The probability of an AI project delivering measurable return on investment (ROI) is a mere one in five; worse still, the likelihood of achieving genuine organizational transformation plummets sharply to just one in sixty.

We stand at a critical crossroads in history, where the choices made by Chief Information Officers (CIOs) and AI leaders will determine whether their organizations achieve unprecedented success or veer off course. These were the core topics addressed at the Gartner IT Symposium/Xpo, held in early November 2025.

AI undoubtedly holds disruptive potential, yet the reality of delivering sustained value remains a formidable challenge. Gartner’s research reveals a sobering truth: the probability of an AI project delivering measurable ROI is only one in five; worse still, the likelihood of achieving genuine organizational transformation drops drastically to a mere one in sixty.

This yawning gap between ambition and tangible value is deeply rooted in the persistent chasm between technological readiness and human capabilities.

Margrethe Vestager, former Vice-President of the European Commission, contextualized this transformation within a stark historical framework at the IT Symposium, stating: “We are in the middle of an industrial revolution. This is deeply thought-provoking—we are still dealing with the consequences of the first industrial revolution, and that happened over a century ago.”

She emphasized that the current technological shift is “far faster than the first industrial revolution”. Navigating this pace requires confronting formidable systemic barriers related to cost, capabilities, and geopolitical dependencies.

Permanent Over-collateralization: AI’s Hidden Cost

A key barrier to achieving excellence lies in financial opacity. Gartner analyst Gabriela Vogel notes that implementing AI systems differs fundamentally from traditional, predictable one-off investments—such as legacy ERP systems—it is more akin to “transition collateral”: a cost organizations must continue to pay long after deployment.

While organizations have a clear grasp of initial financial outlays—the average company spends approximately €1.63 million on “Day One” costs—this clarity quickly fades. Chief Financial Officers (CFOs) report that they “lose visibility of AI project spending after the initial rollout”, aware of Day One costs but completely in the dark about “Day 100” expenses.

Unpredictable additional costs, coupled with the ongoing demands of AI training and context switching, drive expenses into a spiral. For every AI tool purchased, CIOs must budget for 10 unforeseen supplementary costs that were not included in the original plan.

These hidden transition costs include acquiring new datasets to ensure the AI’s foundational accuracy, as well as managing access to autonomous agents. Most critically, the required human resource investment surges dramatically: change management alone can add an extra 100 to 200 days of work for every 100 days of implementation—equivalent to a staggering 200% increase in effort.

This widespread financial ambiguity directly impacts profit margins, particularly in the Europe, Middle East and Africa (EMEA) region. Gartner analyst Rob O’Donohue highlighted a troubling set of statistics, stating: “In EMEA, 73% of organizations are barely breaking even—or even operating at a loss—on their AI projects.” To avoid inadvertently becoming the bearer of negative-ROI business cases, leaders must conduct rigorous analyses to identify which transition costs warrant funding.

Barriers to Autonomous Supply Chain Agents

Some of AI’s most promising value propositions lie in optimizing supply chains and complex B2B negotiations. However, these ambitious goals are constrained by the immaturity of AI agent technology. Despite high adoption rates—in EMEA, 15% of CIOs have already deployed AI agents, with another 45% planning to do so in the near future—the current focus remains firmly on conversational agents.

Gabriela Vogel warns that conversational agents are inadequate for delivering high-value business outcomes, as they lack reasoning and autonomous decision-making capabilities. Vogel made it clear: if organizations “need them to make decisions—and you really should—then conversational agents are not ready yet. Autonomous multi-agent systems must be equipped with decision-making and reasoning capabilities.”

Retailers and large logistics enterprises are not seeking simple dialogue; their true objective lies in complex tasks such as “multiplexed B2B negotiations”, where agents must continuously track customer purchases, automatically trigger multiple Requests for Proposal (RFPs) to replenish inventory, negotiate intricate terms and conditions, and select optimal suppliers. Such high-stakes, complex scenarios demand specialized autonomous agents with professional capabilities.

When discussing the regulatory landscape for such mission-critical business processes, Margrethe Vestager emphasized that the AI Act imposes “zero intervention” on many important commercial use cases. She confirmed that applications such as “optimizing supply chains, improving logistics operations, managing warehousing, or similar scenarios” fall under the “zero intervention” category. The legislative burden in this domain is minimal, focusing solely on the “obligation to know what you are doing”.

Yet even as technology evolves, leaders must remember that AI cannot replace the human element critical to resilience.

Vestager recalled candidly that when discussing strategies for surviving the pandemic with Chief Executive Officers (CEOs), “nobody mentioned technology. They all talked about relationships—the relationships we have with our suppliers, the people we can count on. The relationships we have with our customers, the ones we can turn to for support.”

She stressed that these critical bonds are forged through “human-to-human communication, finding a balance of mutual interests”—not through “machine-to-machine dialogue”. Consequently, CIOs must strengthen the human relationships underpinning their supply chains, even as they pursue the technical capabilities of autonomous agents.

The Imperative of Digital Nations and Sovereignty

Dependency on major technology vendors—whom Gartner refers to as “digital nations” due to their control over land, power, water, talent, and capital, comparable to sovereign states—poses critical risks to AI sovereignty and global supply chains.

Their scale of spending is staggering: large vendors invest more in AI infrastructure every quarter than the annual Gross Domestic Product (GDP) of 47% of countries worldwide.

O’Donohue warned that choosing an AI vendor is no longer a simple transaction but a life-altering decision, comparing it to “getting married, having triplets, and moving to another country”.

Margrethe Vestager acknowledged that geopolitical instability has forced a shift toward “making independent decisions while maintaining openness”. Citing the semiconductor supply chain as a prime example, she admitted: “Europe can never be fully self-sufficient”, as cutting-edge components and critical minerals are not available on European soil. The strategic goal is to gain “a stronger presence in the semiconductor value chain” while remaining globally engaged.

To mitigate dependency, supply chain diversification is crucial. Vestager urged CIOs to ask themselves: “Am I 100% certain that my supplier will always deliver, or should I diversify slightly so that if one supplier fails, there are others I can rely on?” Such precautions are not only to guard against “bad faith or malice” but also to prepare for unpredictable catastrophic events, such as “a tsunami, or even a fire”.

She emphasized that digital dependency is an ongoing risk, analogous to vulnerabilities in energy and raw materials: “We cannot afford to leave dangerous digital dependencies unaddressed. This is not just a business issue; it is a societal one.”

Furthermore, efforts to advance sovereignty are not a passing trend. Vestager rejected the assumption of a return to “normalcy”, describing the pre-existing state as “stable chaos” which has now evolved into “accelerated unpredictable chaos”. Because “the new holders of power will say, ‘Oh, the situation has changed. We can do more than in the world of the past’”, the structural demand for resilience will persist.

To strengthen domestic capabilities, Vestager proposed allowing national governments to include “Made in Europe” provisions in public procurement to stimulate the necessary demand. She called on private-sector pioneers to “bear some of the costs to achieve this goal” and develop genuine European alternatives, particularly in critical services such as cloud computing, while acknowledging the need to balance short-term risks with long-term resilience.

Addressing AI sovereignty requires strategies such as tokenization to anonymize data, ensuring that sensitive real data “never leaves your borders, even when used in models”. Gartner predicts that by 2027, 35% of countries will be locked into region-specific AI platforms that utilize proprietary contextual data; implementing such preventive measures is therefore critical to avoiding lock-in risks.

The value of AI will remain out of reach if technological readiness continues to outpace human readiness. As Vogel warned: “The rate at which AI readiness is growing is far faster than the rate at which human readiness is improving. If all vendors stopped AI innovation today, we would still need several years to catch up.” When technological readiness is high but human readiness remains low—a state many organizations currently find themselves in—delivering value becomes extremely challenging.

Editor’s Note: Readiness refers to the level of capability and willingness demonstrated by an individual when performing a task. The concept was established in the Situational Leadership Theory proposed by Dr. Paul Hersey and Kenneth Blanchard. It spans fields such as disaster psychology, public health emergency response, and government data openness, encompassing two dimensions: capability (knowledge, skills) and willingness (motivation, confidence), with four levels ranging from R1 to R4.

Succeeding in this industrial revolution requires leadership capable of driving profound cultural change. Leaders must not only define what AI can do, but also clarify what humans must do. As Ms. Vestager concluded, leading teams in complex environments demands continuous communication to ensure members understand the goals of their work—a communication that “to some extent, also unites minds”.

By focusing on augmenting rather than replacing human capabilities, leaders can guide their organizations onto the golden path to sustainable value, ensuring that in this era of accelerated, uncertain chaos, they do not merely navigate technological change, but spearhead the essential human transformation required to thrive.