AI between Q4 2024 and Q4 2025.

In the past year I have seen more polished presentations than long-term commitments: sprints abound, sustained goals are scarce. Between the last quarter of 2024 and the close of 2025, artificial intelligence has been portrayed as the magical key to transformation across almost every sector. Yet reality is more complex. Corporate enthusiasm has generated growing investment and spectacular headlines, but not always tangible results. From my position as consultant in global projects, with one foot also in Europe’s mid-sized industry and as an academic, I see a clear tension: the urgency not to fall behind versus the difficulty of integrating AI into business strategy.

What companies actually did in this cycle

Beyond the rhetoric, some recurring patterns have defined AI adoption during this period:

– Rising AI investment, with annual increases between 20 and 30 percent in large corporations. Yet less than 15 percent of projects show a real impact on EBITDA or revenue growth
– Proliferation of pilots and proofs of concept that fail to scale beyond a single department
– Widespread adoption of external foundation models without customisation or own training, creating critical dependence on providers
– Creation of roles such as AI Lead or Prompt Engineer, but without a clear integration architecture with business and operations
– Uneven progress in governance: European companies have responded faster to AI Act requirements, while other markets treat regulation as secondary
– In mid-sized industry, a different pattern: lower absolute investment, but greater pragmatism. AI is applied mainly to administrative automation, process management and internal efficiency optimisation

The European map: sectoral and geographical contrasts

From our experience and contact with regional clients, we observe that AI adoption across Europe reveals important differences that help to explain the landscape:

– Barcelona: centre for video games applied to mid-sized service companies, with strong implantation of multinationals in chemicals and consumer goods
– Lyon and Auvergne: momentum in biotechnology and engineering, with clusters combining academic research and industrial applications of AI
– London: more than 5,800 companies linked to AI, 90 percent of them small or medium-sized. The ecosystem is fragmented but highly dynamic in fintech, legaltech and digital services
– Frankfurt: specialisation in finance and logistics, with intensive use of AI in risk analysis, banking automation and supply chain optimisation

Critique of the hype: running for the sake of running

The main problem of the 2024–2025 cycle has not been lack of business interest, but its excess. Humans have historically learned to manage scarcity, but not abundance—whether of money, work, drugs or, in this case, generative AI capabilities.

Four dynamics explain this excess:
1. The trap of internal marketing: many initiatives were launched more to gain visibility in corporate presentations than to solve real problems
2. The illusion of instant productivity: expectations of immediate ROI that do not hold in practice. AI can cut costs or open opportunities, but requires process redesign and change management
3. Homogenisation of use: most companies rely on the same providers and tools. Without customisation or own training, there is little room for true competitive advantage
4. The flight forward: proliferation of AI departments created in isolation, without connection to operations, marketing or production. This lack of integration limits real business impact

Lessons from mid-sized industry

Mid-sized European industry has shown a more austere but sometimes more effective approach. Recurrent examples include:

– Document automation in logistics, reducing processing time by 20 to 30 percent
– Energy optimisation in manufacturing through consumption prediction systems
– Use of chatbots and virtual assistants in service SMEs, with measurable improvements in customer satisfaction through internal surveys

An academic perspective

In academia, debates emerge that rarely reach boardrooms. Business adoption of AI remains tool-centred rather than culture-centred. Some key issues are:

– Lack of ethical debate beyond regulation: compliance with the European AI Act is important, but does not replace reflection on bias, inclusion or social impact
– The paradox of knowledge: AI multiplies the ability to produce information, but not necessarily to interpret it critically. In many teams, dependence on external tools reduces analytical autonomy
– Recent studies from European think tanks (Bruegel, CEPS, OECD) show that firms that combine AI adoption with internal training achieve more sustainable impacts. Technology without a learning culture does not scale

Looking ahead to 2026: executive recommendations

The challenge now is not to spend more, but to direct better. Some practical recommendations include:

– Define clear metrics before investing. What specific problem does AI solve and how will its impact be measured
– Move from pilot to scale. Identify cases that have already shown results and extend them across the organisation
– Invest in hybrid talent capable of connecting technological vision with business knowledge
– Strengthen governance and risk management, including AI Act compliance, privacy and cybersecurity
– Build differential advantages by customising models and training them with proprietary data
– Develop a realistic narrative for teams and shareholders, avoiding promises that technology cannot fulfil

Conclusion

Between Q4 2024 and Q4 2025 companies have run fast. But what makes the difference is not speed, it is direction. Artificial intelligence is not a sprint, it is a long-distance process that requires strategic clarity, operational realism and continuous learning. 2026 should be the year when initial euphoria gives way to mature integration.

The question is no longer who boards the AI train, but who knows where they want to travel.

References

• McKinsey (2025). The State of AI in 2025.
• Gartner (2025). AI Strategy Survey Q4 2025.
• Accenture (2024). Reinvention in the Age of Generative AI.
• Bruegel (2025). AI adoption in Europe: progress and bottlenecks.
• OECD (2025). AI Outlook 2025.
• European Commission (2024). AI Act Implementation Roadmap.
• Harvard Business Review (2024). When AI Pilots Don’t Scale.
• MIT Sloan Management Review (2025). The Productivity Illusion of Generative AI.
• CEPS (2025). Ethics and Governance in European AI adoption.
• Financial Times (2025). Corporate AI spending rises, but returns remain elusive.
• MIT Technology Review (2025). Generative AI in practice: the end of the hype cycle?
• El País Economía (2025). Europa avanza en regulación mientras las empresas tropiezan con la integración real de la IA.