It’s all gone hyper, of course. Nvidia’s $2 billion arrangement with Nebius to deliver a full-stack AI cloud is not just about flashy GPUs, but about owning the infrastructure that will power AI infrastructure for – for what: a decade; a couple of years? – a while, anyway. At the same time, the GFiber-Astound merger says the fiber market has the same idea: scale, reach, speed. Bigger networks mean lower costs, faster rollouts, and a better grip on the spiralling (AI) demand for bandwidth.
But compute and connectivity are only half the story; there are other subplots, sometimes forgotten. Equinix’s interconnection hub – discussed here yesterday, discussed at greater length here today – reveals a third force in this weird AI future: neutral platforms to orchestrate fragmented infrastructure to manage a multiplying ecosystem. Seems like good logic and a smart play from Equinix, shoring up its own position in this hyperscale parade.
Even if the AI market starts to sound like the IoT market. But hyperscale theory comes unstuck in silos, stacked across the edge-cloud continuum in different networks and data centers. Interesting, also, to see Equinix is looking to tackle familiar last-mile service bottlenecks with telcos at the same time as it is solving complex data center interoperability in the cloud. Maybe, winning at AI will be about this orchestration layer?
Clearly, there are architectural issues everywhere – which should be addressed if cloud-style elasticity is to triumph in rangier AI infrastructure for agents and inference workloads.
James Blackman
Executive Editor
RCR Wireless News
RCR Top Stories
Agents of chaos: Enterprises are scrambling as AI workloads splinter across the cloud-edge landspace. Equinix reckons the answer is in neutral interconnection hubs to orchestrate distributed infrastructure, and bring inference closer.
Trump deadline: The March 11 deadline passed for US agencies to act on Trump’s AI order, requiring the Commerce Department to evaluate state AI laws deemed “onerous” – raising uncertainty and potential federal‑state clashes over AI policy.
AI power crunch: At Metro Connect USA, Marc Ganzi, chief executive at DigitalBridge discusses the power crunch threatening AI infrastructure expansion plans, and shared strategies to get around the crisis.
Telco AI models: General AI models fail to “speak telco,” leaving operators in the lurch. A new initiative led by AT&T, AMD, and the GSMA is betting on open-source collaboration and unique datasets to build the precision tools the industry needs.
Testing HCF: Olivier Côté, product line manager at EXFO, digs into the testing challenges with hollow core fiber (HCF), and outlines how methods are evolving to accommodate its physical peculiarities.
AI-Powered Telecom Infrastructure
Supermicro, in collaboration with NVIDIA, delivers AI-powered infrastructure tailored for telcos, enhancing operational efficiency, network management, and customer experiences. Explore now
Beyond the Headlines
GenAI cloud boom: Cloud infrastructure spending reached $119 billion in the last quarter of 2025, driven largely by generative AI workloads, while new AI-focused “neocloud” providers emerge as growing competitors to hyperscale platforms.
Cyient at MWC: Cyient used MWC to promote a “human + AI” approach to autonomous networks, arguing that combining AI with human expertise will help telcos advance toward Level-4 autonomy and new revenue opportunities.
AI is eating the DC… and networks are on the hook. Data centers are teetering under their workloads, and being distributed far and wide in search of power, and obeyance of compliance and regulation. The challenge is to connect them.
P5G for physical AI: Ericsson has tie-ups with Future Tech and NTT Data to combine private 5G and physical AI in Industry 4.0; its appears to be picking up where Nokia has dropped off. Future Technologies is growing 35% per year.
India sets AI vision: India’s comms minister outlined the country’s telco AI vision at MWC, highlighting rapid 5G expansion, falling mobile data prices, and major national programs to extend connectivity and develop 6G.
What We're Reading
Nvidia and Nebius: Nvidia is investing $2 billion in a deal with Nebius to co‑develop a hyperscale, full‑stack AI cloud, integrating next‑gen infrastructure and software to deploy over 5 GW of compute and accelerate global AI workload growth.
Google cuts GFiber stake: GFiber (Google Fiber) will combine with Astound Broadband to form a major independent US fiber provider, with Stonepeak as majority owner and Alphabet as a minority stakeholder.
Bell pushes Canada AI: Bell and Coveo have a sovereign AI deal to modernize digital services for Canadian governments and regulated industries, combining Coveo’s AI platform with Bell’s AI ‘fabric’ while keeping data under Canadian law.
Deoleo taps Telefónica: Telefónica Tech is partnering with olive oil maker Deoleo to drive digital transformation with blockchain for traceability, AI for quality and operations, and digitalised plant maintenance to boost efficiency and transparency.
1NCE integrates LoRa: 1NCE and Netmore will integrate Netmore’s LoRaWAN connectivity into the 1NCE OS platform, giving customers unified access to both cellular and LoRaWAN IoT coverage through one system.
Events
Join this RCR Wireless News‘ event to understand the current state of the Wi-Fi as we examine a myriad of evolving use cases and monetization strategies being deployed by industry. Register now
Industry Resources
Webinar, September 18th
The journey to a fully autonomous network – The evolution of network automation and how Amdocs is leading the way
