Mobile World Congress (MWC) is massive. 100K people ebbed in and out each day. This was my third time at MWC in Barcelona, and it never fails to amaze me – from the breadth of technologies to the massive displays to the quirky gadgets that one might find at the show.

Aside from it being visually over-stimulating with large, brightly lit booths, it was a story of progress, maturity and opportunity for Ampere. Ampere is now in its fourth year at MWC, and each year, we continue to see more and more interest in our products and partner solutions. This year was special. Ampere and our partners announced production ready solutions for the RAN, especially Open RAN, and have made significant progress toward carrier ready products going to trial this year.

AI is Changing the Compute Landscape for Telcos: Again

Unsurprisingly, AI was top of mind in every booth and conversation this year. Efficiency has always been a key priority for telecom providers when it comes to their compute infrastructure. Coming on the heels of a 5G infrastructure upgrade still being digested, telcos must now face an even bigger challenge with the onset of AI. Managing large-scale networks, often in remote, edge-based environments, has long required processors that draw minimal energy. With AI, the imperative is even more important.

Since the advancement of vRAN and Open RAN, x86 solutions have dominated network builds, providing a balance between performance and efficiency while promising a more open software-defined paradigm. Now, with the rapid integration of AI service into and onto telecom networks, new demands for heavier computing and dynamic service expansion are reshaping these requirements and underlining the need for Open RAN on power efficient platforms.

Telcos are now evaluating AI for a range of applications, from dynamic spectrum analytics, service reliability and load efficiency to new revenue models utilizing proximity to end users for AI services delivered over their networks. AI processing requirements are forcing telcos to upgrade their existing infrastructure to provide more compute power. At the same time, the energy footprint of this infrastructure remains a primary concern for carriers, making these upgrades within realistic power and cost constraints challenging indeed. As telcos look to deploy more AI-enabled services, they need to increase compute performance without exceeding existing power and cooling capacity.

Ampere’s Advantage: More AI Compute, Smaller Energy Footprint

Ampere processors have already demonstrated a strong advantage in energy efficiency for traditional RAN workloads, and this is what many of our partners exhibited at the show. But can this advantage extend to the AI era in telecom? The answer is: yes. Ampere CPUs were designed from the ground up for efficiency and high core density. Our technology combined with software that optimizes computational computing on our platform allows Ampere processors to excel at AI inference tasks from traditional machine learning (ML) to the next generative AI advancements. We now see more and more telcos looking to capitalize on this to address AI at the edge. They are seeing that our platforms give them more headroom to run AI services in addition to traditional RAN workloads while keeping energy usage under control.

Real-World Adoption and Ecosystem Support

Ampere benefits from the vision of our partners that have helped bring our solutions to this market. Through our conversations and presentations at MWC, it became increasingly clear that the industry wants horizontal, open solutions – especially in the era of AI.

To enable this, we announced ecosystem collaborations with Supermicro, SynaXG, Fujitsu, OREX SAI, SUSE and Canonical to bring support for Ampere products across all aspects of the Open RAN solutions stack. In addition, we were joined by visionary customers such as Rakuten and Parallel Wireless, and hosted by the GSMA to discuss the alternatives telcos have when making decisions to deploy open, power efficient infrastructure in RAN. Because of this strong support, we are securing new support from telcos worldwide to help them solve their challenges around compute demand and energy capacity.

A New Era of AI Compute for Telcos

With so much going on at MWC, it can be hard to come out with a key takeaway. This year, however, my takeaway was abundantly clear: telcos are facing a new era driven by AI while still coping with the demands of traditional compute infrastructure. These companies can no longer rely on legacy x86 architectures or power intensive GPUs to meet the growing demands of AI on their networks. They need processors that maximize AI Compute within existing power constraints. Ampere and our partners are delivering exactly that – offering scalable, power-efficient solutions that enable telcos to unlock the full potential of AI without breaking their energy or capital budgets.