THE CEO AGENDA 2027

The business media is debating how many jobs AI will destroy. CEOs should be asking a different question.

Few subjects are receiving more attention than the effect of artificial intelligence on employment. Bill Gates has recently argued for protecting certain “human reserved” jobs, while public anxiety has risen sharply; Fortune cited Pew research showing that 71 percent of American adults believe AI will result in fewer jobs over the next two decades. At the same time, business leaders such as Nvidia CEO Jensen Huang are challenging proposals intended to discourage automation.

The debate is usually framed as a contest between two futures. In one, AI destroys jobs. In the other, technology creates enough new work to replace them.

For an individual CEO, that debate may be too abstract to be useful.

The immediate management problem is that AI is changing the relationship between jobs, tasks, experience and organizational capacity long before anyone knows the final employment number.

Consider a finance department employing 100 people. AI may automate substantial amounts of reconciliation, reporting, variance analysis, research and drafting without eliminating 40 jobs tomorrow. Instead, management may simply stop replacing departures, hire fewer junior employees and expect each manager to supervise larger portfolios.

Five years later the department may still employ 75 people, but it is a fundamentally different organization.

There are fewer entry points.

Managers have wider spans.

Junior employees spend less time producing and more time reviewing machine-generated work.

Expertise is concentrated among fewer experienced people.

The organization appears more productive, but its internal system for manufacturing future expertise has changed.

That is the strategic issue hiding behind the employment headlines.

The job is the wrong unit of analysis

Most workforce planning still begins with positions.

A better AI-era analysis begins with work.

For each important role, management should understand which activities involve information gathering, analysis, creation, coordination, judgment, human interaction, physical execution and accountability. AI will affect each differently.

A role that is 40 percent automatable does not necessarily disappear. It may become a much more valuable role if the liberated capacity is redirected toward customers, judgment or growth.

Alternatively, management may continue demanding the same work and simply expect it faster.

The technology is identical.

The organizational outcome is entirely different.

The CEO question

Instead of asking:

How many jobs can AI eliminate?

ask:

What workforce would we design if AI were already embedded in every important workflow?

The answer should influence hiring before layoffs, career architecture before restructuring, and management design before headcount targets.

Organizations that approach AI primarily as a labour-reduction technology may find the savings attractive.

Organizations that redesign work itself may discover something more valuable.

Employees can save hours every week while the company saves almost nothing

This may become one of the most consequential AI stories of the next phase of adoption.

Companies have invested heavily in AI tools, employees report significant productivity improvements, and yet measurable enterprise economics can remain stubbornly difficult to identify. An August Harvard Business Review article highlighted research in which large-scale AI adoption produced no statistically distinguishable change in earnings or hours despite workers reporting productivity gains.

At first this appears paradoxical.

It isn’t.

Imagine 5,000 employees each save three hours a week through AI. Across 48 working weeks, that represents 720,000 hours of theoretical capacity.

If the organization’s fully loaded average labour cost is $75 an hour, management could describe that as $54 million of productivity.

But no $54 million appears on the income statement.

Employees are still employed.

Salaries are still paid.

Buildings and systems still operate.

The saved hours are absorbed into email, additional analysis, more meetings, faster turnaround, increased expectations or simply a less frantic workday.

All may have value.

None automatically creates $54 million of economic return.

Productivity is capacity, not value

This distinction deserves much more attention in board discussions.

AI can create capacity.

Management creates value by deciding what happens to that capacity.

There are only a limited number of destinations.

The organization can reduce cost, increase output, improve quality, serve more customers, shorten cycle times, accelerate innovation or redeploy employees toward more valuable activities.

Unless one of these occurs, productivity remains largely theoretical.

Stop asking for AI use cases

Many organizations now have hundreds of them.

A more useful question for every major AI initiative is:

Where does the economic benefit actually land?

If marketing saves 20,000 hours, what happens next?

If software development becomes 30 percent faster, does product release frequency increase?

If customer service agents resolve cases faster, does service capacity increase or staffing decrease?

If lawyers produce contracts more quickly, does the organization shorten transactions or merely generate more drafts?

AI investment is moving into its second phase

The first phase rewarded experimentation and adoption.

The second will reward economic conversion.

That requires much closer coordination among the CIO, CFO, CHRO and operating leadership than most technology deployments traditionally required.

The CIO can provide capability.

The CFO can identify economics.

The CHRO can redesign work.

Business leaders must convert the result into customer or financial value.

The CEO increasingly has to connect all four.

The AI ROI problem may therefore have surprisingly little to do with the quality of the AI.

It may be exposing the quality of management.

One of the newest AI risks has almost nothing to do with hallucinations, cybersecurity or privacy

The initial discussion about AI risk focused understandably on inaccurate information, privacy, intellectual property, cybersecurity and bias.

A subtler risk is now attracting attention.

What if frequent reliance on AI gradually weakens the executive capabilities it is intended to augment?

Harvard Business Review recently highlighted research involving experienced evaluators suggesting that AI-supported decision-making can undermine original human judgment. Other current HBR coverage warns executives about delegating not only tasks but aspects of voice, agency and leadership presence to AI systems.

The concern deserves serious attention because senior executives spend much of their careers accumulating something that is difficult to put into a spreadsheet: contextual judgment.

They have seen strategies fail.

They recognize patterns.

They know which number in a 40-page presentation deserves attention.

They understand when a customer is saying one thing but signalling another.

They occasionally reject an analytically attractive recommendation because something does not make sense.

AI can strengthen those capabilities by widening the evidence and challenging assumptions.

It can also weaken them if executives increasingly outsource the difficult cognitive work that created judgment in the first place.

The GPS problem comes to management

Most people who rely heavily on navigation software know the phenomenon.

You can arrive successfully at dozens of destinations without learning how the city fits together.

The technology improves immediate performance while weakening the underlying mental map.

A similar possibility now exists in executive work.

AI can summarize the board package.

Draft the strategy.

Produce acquisition scenarios.

Identify risks.

Suggest questions.

Prepare the speech.

Analyze the competition.

Each individual use can be entirely sensible.

But if the executive increasingly begins with the machine’s framing, the machine may quietly influence which problems are considered, which alternatives are visible and which language leadership uses to describe reality.

Preserve the cognitive work that matters

Executives do not need to avoid AI.

They need to distinguish between cognitive labour worth eliminating and cognitive labour worth preserving.

Producing a first-pass industry summary may deserve automation.

Defining the strategic problem probably should not.

Comparing financial scenarios can be accelerated.

Choosing the assumptions on which those scenarios depend remains executive judgment.

Drafting alternatives can be delegated.

Owning the decision cannot.

The leadership question

Every executive team adopting AI should identify a category of activities for which the organization deliberately preserves human reasoning before AI intervention.

That might include strategic problem definition, executive assessment, major capital allocation, sensitive personnel decisions and irreversible strategic commitments.

The objective is not technological restraint.

It is intellectual fitness.

The more powerful the analytical machine becomes, the more important it may be to preserve the humans capable of disagreeing with it.

Companies keep trying to change employees when the unresolved decisions are sitting around the executive table

When a transformation stalls, leadership tends to look downward.

Employees are resistant.

Managers do not understand the technology.

Training is insufficient.

Culture is slow.

Communication needs improvement.

Recent research highlighted by Harvard Business Review offers a more uncomfortable possibility. A study spanning leadership teams at European IT-services firms found that AI adoption was being constrained not primarily by employee resistance but by what the researchers described as leadership drift: executives individually recognized the need to change but collectively avoided the difficult decisions required to do so.

This pattern will be familiar far beyond AI.

The CEO says AI is strategic.

The CFO wants measurable returns before further investment.

The CIO wants enterprise standards.

Business units want autonomy.

The CHRO worries about jobs and capabilities.

Legal emphasizes risk.

Nobody is necessarily wrong.

Collectively, however, their perfectly reasonable positions can produce paralysis.

Organizations can experiment indefinitely

A pilot is one of the safest decisions in management.

It signals action without requiring much commitment.

Consequently, companies can accumulate AI pilots while leaving the truly difficult questions unresolved.

Which work will disappear?

Which processes will be redesigned?

Will savings be removed or reinvested?

Who owns enterprise AI decisions?

What happens to junior hiring?

How does pricing change?

Which decisions can AI make?

Where will the organization tolerate failure?

These are leadership questions.

They cannot be solved by buying another platform.

The alignment illusion

An executive team can leave an offsite believing it has agreed on AI because everyone supports “responsible adoption.”

That is not alignment.

Alignment becomes visible when executives agree on the trade-offs.

A much stronger exercise is to interview each member independently and ask five questions:

What are the three most important AI priorities?

What should the organization stop doing?

How much economic value should AI produce?

Which workforce decisions will be required?

Which executive is accountable?

If the answers differ materially, the company does not yet have an AI implementation problem.

It has an executive alignment problem.

Canadian CEOs can no longer treat trade policy as something for government-relations departments to monitor

The renewed Canada-U.S. trade conflict has rapidly become one of the dominant Canadian business stories. Current coverage describes escalating tariffs, threatened further measures affecting vehicles and steel, retaliatory Canadian actions, and growing concern among exposed manufacturers and small businesses about the uncertainty surrounding investment and employment.

The most important lesson for management is larger than any individual tariff.

For several decades, many North American businesses treated the continental trading environment as infrastructure: sufficiently stable that it could sit in the background of strategic planning.

That assumption can no longer be taken for granted.

Geography has returned to strategy

A component sourced from the lowest-cost supplier is no longer necessarily the lowest-risk component.

A factory located close to a customer may have value that does not appear in conventional labour-cost comparisons.

Supplier concentration can produce financial efficiency while creating geopolitical fragility.

Domestic production that looked uncompetitive under stable trade conditions can acquire strategic value when borders become economically less predictable.

This does not mean every company should reshore its supply chain.

It means location itself has returned as a strategic variable.

The hidden balance sheet

Most companies maintain financial balance sheets.

Few maintain what might be called a geopolitical balance sheet showing where strategic dependencies reside.

Management should be able to identify:

which revenues depend heavily on a particular foreign market;

which suppliers are effectively irreplaceable;

which products cross a border multiple times during manufacture;

where tariff exposure sits;

where regulatory divergence could create problems;

how quickly alternative sourcing could be established.

The Canadian CEO question

For Canadian organizations in particular, there is a difficult but valuable thought experiment:

If reliable preferential access to the United States were no longer an assumption but an advantage that had to be continually defended, what would we do differently?

The answer might involve customers, investment, manufacturing, procurement, partnerships or international expansion.

Companies do not need to predict the next tariff announcement.

They need strategies that are less fragile when one arrives.

For forty years, companies optimized away the spare capacity that may now be becoming strategically valuable

Traditional management thinking rewarded efficiency for good reason.

Lower inventory improved working capital.

Supplier concentration created purchasing leverage.

Centralization reduced duplication.

Higher asset utilization improved returns.

Global sourcing lowered production costs.

Each decision made sense individually.

Together, they often produced organizations optimized for a reasonably predictable world.

That world is becoming more difficult to assume.

Trade conflicts, geopolitical sanctions, energy volatility, cyber disruption, climate events and rapidly changing technology mean that redundancy can sometimes create economic value rather than merely additional cost. The current expansion of trade and sanctions concerns illustrates how quickly political decisions can alter commercial assumptions.

The cheapest system may not be the most valuable system

Suppose Supplier A costs $10 per component and Supplier B costs $10.50.

Conventional procurement logic favours A.

But if Supplier A represents 100 percent of supply from one geopolitical jurisdiction while B allows a dual-source structure, the additional 50 cents purchases something else:

optionality.

Traditional accounting records the extra cost.

It does not easily record the resilience.

CEOs need a new efficiency frontier

Organizations should not respond by creating expensive redundancy everywhere.

Instead, identify points where disruption would be disproportionately costly.

These often include critical suppliers, energy, specialized talent, cloud infrastructure, financing, logistics, key customers and data.

Then calculate not only the cost of redundancy but the cost of failure.

That changes the economics.

From just-in-time to just-in-case?

The answer is neither.

A sophisticated organization will increasingly operate both systems simultaneously, maintaining extreme efficiency where disruption is manageable while intentionally buying resilience where failure could threaten the enterprise.

The strategic skill is knowing the difference.

CEO succession is no longer simply about finding someone capable of running today’s business

Leadership succession periodically becomes a major business-media story because prominent CEO changes allow investors to debate the future of an entire company through the qualities of one individual. Current coverage surrounding Apple’s next leadership era is a useful example: the succession discussion is inseparable from whether the company can reinvigorate innovation and establish a stronger position in AI while managing enormous scale and complex supply-chain pressures.

That raises a broader governance issue.

Most succession processes begin with the current organization.

What capabilities does the CEO role require?

Which internal candidates possess them?

Which experience gaps exist?

That approach may be backward-looking.

If AI, industry structure, customer behaviour and globalization are changing the company materially, boards should begin somewhere else.

Describe the future company first

Imagine the organization five years after the succession.

Perhaps it has fewer management layers.

Perhaps half of administrative work is AI-assisted.

Perhaps the company operates in different geographies.

Perhaps software represents a larger portion of value.

Perhaps supply chains have regionalized.

Perhaps a major acquisition has transformed the portfolio.

Perhaps the existing core business has become structurally less important.

Now ask:

What CEO does that company require?

The answer may be different.

Succession is a strategy exercise

Boards should therefore connect succession planning explicitly to scenarios.

If Strategy A succeeds, what leadership capabilities become critical?

If disruption accelerates, what qualities matter?

If consolidation occurs, what experience becomes valuable?

Which potential leaders can operate in several scenarios?

The internal-candidate paradox

Long-serving executives possess invaluable organizational context.

They may also be most deeply shaped by the assumptions that created the current business.

External candidates bring different perspectives but less institutional understanding.

The question is not which category is inherently better.

It is whether the board understands what needs to change before evaluating who can change it.

The most important succession document may therefore not be the candidate specification.

It may be the future-state strategy.

Scale can make companies exceptionally good at exploiting yesterday’s innovation

One reason leadership transitions at successful technology companies generate enormous attention is the innovation paradox they reveal.

A company can become extraordinary at operations, distribution, capital allocation and incremental improvement while gradually losing the organizational conditions that created its original breakthroughs. Current coverage of Apple’s leadership transition explicitly frames the question around whether a vast, highly successful company can reinvigorate innovation as AI alters the technology landscape.

The issue applies far beyond technology.

Large successful businesses inevitably build systems designed to protect what works.

Investment committees demand evidence.

Budgets favour established businesses.

Processes reduce variance.

Performance metrics reward predictable outcomes.

Experienced executives understand the current model.

All of these are signs of competent management.

They can also make genuinely disruptive innovation difficult.

The core business has an immune system

A new business competes against an established business for resources, talent and executive attention.

The established business arrives with customers, forecasts, historical financial results and experienced advocates.

The experimental business arrives with assumptions.

Guess which usually wins the budget meeting.

That is why executives often say innovation is strategically important while capital allocation says otherwise.

Measure the innovation portfolio

Leadership should be able to divide investment into three broad categories:

improvements to the existing business;

adjacent growth opportunities;

businesses that could eventually challenge the existing model.

Many organizations discover that nearly all discretionary investment sits in the first category.

That is not necessarily wrong.

It should at least be conscious.

The CEO’s uncomfortable question

If someone were going to make our most profitable business economically obsolete, what would they build?

Then ask:

Why aren’t we building some version of it ourselves?

If the answer is that doing so would threaten current economics, management may have identified precisely why an outsider has the opportunity.

Installing AI into an existing workflow is easier than asking whether the workflow should continue to exist

Current business coverage is increasingly moving beyond AI adoption toward workflow redesign. The World Economic Forum recently highlighted lessons from Chinese businesses attempting to move from adoption toward durable productivity, emphasizing redesign of workflows rather than simply layering AI onto existing work. HBR has similarly focused attention on organizational areas where companies have deployed AI without converting it into measurable business results.

This distinction may define the next competitive divide.

Consider a traditional process containing twelve steps and three approvals.

Management can add AI to steps 2, 5, 7 and 9 and demonstrate significant local productivity improvements.

A new competitor might ask why twelve steps and three approvals exist at all.

It builds a four-step process around AI from inception.

The incumbent automated.

The competitor redesigned.

Legacy processes contain historical constraints

Organizations accumulate workflow because problems accumulate.

An approval exists because something went wrong in 2016.

A report exists because a former executive requested it.

A reconciliation compensates for two systems that do not communicate.

A committee exists because authority is unclear.

AI can make every one of these activities faster.

That does not make any of them necessary.

Eliminate before automating

The correct order is often:

Eliminate → simplify → redesign → automate.

Reversing the sequence can create extremely efficient bureaucracy.

The zero-based test

For every major workflow being targeted for AI, management should ask:

If this process did not exist and we were designing the organization today, knowing AI was available, would we create it this way?

If not, automation should stop until redesign begins.

This is a much harder transformation because it affects roles, controls, reporting lines, incentives and decision rights.

It is also where the larger economic opportunity may reside.

AI has a strategy. Trade has a strategy. Workforce has a strategy. Supply chain has a strategy. The company may not.

This is the deeper pattern running through today’s business headlines.

Artificial intelligence is changing the economics of work.

Trade tensions are changing supply chains.

Geopolitics is changing capital allocation.

Workforce disruption is changing career structures.

Technology is changing strategy.

Succession is changing leadership requirements.

Each issue is important enough to receive its own executive committee, workstream, adviser and presentation.

That creates a new risk.

Organizations can respond intelligently to every individual disruption while producing an incoherent enterprise response.

Everything connects

Suppose management decides to accelerate AI automation.

That affects workforce requirements.

Changing workforce requirements alter real-estate needs.

AI infrastructure increases technology and energy demand.

Automation changes make-versus-buy decisions.

Trade disruption encourages regional manufacturing.

Regionalization changes capital requirements.

Higher capital requirements affect investment priorities.

Investment choices alter acquisition strategy.

The leadership capabilities required to manage the new business begin changing.

There is no clean boundary between these issues.

CEOs need an integrated strategic agenda

The executive committee should periodically place the major disruptions on one page and examine their interaction.

A useful framework might contain six questions.

1. What is changing externally?

Technology, customers, economics, geopolitics, regulation and labour.

2. Which assumptions in our current strategy are affected?

Identify the strategic dependencies.

3. What becomes more valuable?

Capabilities, assets, customers, geographies or relationships.

4. What becomes less valuable?

Legacy processes, facilities, skills or products.

5. Which decisions become urgent?

Not every trend requires action.

6. What should we deliberately not do?

Strategic focus matters even more when disruption multiplies.

From strategic planning to strategic sensing

Traditional strategic planning often assumes management can periodically assess the environment, choose a direction and execute for several years.

The emerging environment requires something closer to a strategic operating system.

The long-term direction may remain durable.

The assumptions beneath it must be tested continuously.

Management should know which indicators matter, which thresholds trigger reconsideration and who owns the decision when circumstances change.