BIMB Securities believes the artificial intelligence (AI) infrastructure boom is far from over, as hyperscalers continue to expand capacity while the pursuit of more capable AI models fuels ongoing upgrades across servers, networking equipment and data centers.

The research house said in a note on last Friday that while brief market pauses and sentiment-driven dips are to be expected, the primary indicators of a structural downturn such as budget cuts, overcapacity in data centers, or falling semiconductor demand have not materialized yet.

“Consequently, the upward trajectory for AI-driven semiconductors remains solid,” it said.

It also said healthy momentum across the supply chain is further confirmed by rising AI accelerator shipments and Nvidia’s ongoing data center revenue growth, proving that the infrastructure buildout remains in a strong growth phase.

BIMB also highlighted that the AI infrastructure buildout looks less like a temporary boom-and-bust cycle and more like a permanent, multi-decade structural shift in the global economy.

Cited PwC, it said global data center capital expenditure could expand from $0.8 trillion in 2026 to $31.6 trillion by 2050 under its base case scenario, with potential upside towards $50 trillion should AI adoption accelerate further.

“Unlike previous technology investment cycles, demand stems from the growing need for servers, GPUs and networking equipment, which typically need replacement every four to six years,

“Indeed, closer to home, we are already witnessing this trend,” said the research house.

It is noted that several players are moving from 800G to 1.6T optical transceivers, reflecting the industry’s continuous push for higher bandwidth and faster data transmission to support increasingly intensive AI workloads.

If AI spending wanes, only some Asia-Pacific tech firms would be resilient – S&P