.jpg)
There is a silent misconception holding back Portugal's ambition: that of thinking of "reindustrialization" as an effort to recover what we already know. As if it were enough to reinforce what already exists, the current chains, the current competitive advantages; but recent history has shown that the future does not reward repetition, but rather preparation.
For better context, let's focus on a simple example: systems such as NVIDIA DGX H100/H200 weigh around 130 kg per server. But when we scale up to entire AI racks, such as the new systems with dozens of interconnected GPUs and liquid cooling, we are talking about equipment with industrial weight and requirements; there are references that a GB200 NVL72 rack weighs around 1.36 tons, with more than three-quarters of the weight being heat dissipation systems.
Artificial intelligence and, in general, high-performance computing are exposing a point that many have underestimated: the limit is not just the chip. The limit is heat, energy, water, cooling, the network, physical space, and resilience.
The signal is clear: the competitive frontier is shifting toward those who master the triad of computing + energy + cooling. And this is not just a "technology" issue. It is an industrial policy issue.
We are living in an era where what seemed like absolute truth last year is already outdated this year. And it's not just about technological trends. It's about structural changes: energy, computing, materials, strategic autonomy, defense, communications, logistics, mobility, health. Everything is being reconfigured at the same time and interconnections are growing.
If the future demands abundant, predictable, and clean electricity; it demands data centers and critical infrastructure; it demands thermal engineering and robust electrical networks; it demands talent in innovation, process management, electronics, systems, materials, and logistics. So, any country that fails to build these foundations quickly will be condemned to buying capacity from third parties, depending on third parties, and always competing "on the sidelines."
What if the next leap happens... outside Earth?
Now here's a provocative thought: if a significant part of the cooling effort exists to dissipate heat in an atmospheric environment and in a terrestrial building, what happens when we imagine computing in environments where thermal architecture can be rethought?
The Space sector is no longer just scientific exploration, it is now economic infrastructure. From georeferencing and synchronization for logistics and mobility; Earth observation for agricultural activity, insurance, maritime economy, civil protection; resilient communications and technological security and autonomy in a geopolitically unstable world.
It is no coincidence that the Portugal Space 2030 national strategy explicitly states that multiple sectors benefit from space solutions and frames the effort as an industrialization and innovation agenda.
The country needs policies that bridge the gap between R&D and manufacturing.
Portugal has universities and R&D centers with real capacity, but the critical point is transforming scientific excellence into a scalable industrial advantage. This bridge is the difference between a country that "does research" and a country that creates the industry of the future. And here we must be frank: we often talk about innovation as if it were just a "department," but in the new economy, innovation is a value chain. It is the ability to design, manufacture, operate, maintain, certify, and export. It is economic sovereignty.
Tomorrow's competitive advantages are not inherited; they are built.
Article originally published on February 12, 2026, in Jornal de Negócios, available here.
