The Logicalis 2026 Global CIO Report surveyed over 1,000 technology leaders worldwide. The numbers paint a picture of an industry moving faster than it can think.
There is a version of the AI adoption story that sounds like success. Companies are investing. Proof-of-concept projects are delivering results. Budget is flowing. The technology is working well enough that over a third of organizations have accelerated their AI initiatives based on early wins.
Then there is the other version of the same story.
More than half of those organizations believe their AI adoption is already moving too fast. Eight out of ten have compromised on governance because they did not have the knowledge to do it properly. More than half say they do not fully understand the risks of the AI they have already deployed. Only 39% are confident they are actively managing the environmental impact of their AI workloads. And 16% have no continuity plan if a key AI provider becomes unavailable.
Both versions come from the same report. The Logicalis 2026 Global CIO Report, published in March, surveyed over 1,000 chief information officers worldwide. The gap between those two stories is the most important thing in it.
The Numbers Worth Sitting With
A few statistics from the report stand out not because they are surprising but because of what they reveal about the structural conditions of enterprise AI adoption right now.
Eighty-nine percent of organizations describe their current approach to AI as “learning as we go.” This is an honest description of how most technology transitions unfold, and there is nothing inherently wrong with iterative learning. The concern is what it implies when combined with the governance figures: organizations are making consequential decisions about AI deployment without a clear framework for evaluating the risks of what they are deploying, and they are doing so at speed, because competitive pressure leaves little room for deliberation.
Sixty-two percent of CIOs report compromising on AI governance specifically because of limited knowledge, not because they decided governance was less important, but because they did not know enough to do it well. This is a different kind of problem than organizations that consciously deprioritize governance for business reasons. It is a skills deficit presenting as a governance deficit. The Logicalis data is consistent with other research in this space: a separate Dun and Bradstreet survey found that 97% of organizations have active AI initiatives while only 5% say their data is ready to support them.
The 16% with no continuity plan if a key AI provider becomes unavailable is a quiet systemic risk. Organizations that have built workflows, products, or internal processes around a specific model or API provider are exposed to a kind of single-point-of-failure risk that most IT governance frameworks were not designed to handle. This is not a hypothetical scenario. Models are deprecated, APIs change, providers shift pricing structures. The organizations that have not modeled what happens if their primary AI dependency becomes unavailable are, in the meantime, taking on that exposure without measuring it.
Why the Constraint Is Skills, Not Budget
The report’s most structurally important finding is the identification of skills as the primary constraint on AI adoption, ahead of funding, infrastructure, and regulatory uncertainty.
A lack of internal technical capability is holding back AI ambitions in nearly nine out of ten organizations. This is a notable reversal from the framing that has dominated AI adoption discussions for the past two years, where the limiting factor was assumed to be access: access to capable models, access to compute, access to data pipelines. All of those constraints have loosened significantly as the model landscape has matured and cloud infrastructure has democratized access to AI capabilities.
The new constraint is the ability to evaluate, configure, govern, and maintain AI systems responsibly. The technology has outpaced the organizational capacity to use it well. This is not a criticism of individual organizations. It reflects the pace at which the landscape has moved. The time from “large language models are a research curiosity” to “large language models are embedded in enterprise workflows” was, measured historically, extremely short. Human expertise does not scale at the same rate as model capability.
The Salesforce 2026 CIO survey, conducted independently, found a complementary pattern: CIOs describe the most important skills for the AI era not as technical capabilities but as leadership, storytelling, and change management. The role of the technology leader is shifting from building and owning systems to orchestrating AI capabilities that are increasingly provided by external providers, while managing the organizational change that adoption requires. That is a different job than the one most CIOs were trained for.
The Governance Problem Is Structural, Not Just Cultural
It would be comfortable to frame the governance gap as a cultural problem: organizations that are not taking safety seriously enough, cutting corners, prioritizing speed over responsibility. The data does not really support that frame.
Seventy-six percent of CIOs describe unchecked AI as a serious concern. They are not ignoring the issue. They are proceeding despite concern, because the competitive and organizational pressures to move forward outweigh the capacity to slow down and build proper frameworks. The constraint is not willingness. It is capability and time.
The Cisco 2026 Data and Privacy Benchmark Study found consistent results from a different angle: 93% of organizations plan further investment in AI governance, recognizing that their current frameworks are inadequate. This is an industry that is broadly aware it is outrunning its governance capacity and is committed to catching up while simultaneously accelerating.
That combination, knowing you are outrunning your safety frameworks and continuing to accelerate anyway, is worth examining directly. It is not irrational. In a competitive market, a company that slows down to build governance while its competitors deploy and learn has a different risk profile than one that deploys and builds governance in parallel. The collective action problem is real. The industry cannot simultaneously slow down to govern properly and maintain the pace of competition. Individual organizations cannot unilaterally solve this.
This is one of the structural arguments for external regulation. When the governance deficit is industry-wide and the competitive dynamics make unilateral restraint economically costly, the mechanisms that can shift the equilibrium are not internal to individual organizations. They are external: standards, regulation, liability frameworks, auditing requirements. The EU AI Act, currently in its enforcement run-up, is one attempt to provide that external structure. The debate about whether it is the right structure is ongoing, but the Logicalis data makes a reasonable case that some external structure is needed, because internal governance is demonstrably not keeping pace with adoption.
The Environmental Dimension Nobody Is Measuring
One finding in the report tends to receive less attention than the governance numbers, but it deserves more: only 39% of CIOs are confident their organization actively manages the environmental impact of AI workloads, and only 41% are confident that energy efficiency is prioritized in AI deployment decisions.
This matters because AI workloads are energy-intensive in ways that are often invisible to the organizations using them. When a company runs inference through an API, the energy cost sits with the cloud provider. It does not appear in the company’s energy reporting, its sustainability disclosures, or its carbon accounting. The externalization of AI’s energy footprint is not unique to AI, but the scale of the workload is growing fast enough that the invisible environmental cost is also growing fast.
The organizations that are neither measuring nor managing the energy impact of their AI deployments are carrying an exposure that will become more visible as disclosure requirements tighten and as AI’s aggregate energy demand becomes a more prominent public issue. Logicalis found that 94% of CIOs plan to lean on managed service providers to help navigate AI governance, scaling, and sustainability over the next two to three years. That number suggests awareness of the problem. The 39% figure suggests that awareness has not yet translated into practice.
What This Moment Actually Looks Like
The picture the Logicalis report paints is not of an industry recklessly ignoring risk. It is of an industry caught between real competitive pressure to adopt AI quickly and real structural constraints in the skills, frameworks, and time required to do it responsibly.
The phrase “learning as we go” is accurate, and in one sense it is fine. Every major technology transition involves learning as you go. The question is whether the learning is happening fast enough relative to the consequences of the mistakes being made along the way, and whether the organizations bearing the consequences of those mistakes are the same ones making the decisions, or someone else.
The 44% who say they fully understand the risks of AI they have deployed are not necessarily right. They may just be the ones who have thought about it enough to feel confident. The honest number may be lower.
Sources:
PR Newswire. Logicalis 2026 CIO Report press release. March 3, 2026.
Salesforce. “AI Adoption Skyrockets 282% as CIOs Enter the Era of Scale.” November 2025 / 2026.
CIO Magazine / Cisco. “The struggle for good AI governance is real.” February 24, 2026.