The race for Artificial Intelligence (AI) is no longer just a software competition. It increasingly resembles a global public works project: energy, chips, networks, data centers, and new industrial plants that, according to Jensen Huang, CEO of NVIDIA, will mark “the largest wave of infrastructure building” seen so far. This idea, repeated during his Davos appearances, encapsulates a pivotal era: the world is building “AI factories,” and the scale of this effort will be measured in trillions.
This is not an empty slogan. Huang describes AI as a platform composed of multiple layers that must scale simultaneously. Better models alone aren’t sufficient: without abundant electricity, semiconductors, cloud capacity, and industrial applications, the promise remains in laboratories. His argument is that, for the first time, digital innovation is directly driving the physical economy: it requires constructing facilities, deploying power, cooling, high-capacity networks, and logistics chains that cannot be improvised.
Within this framework emerges the notion of an “AI factory.” In practice, it would be a center for producing intelligence: infrastructure that converts energy and computing into outcomes (models, inferencing, automation, and, from a business perspective, productivity). The industrial analogy is intentional. If data centers have long been considered the “storage” of the internet, these new installations aim to operate like manufacturing plants—optimized for training and deploying AI systems at scale, with continuous improvement cycles and a critical dependency on energy efficiency.
The discourse also seeks to address the question that has been looming over AI for months: will it destroy jobs? Huang argues that infrastructure demand alone acts as a counterbalance. He doesn’t just mention engineers; he emphasizes trades and technical profiles supporting deployment: electricians, installers, construction workers, metallurgists, network specialists, operations, and maintenance. According to him, AI’s economic impact creates immediate pressures on the labor market because “the urgent need” isn’t just writing code but building and operating large-scale physical installations that consume significant energy and require extreme reliability.
This vision is supported by ambitious figures. Huang has linked the deployment of this new infrastructure to an economic impact estimated at up to $85 trillion over the next 15 years. The core message is that current spending—though already enormous—would simply be the beginning: “just a few hundred billion dollars” within a much larger investment curve. In other words, what currently seems like a computing spending bubble in his narrative would be the prelude to reindustrialization affecting entire sectors.
Europe plays its own role on this map. Huang has pointed out that the continent retains a strategic advantage: its industrial base. His recommendation is direct: invest early to “merge” manufacturing capacity with AI, robotics, and what he calls “physical AI” (intelligent systems acting in the real world). He argues that a significant part of future competitiveness will be decided here, but only with the prerequisite of sufficient energy and infrastructure to sustain this ecosystem.
Meanwhile, market signals already suggest that this race is moving into concrete infrastructure. During December 2025, new data center announcements in Europe revealed a clear pattern: regulatory approvals in Ireland for new connections in Dublin; permits for 73 MW projects linked to AWS; multimillion-dollar partnerships to develop campuses in Frankfurt, Amsterdam, and Paris; and corporate moves indicating consolidation and scale. Spain, in particular, has seen tangible acceleration—from announced projects with hundreds of megawatts planned to investments worth billions and new campuses across various regions, with Madrid emerging as a recurring focal point.
Additionally, NVIDIA is trying to translate its vision into a methodology. In its latest communication, the company presented “blueprints” for gigawatt-scale AI factories supported by digital twins, automation, and continuous optimization systems. The promise is that these facilities won’t just be built faster but will operate as “learning” systems: software and agents adjusting consumption, cooling, and workloads to maximize every watt and GPU.
However, the challenge isn’t purely technological. The concept of AI factories clashes with real-world constraints: electricity availability, permits, construction timelines, supply chain tensions, and reliance on a few critical manufacturers. It also raises uncomfortable questions for companies and governments regarding who controls this infrastructure, where data resides, and the extent to which this new industrial muscle will be concentrated in a handful of regions and suppliers.
Nonetheless, Huang’s diagnosis is compelling because it shifts the conversation: AI will no longer be just a productivity revolution in offices but a transformation of infrastructure on par with electrification, industrialization, and telecommunications. If his prediction materializes, the data center boom won’t be just a technological trend but the beginning of a new industrial economy where the unit of production isn’t a car or a microchip, but intelligence deployed at scale.
Frequently Asked Questions (FAQ)
What is an “AI factory,” and how does it differ from a traditional data center?
An AI factory is designed to produce and operate large-scale AI models, with infrastructure optimized for training and inference (power, networking, cooling, automation), beyond the typical general-purpose data center.
Why does the expansion of AI depend so heavily on energy and the electrical grid?
Because intensive computing requires stable supply, capacity for electrical dispatch, cooling, and long-term planning. Without sufficient and affordable energy, the operating costs of AI skyrockets and growth is hindered.
What job profiles will grow with the massive deployment of data centers and AI factories?
Beyond software engineering, there will be increased demand for network technicians, data center operations staff, cooling and power specialists, cybersecurity experts, maintenance teams, civil works, and electrical installers.
What can Europe do to capitalize on the AI boom without losing industrial competitiveness?
Accelerate investments in energy and digital infrastructure, foster local ecosystems (chips, photonics, networks, cloud), and connect AI development with its historical strengths: manufacturing, automation, and robotics.