Investment in artificial intelligence (AI) is growing rapidly, plans for new data centers continue to multiply, and pressure on corporate leadership to define AI strategies is intensifying. Even so, large-scale adoption still falls short of expectations. The challenge is not the technology itself, but the difficulty of turning it into consistent operational value.

According to McKinsey data, only 7% of companies say they have implemented AI across their organizations at scale. Most remain in the early stages, with pilot projects or limited applications. This gap between expectation and execution raises questions about whether companies can absorb the technology at the pace markets anticipate.

Market expectations and the risk of misalignment

The debate around a possible “AI bubble” has gained traction. Not because the technology lacks value, but because markets may be overestimating the speed of adoption. According to Goldman Sachs Research, concerns about high valuations and capital-intensive investment resurfaced in late 2025, driven by a growth narrative that may not fully reflect business reality.

Technology evolves quickly, but organizational transformation is structurally slower. It requires operational change, internal alignment and sustained investment. When the pace of investment outstrips the ability to execute, a tension emerges that can undermine the sustainability of the growth cycle.

Infrastructure is expanding faster than demand

Infrastructure expansion does not, by itself, guarantee effective adoption. A recent example reported by Reuters highlighted Oracle and OpenAI scaling back a data center project in Texas after financing difficulties and changes in operational requirements.

This kind of situation reveals a mismatch between investment enthusiasm and actual business demand. It is also compounded by physical constraints, such as the heavy energy and water consumption of data centers, as well as delays in connecting projects to power grids, all of which can increase costs and slow deployment.

SMEs are advancing only in limited ways

Adoption is even slower among small and medium-sized enterprises. OECD data suggests that although the use of generative AI is growing, only a minority of SMEs are integrating it into their core operations. In most cases, the technology is being used in peripheral tasks such as content production, administrative support or customer service.

This pattern reflects an approach focused on incremental efficiency rather than structural transformation. Among the main obstacles are a lack of internal capabilities, fragmented data, budget constraints and uncertainty about return on investment.

The bottleneck is organizational

The central difficulty is not technological, but organizational. Many companies are trying to fit AI into existing models rather than redesigning processes and workflows around it. According to McKinsey, only 1% of organizations believe they have reached maturity in this area.

On this point, Microsoft CEO Satya Nadella recently emphasized that successful AI adoption depends on “mindset, skills, tools and data.” The order of those factors is revealing: companies tend to start with the tools, while overlooking the fact that the main challenges lie in organizational culture and internal capability-building.

Digital transformation is still incomplete

Another structural factor is that many companies have still not completed their digital transformation. This is especially true for SMEs, where weaknesses remain in data systems, technology integration and workforce capability.

In this context, AI is being introduced on fragile foundations, making it harder to scale effectively. Successful adoption requires structured data, clearly defined processes and investment capacity — conditions that are not always in place.

Experimentation without strategic direction

Despite the increase in AI initiatives, many remain tactical in nature. Companies are testing solutions, launching pilots and exploring tools, but without clearly defining strategic objectives. Too often, they have yet to identify how AI can improve margins, productivity or competitive advantage.

The result is a significant amount of experimentation with limited impact on overall business transformation.

Between enthusiasm and execution

AI adoption continues to fall short of expectations because it is often treated as a technology purchase rather than an organizational transformation process. Capital is available and infrastructure is expanding, but companies’ internal capabilities are not keeping pace.

The current picture reveals a two-speed economy: accelerated investment at the top, and slow, uneven adoption at the base. Until organizations strengthen their structures, data and processes, the promise of AI will remain distant from operational reality.