Mobile ecosystem is facing major memory problems

Earth, globe, computer chips, silicon, AI
A global memory shortage could raise both the cost of network upgrades and your next phone. (Art by Midjourney for Fierce Network)
  • AI’s memory hunger is reshaping chip supply — and telcos are caught in the crossfire
  • Core network upgrades face a new bottleneck: pricing
  • The memory crunch may soon show up in your next phone

Telcos may want to start stuffing their piggy banks because a global memory shortage is poised to make network upgrades more expensive than ever.

We’ve written before about how memory bandwidth could become a limiting factor for AI growth. But a memory shortage has also emerged as chipmakers rush to pump out enough GPUs to feed hyperscalers’ AI appetite. Scarcity, of course, comes with higher prices and long wait times. 

That’s bad news for telcos looking to upgrade the servers underpinning their core networks. 

“Telcos will need to upgrade their core networks — not just as a result of 6G, but also because new AI-based service offerings in the next 1-2 years will require much more processing power in the core networks that, in some cases, haven’t been fully upgraded yet to support such services,” J. Gold Associates founder Jack Gold told Fierce. 

Costs and competition

To upgrade their networks, telcos will need servers, which contain both processors and memory. Processors like CPUs have a small amount of memory built into them. 

Obviously, though, telcos aren’t the only ones trying to get their hands on servers, CPUs and memory. 

While cloud server CPU demand has dropped in recent years thanks to AI-related hype around GPUs, analysts are now forecasting a resurgence in focus on CPUs as replacement cycles come into play. 

New Street Research’s Pierre Ferragu and team predicted that cloud CPU server shipments will bottom out at around 5 million in 2026 before climbing steadily back up to an estimated 8.4 million by 2030.

As Gold noted, telcos and cloud companies tend to get their servers from different suppliers. But the chips inside those servers? There’s a bit more overlap there.

Intel, which supplies Xeon processors for Nokia’s edge appliance series and Ericsson’s Cloud RAN, also has big-name supply deals with AWS and Google Cloud. Similarly, AMD’s CPUs power Nokia’s Cloud Platform and some Ericsson kit, but the chipmaker has been on a roll lately of scoring major cloud deals with the likes of OpenAI, Meta, Oracle and more.

Mix all these ingredients together and you end up with cost increases as operators start to upgrade their networks for AI.

“With any shortage, whether in CPUs or memory, it affects the prices,” Gold said. “And during any upgrade cycle (like 5G to 6G), the capital expenditures are very high for the telcos, and in this case will be amplified by the shortages, which may ultimately increase costs to the subscribers.”

Futurum Group Research Director Brendan Burke agreed. "Telcos will increasingly find themselves squaring off with cloud titans for CPUs and they'll lose that fight on purchasing power alone unless they plan ahead," he told Fierce. "The memory shortage will create a pricing dilemma that operators cannot fully avoid. The uncomfortable truth is that telcos are price-takers in both compute and memory, while being price-setters to neither their infrastructure vendors nor their device OEM partners."

Analysts and Intel CEO Lip-Bu Tan have predicted the memory shortage will persist at least through 2028. Whether that timeline will hold remains to be seen.

So what's an operator to do? "The operators who recognize this structural disadvantage earliest and build supply chain resilience through diversified architectures and long-term agreements will be best positioned," Burke said. 

Gold added that there may be time for the markets to course correct before 6G deployments ramp. “It's not clear yet how all this plays out for 6G upgrades as they are still a few years away and a lot can change,” Gold said.

Mobile device dilemma

But chips aren’t the only place telcos could feel the squeeze from the memory shortage. This shortage is about to ripple across the mobile device ecosystem.

“With large [AI] models requiring so much high-bandwidth memory, the corporations that dominate in this space are focusing their manufacturing efforts on building more HBM. Not good news for the memory needed for computers or phones,” Mobile Ecosystem Forum CEO Dario Betti told Fierce.

He noted that while Apple and Samsung may have enough market power to get what they need, other brands may struggle. Consumers, he added, may also start to take notice as devices cost more and offer less in the way of performance advancements. Emerging markets and volume-driven brands will likely feel price increases the most. 

“It is not just device manufacturers who are being affected by the AI memory crunch. It is impacting the mobile ecosystem as a whole. We may see, for example, that smartphones will not get faster every year as consumers have [become] accustomed to. We may see storage shrinking, and memory being capped for some models,” he said.

Indeed, the impacts of the shortage are already being felt by smartphone vendors. Counterpoint Research data showed global smartphone shipments dropped 6% in Q1 2026, while IDC pegged the decline at 4.1% over the same period.

“The smartphone market has entered one of its most challenging periods, driven by acute memory supply constraints that are directly impacting both shipments and demand,” IDC Senior Research Director Nabila Popal stated. “Limited memory availability is forcing shipment reductions, while sharply higher memory prices are pushing up bill‑of‑materials cost and forcing price hikes by many top brands.”

What the lack of consumer appetite for pricey new phones might mean for operators remains to be seen. But the memory shortage is certainly something they’ll have to navigate. 

Update 4/16/26: This story has been updated to add comments from Futurum's Brendan Burke.