A reminder, if needed, that the AI boom is no longer just about chips and models, but about the network plumbing that lets it all work: Nvidia is to invest $500 million, rising potentially to $3.2 billion, to take an expanding stake and a priority supply from US photonics manufacturer Corning. AI training workloads have been forced out of servers, racks, and clusters as they multiply; they need to be connected as if they are still in the same geography, facility, cabinet – which means light-speed fibre optics and ultra-flex programmability. Which is what we have been talking about for most of the week – with Lumen yesterday buying up the control plane, and Zayo buying up the route miles.
But this is different – it is more about the ‘scale-up/out’ in racks and factories, than the ‘scale-across’ (DCI) between clusters, where Lumen is putting special focus, or the broader long-haul and metro-access, which Zayo has just extended. At the same time, Nvidia has the whole picture covered, of course. It has invested $2 billion in Marvell to integrate Marvell’s custom silicon and optical components. It has just pumped $2 billion each into Lumentum and Coherent to target the lasers, optical engines, and coherent tech that power both scale-out intra-factory links and scale-across inter-factory (DCI) networks – deals also structured as hybrid supply/equity contracts.
For GPU networking, it has acquired and developed (or just developed) its own proprietary NVLink and NVSwitch for ‘on-node’ scale-up GPU interconnects, the InfiniBand fabric (from Mellanox, part of a $7 billion deal) for scale-out factory clusters, the Spectrum-X Ethernet platform (based on Mellanox IP) for multi-tenant cloud connectivity, and the Kubernetes-based Run:ai Orchestration platform ($700 million in 2024) for software-defined workload orchestration. It has an eye on the networking conundrum at the metro and enterprise edge, too – where its AI-RAN work with Nokia goes, and its collaborations with AT&T, Softbank, Cisco, T-Mobile variously on AI-RAN or AI-grid structures.
It has tie-ups with Siemens and others inside factories, as well – in case we forget where lots of these systems and workloads are headed. So Nvidia is everywhere. Well, of course it is. It is securing the essential digital pipework in AI data centers – by buying up the supply ahead of time, and ensuring demand stays high. The deal with Corning is symbolic, as well – just because Corning, 175 years old, practically “invented” (first commercialized) low-loss optical fiber, back in the 1970s – which is the breakthrough the entire fiber telecom industry has run with. The same materials science informs the most demanding layers of AI infrastructure today – which is what Nvidia is buying.
There’s also the whole sovereign made-in-America manufacturing angle: three new plants in Texas, 3,000 “high-paying” jobs. Like we said in the piece: it might just be that, when the dust settles, the year-end reviews about the plotting and digging in AI infrastructure conclude that this is the most important deal: about good old networks, and breaking through the glass ceiling.
James Blackman
Executive Editor
RCR Wireless News
RCR Top Stories
US glass making: Nvidia has agreed a multi-year supply deal with Corning that could grow into a $3.2bn equity position, alongside a major expansion of US optical fibre manufacturing capacity aimed at meeting surging AI infrastructure demand.
Q1 SKT scores: SK Telecom posted stronger Q1 results as 5G subscriber growth recovered and AI data center revenue surged, reinforcing the operator’s strategy to expand deeper into AI infrastructure and enterprise services.
Edge AI systems: Blaize and Winmate have a new deal to develop edge AI systems. The partnership targets defense, maritime, and other industries that require secure decision-making independent of cloud-based data links.
RCR AI roundtables: Smarter minds, smaller rooms, bigger ideas – RCR is hosting a series of invite-only roundtables about AI infrastructure on October 21 in Dallas. Attendees will leave with fresh perspectives and actionable ideas.
WEBINAR ALERT: Join RCR and VIAVI on May 21 to discuss how quantum-era technological progress and geopolitical dynamics are elevating telecom security from an operational concern to a strategic priority.
Powering Sovereign AI at Scale
AI is central to telecom and national infrastructure, but must be secure, compliant and sovereign. Read this white paper to see how AI Factories enable trusted scale and new revenue.
Beyond the Headlines
Lumen buy-in: Lumen is to acquire cloud platform Alkira for $475m to strengthen its ‘east-west’ DCI strategy, and bolster its programmable NaaS roadmap to cover all the compass points – from enterprise to cloud, US to everywhere.
Vodafone buy-out: Vodafone will buy CK Hutchison’s stake in VodafoneThree for £4.3bn, taking full ownership of the UK’s largest operator – in a move to accelerate integration, sharpen decisions, and drive consolidation.
Nokia sell-off: Cut-price FWA sale shows Nokia’s shift to AI infra, sets the tone for further disposals, and gives Inseego an instant double-your-business boost – which poses questions for the San Diego firm, as well.
Battery storage: When goals move beyond simple standby backup and require an active, value-generating asset, Battery Energy Storage Systems (BESS) may be the way to improve day-to-day data center economics.
iPhone moment? Most factory-floor ‘physical AI’ runs on lightweight edge compute, says NTT Data – but 5G sensing, via SRS, could unlock a new wave of demand for private 5G, tying edge IoT and AI into a single enterprise stack.
What We’re Reading
Sovereign sharing: Telefónica is launching a multi-sector data platform and demo center for organizations to securely share and analyse data across industries. The system emphasises data sovereignty, interoperability, and governance.
Auto-networks: HPE is launching new autonomous networking capabilities that use agentic AI to turn enterprise networks into self-driving systems. The tech can detect, diagnose, and fix issues in real time, reducing manual intervention.
FWA uplink tests: Samsung and Qualcomm have achieved an industry-first validation of 5G Power Class 1 performance on virtualized RAN, boosting FWA uplink speeds and extending coverage through higher transmission power.
Agentic oversight: ServiceNow is expanding its AI Control Tower with Nvidia to extend oversight of agentic AI from employee devices to large-scale data center workloads. They aim to unify security, monitoring, and compliance.
HaLow house: Morse Micro has named Gateworks as the first global design partner to speed development of Wi-Fi HaLow solutions with faster production-ready design and deployment of Wi-Fi based industrial IoT systems.
Events
Virtual Program
Test & Measurement Forum, May 19
Join the annual Test & Measurement Forum where industry leaders, innovators, and engineers come together to tackle the most pressing test and measurement challenges in telco today. Register now
Telco AI Forum, June 16th
Telco AI Forum brings together operators, vendors, hyperscalers, and academia to explore how the evolution of the industry and partnership ecosystems is laying the foundations for AI-native 6G networks and unlocking ROI. Register now
Quantum Safe Networks Forum, July 14th
Quantum Safe Networks Forum brings together telecom operators, cybersecurity experts, and industry analysts to explore how to build resilient, future-ready infrastructure in the face of quantum disruption. Register now
CCA’s 2026 Mobility & Connectivity Show
Join industry stakeholders and innovative leaders in the communications service provider community this spring at CCA’s 2026 Mobility & Connectivity Show, April 14-16 in Louisville, KY. Register now
RCR Roundtables AI Infrastructure, October 21st, Dallas, Texas
Join 50 senior data center, energy and AI leaders at the Ritz-Carlton Dallas on October 21 for invitation-only roundtables on powering and scaling AI. Request your invitation
Industry Resources
Webinar, May 21st
Securing telecom infrastructure for the quantum era
Webinar, June 2nd
Scaling optical networks for the AI and hyperscale era
Report
AI in testing – Developing trust, delivering results
Whitepaper
Powering sovereign AI at scale
Whitepaper
Scalable database design for 5G and beyond
Report
Test, measurement and service assurance in the AI era
Summit Access
GSMA Device Enablement Summit: How operators can fix device-network fragmentation
