Some thoughts on today’s news: that telcos are doubling down to support AI-era workloads, while cloud providers are discovering just how expensive the infrastructure build-out can be. On one hand, AT&T has said (on its 150th birthday) that it will invest $250 billion over five years to expand its US fibre, mobile, and satellite networks. It already passes 32 million locations with fiber; it wants to reach 60 million by 2030. Fiber, by the way, is just more important than mobile right now – at least until the training slows, and the inference starts. Meanwhile, Verizon told RCR at MWC about how to mix global long-haul fiber and dense metro fiber, also with enterprise edge networks. It told the same story at PTC in January, but there is more in it this time – and it is worth the read. But the AI buildout is expensive.
Oracle is preparing thousands of job cuts, apparently, as it grapples with a cash crunch from massive spending on AI data centers; separately, it has denied there’s any jeopardy in its work with OpenAI to expand its site in Abilene, Texas. Networks and data centers are different things, which have mostly evolved separately. AI is tying them together. The Oracle example shows the economics are still unclear – and that the new era of agents and inference, for which metro infrastructure is being upgraded, can’t come soon enough. But for once, the network build-out looks like a surer gamble. This is not another MEC story for telcos; programmability across fiber and 5G will be required, and might be monetizable. No matter where the workloads land – which no one actually knows yet.
James Blackman
Executive Editor
RCR Wireless News
RCR Top Stories
Verizon’s fiber diet: Global, programmable, and dense – from cloud to edge: Verizon talked at MWC about how its sees the AI stack evolving for telcos, and why its investments in backbone fiber, metro access, and private networks will win.
China Telecom on AI: China Telecom outlined its AI transformation strategy at MWC last week, highlighting cloud infrastructure, computing power platforms, and large-scale 5G network sharing with China Unicom.
AI DCs go nuclear: Nuclear is the only carbon-free baseload power capable of meeting 99.999% reliability for modern data centers. BYOP and BTM bring operators flexibility and options for getting around power grid constraints.
MNOs for sovereignty: Turkcell told MWC last week that telecom networks are a foundation for data sovereignty, as AI infrastructure, data governance, and sovereign cloud partnerships reshape the global digital economy.
Keysight roundup: Keysight demoed three solutions at MWC with partners: an AI-RAN test solution with Ericsson, a prototype for AI-driven uplink performance with MediaTek, and a pre-6G interoperability validation solution, also with Ericsson.
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
Bridge for DC power: Bridge Data Centers is exploring nuclear for AI data centers through a research partnership with A*STAR’s IHPC and engineering firm HY, as operators seek reliable low-carbon power for growing compute demand.
Intel on Wi-Fi 7: Intel’s Wi‑Fi innovations at MWC focused on how next‑gen Wi‑Fi (Wi‑Fi 7 and beyond) will support richer connectivity and complement cellular 5G, especially for capacity‑heavy enterprise and AI‑driven applications.
Cyient on AI nets: Cyient outlines a three‑layer approach to AI and autonomous networks that embeds intelligence into network engineering, operations and business outcomes to modernise telco infrastructure and move beyond AI hype.
Mid-term 5G pay-offs: Ericsson offered a measured view of AI’s impact on telecoms at MWC: fiber may lead the infrastructure boom today, but AI will also shift through the mid-cycle 5G evolution in AI-driven RAN optimization.
HCF for AI DCs: Hollow‑core fiber could ease the AI data center land and power crunch by enabling low‑latency, high‑speed optical links that support distributed facilities, though manufacturing costs and ecosystem maturity are barriers,
What We're Reading
AT&T’s $250bn pledge: AT&T is to invest $250 billion over five years on US fiber, mobile, and satellite networks to boost coverage, capacity, and resilience while supporting rising demand from AI, cloud services, and connected devices.
Phantom AI funds: The UK’s multibillion-pound AI strategy relies partly on exaggerated or unverified private investments, with some projects involving reused data centres, uncertain funding, or undeveloped sites.
Oracle denies AI crash: Oracle has pushed back against reports that its Stargate AI project with OpenAI is faltering, saying the broader plan to build 4.5GW site remains on track despite canceling a planned expansion at its Abilene campus.
DE-CIX in Bulgaria: Internet exchange operator DE-CIX has strengthened its work with BIX.BG in Bulgaria to expand its presence in the Balkan market and integrate Sofia’s peering ecosystem more closely into its interconnection platform.
MariaDB buys GridGain: MariaDB is acquiring GridGain to support enterprise AI use cases, combining MariaDB’s platform with GridGain’s in‑memory computing for high‑speed transactional and analytical workloads.
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
