As artificial intelligence (AI) shifts from answering questions to making decisions, pressure on digital infrastructure is rising quickly, according to Alexey Navolokin, general manager for APAC at AMD. 

He said the next wave of development, known as “agentic AI,” will demand far more than today’s systems were designed to handle.

Navolokin described agentic AI as systems that can reason, plan and act across multiple platforms. Instead of merely suggesting travel options, for example, such systems could book flights, update calendars, send reminders and adjust schedules in real time.

This shift from reactive tools to proactive digital agents means AI workloads will run longer, involve more steps and require constant coordination, he explained.

Navolokin said this evolution is driving a surge in computing demand, likening it to adding billions of virtual users to global networks. 

For countries such as the Philippines, which face demographic pressures including an aging population and a tightening labor force, he noted that AI-driven productivity gains are increasingly vital, but only if infrastructure can keep pace.

While graphics processing units (GPUs) often dominate AI discussions, Navolokin stressed that central processing units (CPUs) remain essential. 

CPUs manage data movement, memory and coordination of tasks across systems. He added that many AI applications, including language models, fraud detection and recommendation engines, can run efficiently on CPU-based servers when supported by high-performance processors.

Connectivity is another critical layer, he said. Advanced networking components and high-speed interconnects help move large volumes of data between processors, reduce delays and keep distributed systems functioning as one. 

As AI systems become more modular, combining different processors and memory types, Navolokin said heterogeneous system design is becoming standard.

He also underscored the importance of open systems. Open software platforms and hardware standards, he said, allow developers and organizations to build and scale AI without being tied to a single vendor, lowering costs and supporting innovation.

With AI adoption accelerating, Navolokin said investments in scalable, open and well-integrated infrastructure will determine how effectively nations harness the next generation of intelligent systems.