By Javier Surasky
What if the next technological revolution didn’t just transform our lives but also shape the future of global equity? The European Union (EU) has taken a bold new step in its quest to lead in artificial intelligence (AI) by setting in motion a process that will culminate in the adoption of its AI Continent Action Plan, a complement to the EU AI Act, in force since August 2024.
With
countries like Taiwan dominating the production of AI chips, others leading in
specific hardware niches, and Africa and Latin America and the Caribbean
providing critical natural and energy resources for AI operations, can the EU
balance its race against giants like the U.S. and China without integrating
external poles beyond the Union? Can the EU ignore the debates and frameworks
established at the UN without weakening multilateralism and harming itself?
This blog examines how the EU’s AI strategy aims to transform the global landscape, its strengths, and its limitations, while seeking to address these
questions.
Let’s start
by recalling that during the AI Action
Summit in Paris in
February 2025, European Commission President Ursula von der Leyen unveiled the
AI Continent Action Plan, a roadmap to transform industries and nurture
European talent to drive AI innovation.
The Plan
has five pillars: computational infrastructure, data access, AI adoption,
skills development, and regulatory simplification. It promotes initiatives like
“AI Factories” and “Gigafactories,” backed by a €200 billion investment through
the European InvestAI program. This new Plan builds on the EU AI Act, where
establishing ethical AI governance and fostering public and investor
trust—primarily through a risk-based but industry-friendly regulatory
approach—are key elements.
AI
Factories are not “factories” in the traditional sense but rather ecosystems
for production and research that combine high-performance computing with
quality big data and specialized human talent. So far, 13 AI Factories have
been established in Germany, Austria, Bulgaria, Slovenia, Spain, Finland,
France, Greece, Italy, Luxembourg, Poland, and Sweden.
Conversely,
Gigafactories are facilities equipped with 25,000 to 100,000 cutting-edge
processors and massive storage systems, optimized for training complex AI
models.
This EU
positioning has an unspoken backdrop: Europe’s lag behind the AI advancements
of the U.S. and China, as well as its comparative weakness in competing for
resources produced in third countries, such as critical hardware—from chips to
semiconductors—and access to natural resources essential for an accelerated
race toward AI development and intensive use.
Against
this reality, the EU seeks to position itself as an option that prioritizes
clear and robust governance rooted in ethics and data safety. While the U.S.
imposes tariffs to incentivize foreign technologies to be produced on its soil
and restricts the export of those primarily developed domestically—disrupting
pre-existing global trade agreements—the EU appears to follow a similar
strategy, but instead of tariffs, it relies on data access restrictions for EU
citizens’ data, justified by axiomatically positive values like security and
citizen protection.
But is the
new AI Continent Action Plan properly aligned with the principles of the EU AI
Act? And what does it mean for the world beyond Europe’s borders?
The first
thing we notice when examining the Plan and the Act is their strong coherence,
promoting an innovation framework interwoven with ethical considerations.
Thus, the
AI Act’s classification of AI systems by risk level (prohibited, high-risk,
limited risk, minimal risk), with different requirements for each group, is
integrated into the Action Plan through five major lines of work:
- Regulatory simplification
through the establishment of an AI Act Support Service to assist
companies, especially SMEs, in complying with the Act’s requirements.
- Ethical AI development
supported by AI Factory programs and Data Labs, aimed at creating AI
models that meet the AI Act’s requirements on data quality and the
reduction (or elimination, where possible) of biases.
- The promotion of key AI sectors
is reflected in the new Plan by encouraging AI adoption in critical areas,
with schemes ensuring that developed applications meet the standards
required for their respective risk categories under the AI Act.
- Data governance, where the Plan
aligns with the EU’s
General Data Protection Regulation (GDPR) of 2018 and the relevant
provisions of the AI Act to ensure interoperability, compatibility, and
privacy in data use.
This
regulatory alignment, however, leaves room for potential tensions: it remains
to be seen how the Plan’s push for rapid innovation—fueled by the concept of AI
Factories and the goal of tripling the operational capacity of data centers in
Europe—will balance with the ability of smaller, resource-constrained companies
to meet the AI Act’s demands, such as conducting risk assessments for AI
systems.
In a
familiar, unresolved tension, the level of data protection provided by the EU
framework may end up slowing the availability of data needed for an ambitious
data collection and management program, such as the establishment of Data Labs
and a 300% increase in data centers.
All of this
is part of Europe’s effort to maintain a privileged position in the global race
to dominate or capitalize on AI’s benefits, where the U.S. leads in innovation,
China leads in AI deployment, and other actors play intermediate roles in a
complex, interconnected global digital ecosystem. States in Asia and the Middle
East lead in hardware and software niches, while countries in Latin America and
the Caribbean and Africa serve as suppliers of the natural and energy resources
indispensable to keep the system running.
Here, a new
tension emerges that the AI Continent Action Plan fails to resolve: by
positioning itself as a leader in ethical and secure AI regulation, the EU
risks falling behind faster-moving, less regulated markets like the U.S., or
highly centralized, state-driven, and subsidized ones like China. It also risks
lagging behind specialized niches, such as Canada’s leadership in deep learning
research, Singapore’s advancements in AI for smart cities, Israel’s military AI
applications, or Japan’s robotics, among other notable cases.
The EU
currently lacks its own “niches” of competitive advantage in AI, whether in
software or hardware, and it’s doubtful that a purely “ethical advantage” will
grant it greater weight in the race for AI development and adoption. Even more
worry, the EU has a deficit in required natural and energy resources.
This last
point raises questions about how the EU’s AI Continent Action Plan addresses
countries that supply natural resources, primarily developing nations. The
answer is startling: neither the AI Act nor the Action Plan tackles this issue,
despite the fact that, in the current state of technological evolution, lithium
from Argentina, Chile, and Bolivia, and copper from Peru and Chile, are
indispensable to sustain the growth the EU aims to achieve in the digital
realm. The same could be said of rare earths and other minerals from various
African and Latin American and the Caribbean countries.
Could a
win-win exchange mechanism be created, where resource-producing countries
partner with the EU, trading natural resources for technical and human capacity
in AI or digital services? Nothing suggests this path is impossible, and it
could create strengths on both sides of such an alliance. It’s worth noting
that these natural resources are also of interest to the U.S. and China, so the
EU shouldn’t wait for its “competitors” to act first.
For certain
countries in Latin America, the Caribbean, and Africa, Europe could be a
preferred partner in this regard, as it demonstrates a greater commitment to
environmental sustainability than the other two major contenders in the AI
race, even if European standards still have room for improvement.
In fact,
the European Plan’s failure to include a chapter on sustainability in supply
chains, ignoring the impact of its measures on the regions supplying the natural
resources needed to make it operational, is a poor start and a major political
misstep.
This
becomes even more serious when we see that neither the AI Act nor the Plan
adopted by the EU reference the work of the UN’s High-Level Advisory Body on
AI, UNESCO’s Recommendation on the Ethics of AI, or the 2030 Agenda for
Sustainable Development. They don’t even build connections where their contents
mutually reinforce each other, which happens on several occasions.
Currently,
three public consultations are underway in the EU to shape the final version of
the AI Continent Action Plan: one on the Cloud, another on the European AI
Strategy, and one
on Gigafactories. In May 2025, a third consultation
on the EU’s data strategy will open. These consultations offer an opportunity to address the identified tensions. They democratize the discussion and offer a
chance to rethink the elements we’ve highlighted, for the benefit of the EU
itself, but also for third countries that could become allies if the right
incentives are created.
The EU
needs to revisit its colonial and self-centered thought patterns, which
persist, to incorporate perspectives that consider how the interests and rights
of third parties can, when addressed, bolster the legitimacy of its ambitions
and strengthen its leadership opportunities. It is not a zero-sum game—quite
the opposite. To grasp this, European leaders must broaden their frameworks of
thought, and no artificial intelligence can help them do so, as what is needed is precisely to think more humanely.