The shift toward “physical artificial intelligence” applications such as autonomous vehicles, robotics and industrial automation could reshape demand for telecom networks and support a stronger outlook for ASEAN and China operators, Maybank Investment Bank said Thursday.
The research house said in a note that the next phase of the AI cycle is moving beyond digital services into real-world systems, including digital twins, connected machines and industrial control networks, creating new requirements for low-latency, high-reliability connectivity.
Maybank said this evolution could bring communications networks back into focus after several AI cycles in which value largely accrued to semiconductor, data center and power infrastructure players.
Under its base case, Physical AI traffic could account for about 32 percent of mobile data traffic in developed markets and China by 2031, with workloads increasingly driven by uplink-heavy and service-level agreement (SLA)-sensitive applications rather than traditional consumer usage.
The bank said such demand is structurally different from best-effort consumer traffic, as it is tied to real-world outcomes and therefore requires stronger service assurance, edge computing and orchestration capabilities.
“Physical AI workloads require lower latency, higher reliability and closer integration with edge infrastructure, making them more premium in nature,” it noted.
It added that potential revenue uplift could be meaningful even if monetization remains conservative.
If priced at around 50 percent of consumer mobile tariffs, Physical AI could still contribute about 16 percent of total mobile revenues by 2031 and lift industry-wide mobile revenue growth into the high single-digit range, compared with current low-single-digit expectations.
Maybank estimated that Physical AI traffic could reach around 550 exabytes by 2031, equivalent to roughly one-third of Ericsson’s projected mobile traffic in developed markets and China.
The bank said such traffic could support a comparable contribution to mobile revenues over time because Physical AI workloads are likely to be more valuable than traditional consumer data usage.
Unlike consumer traffic, Physical AI applications are expected to be increasingly uplink-heavy, enterprise-focused and, in many cases, mission-critical.
As a result, operators may be be able to charge a premium for connectivity services bundled with deterministic latency, edge computing, orchestration capabilities and service-level guarantees.
Maybank also noted that telecom networks operate in a constrained environment where spectrum is limited, infrastructure is regulated and nationwide networks are difficult to replicate.
If Physical AI creates a new category of high-value and difficult-to-substitute demand, it could strengthen operators’ pricing power while improving the quality of telecom revenues.
It also said terrestrial telecom networks are better positioned than satellite systems to serve Physical AI demand, particularly in dense urban environments such as factories, logistics hubs, airports and road networks where fiber backhaul and network densification are critical.
Even under a fully deployed low-earth orbit satellite constellation scenario, effective capacity in populated areas would remain limited compared with expanding terrestrial infrastructure, it noted.
On investment cycles, Maybank said concerns over a value-destructive 6G upgrade cycle may be overstated, estimating that cumulative 6G capital expenditure could be lower than 5G, as much of the foundational infrastructure has already been built.
The research house said the clearest beneficiaries are developed market and Chinese telecom operators, including Singtel, China Mobile, China Telecom and China Unicom, which are expected to benefit from stronger enterprise adoption and early deployment of robotics and edge AI systems.
According to Maybank, they combine dense fiber and mobile infrastructure, high enterprise digitization, stronger policy support, and a greater likelihood of deploying autonomous vehicles, robotics, smart factories and edge AI at scale.
For emerging ASEAN markets, the outlook is more mixed but stable. Maybank said risks from satellite-based competition appear limited, while growth from 5G fixed wireless access (FWA) in markets such as Indonesia and the Philippines remains intact.
While satellite connectivity should remain relevant in remote and underserved areas, Maybank does not believe it materially alters the economics of dense urban and suburban broadband markets where terrestrial networks retain clear advantages in capacity, indoor coverage and cost efficiency.
In fact, it continues to see scope for 5G-led fixed wireless access growth in markets such as Indonesia and the Philippines, where wireline penetration remains relatively low and mobile operators are better positioned to address incremental household broadband demand.
Meanwhile, capital expenditure risks linked to 6G rollout are likely to be manageable for Maybank, supporting a broadly constructive regional telecom outlook as Physical AI adoption accelerates.
According to the research house, the risk of a value-destructive 6G capital expenditure (capex) cycle is lower than in prior generations.
Much of the heavy lifting has already been done through 4G and 5G, including site densification, fiber backhaul, and broader spectrum holdings.
“As a result, the 6G cycle is likely to be more evolutionary than revolutionary, with lower radio and passive infrastructure intensity partly offset by higher software, cybersecurity and
edge-related spend,” it noted.
On balance, this should still allow capex-to-sales ratios to continue trending down over time, supporting free cash flow and earnings across the sector, it added.