AI Sets a New Mandate for the Modern CEO
Fresh KPMG research signals a clear shift: AI is now a board-level operating priority, not a side project. US CEOs are upbeat on growth-86% confident in the country and 84% in their own companies-yet 79% have already changed strategy this year. That mix of confidence and decisive movement is the new norm.
As Tim Walsh, Chair and CEO of KPMG US, puts it, "The CEO role is evolving and increasingly multi-dimensional as expectations become more expansive. But CEOs are leading with confidence, leaning into new skillsets and adaptability as they position their organisations for success in years to come."
Confidence with a bias for action
Optimism alone isn't a plan. The data shows leaders are adjusting quickly to market pressure and shifting customer behavior. The CEOs who win will keep optionality high-scenario planning, flexible budgets, and sharper portfolio decisions-while moving fast on AI where value is clear.
AI moves from experiment to a budget line
Seventy-four percent of CEOs rank AI as their top investment to handle economic uncertainty, and 69% plan to allocate 10-20% of next year's budget to AI. Yet only 13% feel "very confident" they're ahead on adoption. The gap is execution.
KPMG's takeaway is direct: improve data quality, partner well, and put real money behind agentic AI to capture exponential value. That means funding data pipelines, standing up repeatable patterns (not one-off pilots), and securing the right strategic vendors.
Risk is rising-and so are expectations
Risks are stacking up. Eighty-nine percent say tariffs will materially affect performance in the next three years. Eighty-four percent worry an AI-native player will displace an incumbent in their industry. The message: defend and build at the same time.
- Fraud detection and prevention is an extreme concern for 65% of US CEOs
- Identity theft and data privacy are priorities for 52%
- Forty-five percent are actively addressing vulnerability to cyberattacks
The smart move: use these pressures to accelerate work on digital resilience and supply chain redesign, and adopt a recognized AI risk framework to keep trust front and center. For a practical reference, see the NIST AI Risk Management Framework.
The hybrid AI-human workforce is here
More than half of CEOs are concerned about the cultural impact of AI. They're right to be. In the near term, 86% expect managers to oversee multiple AI agents as part of their roles.
This isn't just technical integration. It's management design. You'll need clear governance, transparent decision rights between people and agents, new performance metrics, and a plan to upskill managers who will supervise AI systems-just like they supervise teams.
What to do in the next 90 days
- Pick 3-5 high-leverage, automatable use cases tied to P&L (e.g., sales ops, FP&A, fraud, customer support). Fund them end-to-end, not as pilots.
- Upgrade data quality where it hurts outcomes most (labels, lineage, access controls). Tie budget to measurable accuracy and latency targets.
- Appoint a single executive owner for AI governance with authority over standards, approvals, and incident response.
- Run one focused agentic AI pilot (clear work boundaries, human-in-the-loop, audit trails). Measure cycle time, error rate, and user adoption.
- Consolidate your vendor mix. Standardize on a small set of model providers, vector databases, and orchestration tools to cut integration time.
- Tighten cybersecurity around AI: secrets management, model access logs, prompt injection testing, and red-team exercises.
- Model tariff and regulatory scenarios. Stress-test pricing, sourcing, and inventory plays now, not later.
- Define the operating model for managers overseeing AI agents: span of control, escalation paths, and compensation aligned to outcomes.
- Publish "AI guardrails for teams" in plain language so employees know what's encouraged, what's restricted, and how to get help.
Metrics that keep you honest
- Percent of revenue influenced by AI-enabled products or processes
- Cycle-time reduction in top processes (quote-to-cash, month-end close, claims)
- Accuracy and exception rates for AI-assisted decisions
- AI incident rate and time-to-contain
- Data quality error rate in priority datasets
- Share of teams with an AI-literate manager and documented agent oversight
- Security posture improvements tied to AI use (e.g., phishing detection efficacy)
Why this matters now
AI-native challengers won't wait. The incumbents that move first-on data, agentic workflows, governance, and talent-will separate from the pack. As Bill Thomas, CEO of KPMG International, said, "Ultimately, the leaders who can embrace market volatility and focus investments in the right strategic areas for their organisation will be the ones best placed to unlock new opportunities and build sustainable long-term growth."
Skills and training for your teams
If you're standing up AI programs across finance, operations, or marketing, equip your people with focused training. A practical place to start is role-based learning paths: see Courses by Job at Complete AI Training for targeted options that map to common enterprise use cases.
Your membership also unlocks:
