AI is making telecom more expensive, warn Ericsson and Nokia

Ericsson and Nokia have issued warnings regarding the rising costs of telecommunications infrastructure driven by the global surge in artificial intelligence development. The two major European vendors indicated that the massive demand for high-performance computing components has created a competitive supply environment. This trend is impacting the availability and pricing of essential semiconductors used in mobile network equipment. As chip manufacturers prioritise the more lucrative AI data centre market, traditional telecommunications hardware providers face increased costs to secure necessary inventory.

The shift in the global semiconductor landscape has forced vendors to re-evaluate their supply chain strategies. Ericsson noted that while the industry has moved away from the extreme shortages seen during the pandemic, the specific appetite for AI-capable chips is putting fresh pressure on production lines. Nokia shared similar sentiments, highlighting that the convergence of networking and compute power requires more sophisticated silicon. These advanced components are currently in high demand from tech giants building out large-scale language models and generative AI platforms.

Financial analysts suggest that these inflationary pressures on hardware components may eventually be passed down to mobile network operators. While vendors typically absorb some fluctuations in production costs through long-term contracts, the prolonged nature of the AI boom makes price increases for base stations and core network equipment more likely. This comes at a time when many telecommunications providers are already cautious about their capital expenditure following significant investments in 5G spectrum and initial deployments.

Furthermore, the integration of AI features directly into the Radio Access Network is increasing the complexity of hardware requirements. Modern equipment now requires higher processing capabilities to handle real-time traffic optimisations and energy-saving algorithms powered by machine learning. This architectural shift necessitates more expensive chips per unit, further elevating the baseline cost of network densification and modernisation projects. Vendors are now tasked with balancing the delivery of these advanced features against the rising cost of the underlying silicon.

Industry observers expect the situation to persist as long as the demand for AI training and inference hardware remains at its current peak. The supply of specialised processors and high-bandwidth memory continues to be constrained, leading to longer lead times and higher premiums for high-end components. Both Ericsson and Nokia have indicated they are working closely with their silicon partners to mitigate these risks, though the broader market forces remain outside their direct control.

The long-term impact on the speed of 5G-Advanced and 6G development remains a primary concern for the sector. If the price of hardware continues to climb, some operators may choose to delay their next phase of network upgrades or scale back the scope of their site installations. This could lead to a fragmented rollout of next-generation features depending on the financial resilience of specific regional markets. The industry will likely monitor chip pricing closely over the coming quarters to determine if current inflationary trends represent a permanent shift in the cost of connectivity.