[AI shock] AI becomes a growth litmus test as returns pressure mounts, outcomes hinge on strategy

Global CEOs’ confidence in revenue growth sinks to a five-year low
Pressure mounts to show returns on AI spending, with few firms seeing real financial gains
Rushed AI rollouts backfire, hurting service quality and driving up costs

Global corporate CEOs’ confidence in revenue growth has fallen to its lowest level in five years. As pressure to show results from artificial intelligence (AI) investment intensifies, expectations have sagged—especially among companies that have failed to translate AI spending into financial improvement. Experts say firms that rushed to adopt AI without first building the fundamentals are increasingly facing a backlash after repeated failures to boost operational efficiency.

CEOs lose confidence amid rising uncertainty

According to Reuters on the 20th (local time), PwC, a global accounting and consulting firm, released the results of a survey conducted late last year of more than 4,000 CEOs across 95 countries and territories. The data show that only 30% of CEOs said they were confident about revenue growth over the next 12 months, a sharp drop from 38% in 2025 and the lowest level of confidence in the past five years.

PwC said management uncertainty has reached an extreme as geopolitical instability and fears of trade wars collide with another challenge: recouping AI investment. Many companies, in fact, are struggling to generate financial results from AI spending. In the survey, just 12% of CEOs said AI had contributed to both cost reductions and revenue growth. Another 33% said they had seen benefits in either costs or revenue, while 56% said they had yet to confirm any meaningful financial impact. The figures suggest that while AI adoption experiments are spreading quickly, only a small number of companies have managed to embed AI into their business models.

Performance also varied clearly depending on how AI was deployed. CEOs at companies that experienced gains in both costs and revenue said they apply AI broadly across products and services, demand generation, and strategic decision-making. By contrast, companies whose adoption remained at the pilot stage often reported little or no financial payoff—highlighting how the breadth and depth of AI use are increasingly translating into a widening performance gap.

Why AI adoption outcomes are diverging

Mohamed Kande, PwC’s global chair who oversaw the survey, recently shared his view of the current market in an interview with Fortune in Davos, Switzerland. He said the CEO role has changed markedly over the past year, arguing that a “tri-modal” challenge has emerged in which leadership must do three things at once—run the current business, transform that business in real time, and build an entirely new business model for the future—adding that he has not seen conditions like this over the past 25 years.

Kande said companies went through particularly sharp change as the period from 2024 into 2025 unfolded. “Not long ago, the question was whether AI can be adopted and whether it should be,” he said, adding that “now no one asks that question—everyone is jumping in.” He pointed to the breakdown of “fundamentals” such as data readiness, business processes, and governance as a core reason many firms are struggling to see results. “AI moves too fast, and that speed has made people forget the basics of implementation,” he said. The key to AI adoption, he argued, is not the technology itself but execution—and execution ultimately comes down to strong management and leadership.

He also stressed that the shifting CEO role is reshaping organizations. With the traditional apprenticeship model beginning to break down, he said companies will need to redesign career paths. In other words, the career ladder in which new hires build experience through routine tasks is being disrupted by AI. He said AI will take on a significant share of “doing the work,” and urged people to learn system thinking rather than focusing solely on task execution—arguing that training and growth paths on the job will have to change accordingly.

Global industry cases of AI adoption

The gap in “fundamentals” cited by Kande becomes clear when looking at real-world industry examples. Big Tech companies with the infrastructure to deploy AI efficiently have improved productivity after adoption and are moving ahead with aggressive layoffs. Amazon said in October last year that it would cut 14,000 jobs—the largest reduction in its history—to focus on future investments including AI. Microsoft, which cut a total of 15,000 jobs the same year, saw CEO Satya Nadella stress the need to “reset the company’s mission for the AI era.” Salesforce has also reduced 4,000 customer support roles using AI, saying the technology is already handling about 50% of the company’s work.

Microsoft has publicly stated that it has driven gains in both productivity and revenue by rolling out AI across the organization. Judson Althoff, Microsoft’s chief commercial officer, said at an internal event last July that AI adoption alone saved more than $500 million in customer service operations, while lifting both employee and customer satisfaction. He added that AI has automated early-stage consultations and sales processes for small and midsize businesses, already generating tens of millions of dollars in revenue. The claim is that AI is no longer just replacing tasks, but delivering tangible revenue growth and time savings across sales, customer service, and software development.

By contrast, there are also cases where AI adoption has led to higher costs and declining service quality. Australia’s Commonwealth Bank replaced customer service staff with AI chatbots in July 2025 and laid off 45 employees, only to reverse course after facing higher call volumes and worsening staff fatigue. U.S. fast-food chain Taco Bell also introduced voice AI at drive-throughs, but moved to reassess its use after a surge in customer complaints. Such failures are increasingly common. A report published last May by organizational design platform Orgvue found that 55% of companies that replaced workers with AI later concluded that the rollout had come too early.

Experts attribute these problems to what they call the “doorman fallacy”—the mistaken belief that simplifying or automating employee roles will solve all issues. On this point, Gediminas Lipnickas, a marketing lecturer at the University of South Australia, said that while AI adoption judged purely on efficiency may reduce short-term costs, it can damage customer experience and long-term performance. “AI delivers the greatest impact not when it replaces humans, but when it is combined with human judgment,” he said.