Visabeira Group Uses Accelerated AI to Transform Operations and Economies

Hewlett Packard Enterprise

A good definition of technology is “human dreams made true.” But delivering on those dreams requires a lot of effort—and a lot of computing power.


At Visabeira Group, we deliver on our mission of “Performing Dreams” every day. A Portuguese multinational, Visabeira operates in various sectors, from telecom and energy to construction and hospitality. The group has achieved impressive scale, reaching €2.3 billion in revenue, employing more than 16,000 people in 16 countries, and marketing products and services in 119 nations across five continents.


Under my leadership as head of innovation, Visabeira has focused heavily on integrating next-generation technologies. In addition to growing revenue, our approach involves rethinking business—particularly by harnessing the power of AI and high-performance computing (HPC). However, as Visabeira moved forward on this journey, traditional cloud services couldn’t keep up with our demands.

Scaling to support AI’s complexity

Visabeira used to rely on top-tier hyperscalers like Amazon Web Services and Google Cloud, using virtualized GPUs on regular hardware. These platforms were highly scalable, but Visabeira’s use cases, such as renderings for industrial digital twins and model training for agentic AI, were becoming increasingly complex. For example, we work with 72 billion parameters in our model training.


The added complexity made it impossible to run our use cases in hyperscaled environments, which lack the low latency and high computing power we needed. Extended reality (XR) renderings are effective only if they look realistic. Our renderings, by contrast, were pixelated, like something from old-school Mario or Minecraft.

“We knew Visabeira wasn’t the only one with this challenge. As AI solutions multiply, more companies will hit the ceiling of what hyperscalers can provide. But the alternatives are inaccessible to most SMBs.” —Paulo Soeiro Ferreira, Head of Engineering and Innovation at Visabeira Global


We knew Visabeira wasn’t the only one facing these challenges, especially in the Portuguese market. As AI solutions multiply, more companies will hit the ceiling of what hyperscalers can provide, but the alternatives are inaccessible to most SMBs. By bringing an accelerated parallel computing platform to the Portuguese market, Visabeira could open the playing field of AI to an increasing number of SMBs.

Shoulder-to-shoulder with a future-forward partner

Visabeira decided to invest in supercomputing platforms with HPE to meet these demands. Moving to accelerated parallel computing was a simple decision because it would allow us to perform millions of tasks simultaneously, resulting in vastly improved performance.


Our selection of HPE as a partner was also straightforward. Several factors pointed us towards HPE:

  • HPE’s established network of credible business partners and service providers assured us we were partnering with a vertically integrated organization capable of supporting us through any number of challenges
  • HPE has a significant Portuguese presence, including specific support for HPC
  • HPE came recommended by Visabeira’s academic partners, especially the University of Aveiro
“HPE’s established network of credible business partners and service providers assured us we were partnering with a vertically integrated organization capable of supporting us in the challenges to come.” —Paulo Soeiro Ferreira, Head of Engineering and Innovation at Visabeira Global


With all this in mind, Visabeira adopted the HPE ProLiant DL385 Gen11 and HPE’s latest-generation GPU-based HPC AI, including NVIDIA OVX and NVIDIA enterprise AI solutions. Combined with HPE Professional Services, these solutions deliver unprecedented computing power while significantly improving latency — both vital for Visabeira’s intensive AI applications.

From a user perspective, implementing a cutting-edge system can sometimes involve a massive learning curve that requires weeks of training. Implementing HPE AI solutions was the opposite: it was a simple process designed for engineers to pick up and use, allowing Visabeira to experience results that much faster.

Increased productivity, accelerated timelines, and 96% faster AI training

From accelerated parallel computing for rendering elaborate XR environments to model training for agentic AI, our results have been impressive. The move to HPE’s platform has drastically reduced the time required for training AI from eight hours for a single case to just 20 minutes—a decrease of 96%.


Why does that matter? Let’s examine what accelerated computing enables. The scale of the Visabeira Group makes understanding our internal operations one of our greatest challenges. Using agentic AI, Visabeira can now deploy digital team members to detect previously unnoticed organizational efficiencies and solve what seemed like impossible problems.


Agentic AI can also provide decision-support tools, like group-wide accounting metrics and insights. Despite Visabeira’s size, we could close our accounting every day. Daily updates on actual versus forecasted revenues will allow management to respond to events in the moment, instead of after the yearly report.


The HPC platform also enables spatial computing for industrial digital twins, in addition to high-definition renderings for digital twins and other immersive environments. Our pixelated Minecraft world has been replaced with an environment so realistic that it makes you want to reach out and touch it.


We have also seen a clear return on investment through increased productivity and accelerated project timelines. Instead of struggling to port data and applications into the platform, our data scientists can focus on making every AI project run as smoothly as possible.

Driving sustainable change

It’s important for us to stay in control of all of these opportunities, so we are currently developing an internal AI transformation office to organize and orchestrate the integration of AI capabilities across the group. With such an organizational catalyst, we can take full advantage of AI’s transformative potential.


HPE’s professional consultants have been very supportive throughout our journey. They are top-tier professionals who helped us adapt to the new platform, accelerating our AI adoption cycle in the process. This transformation will affect not only Visabeira’s internal operations but also have a ripple effect on AI adoption throughout the Portuguese market.

“Shifting from linear to circular economies is the only way to save the planet. But circular economies are complicated, and will require high-performance computing to manage.” —Paulo Soeiro Ferreira, Head of Engineering and Innovation at Visabeira Global


Zooming out even further, we can foresee supercomputing’s role in driving greater economic transformation. At some point, we will have to shift from linear to circular economies to increase resiliency and maximize resources. While we can’t ignore AI’s high energy needs, circular economies are complicated and will require HPC to manage them. We can’t ignore the need for more sustainable business models, and HPC will get us there faster.


Visabeira’s partnership with HPE puts us in a great position to succeed in a rapidly evolving future. With low latency, HPC services and an innovative vision, Visabeira Group is setting the stage to be a leader in the broader movement towards more AI-driven economies.