MATRIX GO in the MEDIA
Original Article at: InforChannel: https://inforchannel.com.br/2025/09/20/infraestrutura-para-ia-e-preciso-ter-uma-base-solida-para-ia-do-amanha/
Artificial Intelligence is spreading across all sectors of the economy and everyday life. Soon, every website, app, or business will have its own AI assistant, operating in real time, personalizing experiences and optimizing decisions. But for this advancement to reach scale, a technical challenge remains: infrastructure.
To meet the growing demand for processing, storage and connectivity, a leap in quality will be necessary in the systems that support this new digital era.
Telecommunications companies, in particular, will play a decisive role. They need to invest now to ensure that their networks and data centers are prepared to support the AI services that will emerge in the coming years.
To delve deeper into this topic, we listened Nicola Sanchez, CEO of Matrix Go, a company specialized in Artificial Intelligence systems.
He shared his perspective on the main obstacles, ongoing technological transformations, and the geopolitical landscape already emerging around AI infrastructure.
Lack of chips, energy and talent: the bottlenecks of AI
"One of the main bottlenecks is the shortage of hardware—such as GPUs and chips—for scaling AI," says Sanchez. According to the expert, "not only is this infrastructure expensive, but it's still quite limited at the moment."
According to him, companies like Nvidia, Google, and Intel have been expanding their factories to meet the explosion in demand, but the pace of production still doesn't keep pace with innovation. At the same time, concerns about energy consumption are growing, becoming a critical factor for the sustainable expansion of AI models.
"This is a point that makes evolution expensive and, in some locations, even unfeasible. Energy is a real physical limit," Sanchez emphasizes.
Human capital is also an obstacle to AI development. Sanchez illustrates this with an example from his own organization: "At Matrix Go, we train our entire team—without exception—to understand how AI engines work and operate. But I know this isn't the reality for many businesses. If there's already a lack of basic knowledge, imagine the technical and operational knowledge."
It is understood that companies and startups have been reacting to this scenario with accelerated training programs, in an attempt to mitigate the deficit of professionals with specific skills in Artificial Intelligence.
However, we need to go further. In Sanchez's view, it's essential to encourage educational programs to ensure we have an increasing workforce specialized in AI.
Futurecom, a leading connectivity and innovation event for Latin America, is preparing to celebrate its 30th edition from September 30 to October 2, 2025, at São Paulo Expo. There will be over 250 exhibiting brands, two content arenas open to visitors, and five conference auditoriums for the Future Congress, Future Cyber, and Future GOV tracks.
The future is not with the same old chips
It's worth noting that AI infrastructure requires more than robust networks: it requires chips that can handle the complexity of models and the speed of inference. According to Sanchez, this won't be possible with traditional components.
“It is natural that conventional hardware has limitations in the face of the global scalability of AI and its real-time applications,” he observes.
In this context, three emerging technologies stand out:
Application-Specific Integrated Circuits (ASICs): customized chips for specific AI tasks, with extremely high energy efficiency;
Field-Programmable Gate Arrays (FPGAs)): reprogrammable, offer flexibility and high performance in targeted applications;
Neural Processing Units (NPUs): processors specialized in neural networks, increasingly present in Data Centers and Edge devices.
These innovations are reshaping data center design and directly impacting the decisions of operators and technology providers.
Infrastructure as a strategic power
More than just a technological asset, AI infrastructure has become a factor of geopolitical power. Sanchez warns that countries like China, the US, and India are in a veritable race to dominate this territory.
"I see that infrastructure already represents a new strategic power. The battle has already begun, silent but intense," says the executive.
This movement is driving the development of leaner, more efficient AI models, with the potential to further accelerate the adoption of the technology on a global scale.
"Every website, app, and company will have its own AI assistant. The infrastructure needs to keep up with this scenario," predicts Sanchez.
However, the energy issue remains a barrier. According to our interviewee: "OpenAI itself has already warned that the limits of AI lie in availability and energy efficiency. This will remain a key issue in the coming years."
The mission of telecoms: building the digital soil
To support the explosion of AI-based services and applications, telecom operators will need to rethink their role and invest in areas such as:
Expansion and modernization of networks, especially with the arrival of 5G and 6G;
Edge computing integration, bringing processing closer to the end user;
Partnerships with data centers and clouds specialized in AI;
Adoption of cleaner and more efficient energy sources; and
Training technical talent to develop and operate these solutions.
"Infrastructure is no longer a technical detail; it has become the backbone of the digital future. Whoever masters this foundation will dominate the services and experiences that will shape the world in the coming years," concludes Sanchez.
