning work actually requires.

Statistics Canada puts business AI adoption in this country at 12.2%, up from 6.1% a year earlier, with another 14.5% planning to adopt in the next twelve months. Canada invented an outsized share of modern AI. Hinton at Toronto, Bengio in Montreal, the Vector Institute here and Amii in Edmonton. The institutes, the research labs, the talent are all here.

The 12.2% is not Canada's innovation report card. It is Canada's CEO report card. Adoption is stalling on the org chart, one decision at a time, and the decision is who owns AI.

The strategy is the CEO's. The execution is somebody else's.

The argument is not that CIOs and CTOs are the problem. They are not. They are the operators who will deliver the work once the CEO is in the sponsorship chair. AI strategy and sponsorship belong to the CEO. AI execution belongs to a senior executive accountable for delivery, and that can be the CTO, the CIO, the CDO, the COO, a Chief AI Officer, or another senior role.

Strategy is the decision about which workflows are worth rebuilding, which functions absorb the change, which trade-offs get accepted, and which competitive bets get placed. That decision sits with the CEO because the CEO is the only chair with the authority to redirect work across functions and reset the leadership team's priorities. Execution is the delivery: the platform choices, the data work, the model and vendor decisions, the engineering, the operations. That belongs to whichever senior executive's function the work sits closest to today. The owner can shift as the focus moves: customer experience, then the data platform, then product, then AI that acts on its own.

What CEOs are getting wrong is collapsing the strategy decision into the execution function and routing both to IT. You do not stop owning the strategy because you hired somebody to run delivery. The strategy still has to be sponsored, defended, and re-prioritized at the leadership table every quarter, and only the CEO can do that.

How to tell AI is parked in the wrong place

Run this against your own organization in 30 seconds:

• Waiting on a hire. You are waiting for a senior AI hire to start before the real sponsorship work begins.

• Parked in IT. AI sits inside IT's roadmap, gets reviewed at IT steering meetings, and is measured on cost.

• Built as cost takeout. Your AI business case was built from the bottom up around headcount reduction or cost takeout.

• Losing the calendar fight. The ERP upgrade is in flight and AI keeps losing to it.

• Parked behind the parent. You are the Canadian president of a global subsidiary and global owns the AI strategy.

If two of those land, the parking spot is the problem, not the timing and not the technology.

The most common version right now is the first one. One CEO had spent eight months interviewing for a Chief AI Officer before realizing the work he was waiting on was work he could sponsor himself. Which customer service or claims or contract-review workflow gets rebuilt first. Which leaders sponsor which workflows. Where the line gets drawn between what AI decides and what a person decides. By the time the candidate signed, his leadership team had aligned on three workflows to rebuild. The hire walked into a programme already in motion instead of a blank roadmap.

The Canadian-division pattern is harder. Global owns the AI strategy, controls the data access, or sets the platform decisions, and the Canadian division is told to wait. Most Canadian-division presidents have more room than they assume, in the parts of the operating model that are genuinely Canadian: the customer onboarding workflow, the Canadian regulatory workflow, the sales adjustments made for Canadian buyers. The conversation worth having with head office is not "let us run our own AI strategy in parallel." It is "here is the specific Canadian-only work we can rebuild inside your policy frame, and here is what you will learn from it." Some divisions have real room. Some do not. The honest move is to name the constraint precisely rather than wait the parent out.

The strongest case for technology-function ownership, taken seriously

The case for putting AI under the technology function is being made well. Davenport, Bean, and Gopal in HBR argue for a Chief Data, Analytics and AI Officer owning the strategy, the stack, the data work, and the return. The CIO-community view is sharper still: AI that acts on its own across systems is infrastructure, data and security are CIO disciplines, and separating AI from the CIO organization is impractical. Both positions are correct on the question they are answering, which is who runs the work. Neither answers the prior question: who owns the decision the work executes against.

The disagreement is narrower than it looks. The CEO does not have to run delivery. The CEO does have to own the strategy. The HBR piece itself concedes that this role should sit "closer to business functions than to technology operations," which is the same point under a different name.

The cautionary tale is Klarna. Aggressive CEO sponsorship in 2023 and 2024, an AI push the CEO fronted personally, down to cloning his own voice for the customer service line, two-thirds of customer interactions handled by a chatbot, and then in May 2025 Sebastian Siemiatkowski publicly walked the strategy back and started re-hiring human agents. CEO ownership is necessary. It is not sufficient. CEO ownership without an honest execution owner produces the Klarna outcome.

What CEO-sponsored AI looks like when it is working

Tobi Lütke published the internal Shopify memo "Reflexive AI usage is now a baseline expectation at Shopify" in April 2025. Two lines matter, verbatim: "Using AI effectively is now a fundamental expectation of everyone at Shopify." And: "Before asking for more headcount and resources, teams must demonstrate why they cannot get what they want done using AI." The memo wires AI usage into Shopify's review cycle. The CEO wrote that policy, not the CIO. The CTO and engineering leadership run the delivery. The strategy is on the CEO's desk.

Jamie Dimon's 2025 annual letter to JPMorgan shareholders, published in April 2026, frames the same shape at scale: "AI will affect virtually every function, application and process in the company." Dimon does not name a single executive as the AI owner. He coordinates it as a company-wide priority with senior executives running specific parts of the delivery underneath.

The BCG AI Radar 2026 survey puts a number on the pattern: 72% of CEOs now identify themselves as the lead decision-maker on AI, and half say their own job stability depends on getting it right this year. Those CEOs are not running delivery. They are owning the decision the delivery executes against.

The question to take into your week

Look at where the AI initiative on your desk has been parked. If it went to IT and you are getting roadmap slides with no movement in the business, the parking spot is the problem. If it is waiting on a senior AI hire, the parking spot is the problem and the calendar is bleeding. If you are the Canadian president of a global subsidiary and it is parked behind the parent, the specific Canadian-only work is the conversation worth carrying back to head office.

The move this quarter is to pull the initiative out of wherever it has been parked and re-sponsor it from your office, with the senior execution owner, current or future hire, reporting up into your programme rather than sitting next to it. That means:

1. Choosing the workflows. Which two or three get rebuilt first.

2. Sponsoring them. Named leaders, on the record, at your leadership table.

3. Drawing the line. What AI decides, what a person decides.

4. Knowing enough to ask hard questions. Not building it. Interrogating it.

5. Getting the leadership team aligned. Before the hire, not after.

Most of the work that matters in the next twelve months can be done before any senior AI hire arrives.

If the next conversation needs to be about ownership before it is about technology, an AI Jumpstart is the engagement we run with CEOs and leadership teams at exactly that point. Three weeks. We watch how the work actually runs in three areas of your operation, come back with a scored shortlist of where AI would pay off, and end in a go or a no-go. Either outcome is more useful than another quarter of status slides: a decision to rebuild a workflow, or a decision to stop.