- Rakuten and Intel are repositioning vRAN as AI-native infrastructure, embedding automation directly into network operations
- The virtualisation driven by AI raises questions on operation complexity and overall cost structures
What happened: From virtual to AI-native
Rakuten and Intel have announced a renewed push towards what they describe as an “AI-first” virtualised radio access network (vRAN), according to a report by Telecoms.com. The initiative builds on Rakuten’s long-running effort to virtualise mobile networks and Intel’s role as a supplier of x86 processors optimised for telecom workloads.
Rakuten, best known outside Japan for e-commerce but also the operator behind a fully virtualised mobile network, is positioning AI as the next layer in network software. Intel, meanwhile, is pitching its latest CPUs and accelerators as capable of running RAN functions alongside AI workloads on common infrastructure.
According to the companies, AI could be embedded across RAN operations, from traffic optimisation to energy management, reducing manual intervention and improving utilisation. The framing matters: virtualisation alone has not always delivered the cost savings operators expected, largely because software complexity replaced hardware simplicity.
The announcement lands as operators face rising compute demands driven by video, cloud gaming and AI-heavy applications. It also coincides with heightened scrutiny of how large platforms such as Meta and YouTube rely on telecom networks to distribute increasingly compute-intensive services, pushing costs back onto carriers.
Also Read: Samsung completes first commercial vRAN call, pushing toward AI-native, 6G-ready networks
Also Read: Samsung and Orange France blaze trail with 5G vRAN
Why it’s important
Virtualisation is entering a more political and economic phase. Courts and regulators are examining the balance of value between content platforms and network operators, while telcos are looking for structural ways to reset their cost base. AI-native vRAN is being presented as one such lever.
If AI can genuinely automate network planning and real-time optimisation, operators could reduce operating expenditure and delay capital upgrades. From a financial perspective, that could improve margins in a sector where returns have lagged traffic growth for years. The risk, however, is that AI simply adds another layer of dependency on large silicon and software vendors.
Rakuten’s experience gives the partnership credibility, but it also highlights the challenge: virtual networks are harder to run, not easier. Whether AI simplifies that complexity or merely masks it will determine if “AI-first” vRAN becomes a turning point or another incremental upgrade.
