65 Percent of CEOs Worldwide Fear They're Investing Too Little in AI, Cisco Survey Finds

Two-thirds of CEOs at the world's largest companies admit their organizations are investing in artificial intelligence too slowly. That's according to Cisco's latest survey, which polls thousands of CEOs each year about their approach to AI. The share of those concerned rose by 12 percentage points over twelve months, showing that pressure to speed up deployment is growing faster than companies' ability to keep pace.

Mounting pressure to move faster

Cisco's AI Readiness Index has now run this survey for a second consecutive year, making it possible to compare CEO sentiment year over year. The result is clear: optimism about AI's potential is rising, but so is the fear of missing the moment. In 2026, 69 percent of respondents consider AI adoption a prerequisite for running a modern business, not an optional add-on to strategy.

Interestingly, uncertainty about competence has dropped at the same time. The share of CEOs admitting their own understanding of AI limits the decisions they make fell from 74 to 47 percent, while the share citing a lack of knowledge as a decision-making barrier dropped from 74 to 49 percent. CEOs feel more confident talking about AI than they did a year ago, but that confidence hasn't yet translated into a sense that their company is investing fast enough.

Three barriers to deployment

The survey identifies three specific obstacles that most often slow AI scaling at large organizations. The first is infrastructure: 53 percent of CEOs fear their current network and computing resources won't support the scale of planned deployments, and only 22 percent rate their infrastructure as fully ready.

The second barrier is data. 34 percent of respondents cite fragmentation and dispersion of data as the single biggest obstacle, more than any other factor in the survey. Only 19 percent of organizations report full, centralized access to company data for their AI systems, meaning most companies are feeding their models an incomplete or scattered picture of their own business.

The third and newest concern is security and control over autonomous AI agents. As more companies deploy agents that make decisions without ongoing human oversight, the question of who is responsible for their mistakes looms larger. Only 31 percent of CEOs today feel equipped with the tools to actually control what these systems do.

Move fast with AI, but always anchor decisions in human values - one of the CEOs surveyed, Cisco AI Readiness Index 2026 study

What it means for business

The study's authors note that organizations that have managed to solve the infrastructure, data, and agent-control problems simultaneously are gaining a clear competitive edge over the rest of the market. That confirms a pattern also visible in other analyses this year: the gap between companies that use AI and those that actually see financial returns from it comes mainly from technological foundations, not simply from the decision to buy a model license.

For Polish companies, Cisco's findings serve as an important benchmark, since domestic surveys have for months shown a similar pattern of deployment lagging behind stated ambitions. Companies in Poland report disappointment with the results of their AI projects so far more often than the EU average, while at the same time planning to increase budgets for the coming year, in line with the global trend Cisco describes.

The survey also signals a shift in boardroom priorities for 2026. Modernizing network and computing infrastructure to handle the loads generated by AI models is becoming CEOs' top priority, overtaking earlier concerns focused mainly on talent and employee training. That signals the phase of experimenting with individual AI tools is giving way to a phase of investment in the hard infrastructure needed to scale AI across an entire organization.