Artificial intelligence has ceased to be a topic exclusive to technologists and venture capital funds and has become one of the axes of global geopolitical and economic competition. In 2025 and 2026, the race for AI infrastructure — data centres, chips, energy, talent — is mobilising hundreds of billions of dollars in investment, reordering the geopolitics of semiconductors and raising fundamental questions about who will control the computational resources of the next technology cycle.
For companies, funds and institutions, this has consequences that go well beyond the technology sector. AI is beginning to affect industry, energy, defence, finance, healthcare and logistics. And the capacity to access computing infrastructure, high-end chips and specialist talent is emerging as a new form of competitive advantage — or strategic dependence — of the first order.
The AI infrastructure investment supercycle
The major technology operators have committed unprecedented capital expenditure. Microsoft, Google, Amazon, Meta and Oracle have announced plans that, taken together, run to hundreds of billions of dollars in data centres for 2025–2026 alone. To this must be added public investment projects in the United States, Europe and Asia, and sovereign funds repositioning part of their portfolios towards computational infrastructure.
This wave of investment is driving demand for advanced chips — especially Nvidia’s GPUs and their competitors — as well as for energy, industrial land, cooling and specialist talent in language models, distributed infrastructure and cybersecurity.
Goldman Sachs estimated that data centre energy demand could grow by 165% between 2023 and 2030, making AI one of the largest vectors of new electricity demand at a global level.
Goldman Sachs – AI to drive 165% increase in data center power demand
The war for advanced semiconductors
At the heart of the technology race lies control of the most advanced chips. The United States has progressively tightened restrictions on the export of high-end semiconductors to China, citing national security grounds. Rules that began by affecting certain Nvidia chips have been extended to a broader range of components and manufacturing tools.
China, for its part, has accelerated its self-sufficiency programme in semiconductors, with massive investment in companies like SMIC and the development of its own lithography tools. Progress is real but uneven: China can manufacture chips from previous generations at scale, but remains behind in the most advanced processes that require ASML equipment and technologies that export restrictions are making harder to access.
The result is a chip ecosystem that is gradually bifurcating into two architectures: one oriented towards the Western market and another towards the Chinese market. This creates inefficiencies, raises design costs and puts many companies — especially those with simultaneous presence in both markets — in difficult positions regarding which technology platforms to adopt.
The International Energy Agency has warned that global electricity demand from data centres could double before 2030, with direct implications for the energy planning of many countries.
IEA – Electricity 2026
Europe: between regulation and the risk of falling behind
Europe has bet on being a pioneer in AI regulation with the EU Artificial Intelligence Act, which has been in force since 2024. The regulatory approach is ambitious and makes sense from the perspective of rights protection and risk management. But it also generates a legitimate debate about whether this framework helps or hinders Europe’s capacity to compete in the development and industrial-scale adoption of AI.
The European Union has launched initiatives such as AI Factories and computational infrastructure investment plans, but the gap with the United States and China in terms of concentration of private investment and installed capacity remains considerable. Being cut off from the most advanced AI models for regulatory or technological dependency reasons is a real risk that several European governments are already analysing with concern.
What is clear is that Europe needs to advance simultaneously in intelligent regulation, infrastructure investment and talent development if it wants to maintain a relevant position in the AI value chain.
European Commission – AI Act and AI Policy
Which sectors are already feeling the impact of AI
Industry and manufacturing
AI applied to machine vision, quality control, predictive maintenance and process optimisation is changing industrial productivity.
Logistics and supply chain
Route optimisation algorithms, demand forecasting, disruption detection and dynamic inventory management are significantly improving logistical responsiveness.
Finance and insurance
Risk analysis, fraud detection, product personalisation and back-office process automation are all being transformed by language models and machine learning.
Defence and security
AI has become a central component of military modernisation: intelligence analysis, autonomous systems, offensive and defensive cybersecurity, and simulation.
Energy as an unexpected bottleneck
One of the problems attracting most attention is energy. Next-generation data centres consume massive amounts of electricity, and demand is growing at a pace that is putting pressure on power grids in many regions.
In the United States, grid operators and regulators are revising demand forecasts upwards, accelerating the approval of new generation plants — including nuclear — and adjusting connection timescales for large data centres. In Europe, the availability of renewable energy and grid capacity are becoming primary criteria for deciding where to build AI infrastructure.
This creates opportunities for power generation, storage and energy efficiency companies, and elevates energy security analysis to a first-order factor within any AI adoption or investment strategy.
How companies should position themselves
For a company that is not strictly a technology firm, the most relevant message is that AI is no longer a future bet: it is a present competitive variable. Organisations that now build intelligent AI adoption capabilities — in their operations, decision-making processes and knowledge management — will be better positioned over the next five years.
Some sensible positioning levers include:
- identifying internal processes with the greatest potential for automation or improvement through AI,
- building data governance as a prerequisite for any AI adoption,
- assessing technological dependencies and diversifying platform providers,
- incorporating regulatory risk analysis into any AI technology investment or adoption,
- monitoring how competition is evolving in their own sectors to avoid growing exposed to an increasing technology differential.
A race that has barely begun
The speed of AI model advancement, the scale of infrastructure investment and the intensity of geopolitical rivalry for control of computational resources all suggest that we are in the early phase of a long cycle. The next five years will define which companies, countries and regions achieve a relevant position in this new layer of global economic competition.
Ignoring that reality is not strategic neutrality: it is being left out of a conversation that is already reordering competitive advantages across many sectors.