AI infrastructure is rapidly evolving across multiple fronts, as shown our top stories for today, which underscore both the supply constraints and massive investment commitments happening in the space. Just how tight is the GPU market? So tight that Nvidia is leasing back 18,000 of its own GPU servers from Lambda for a cool $1.5 billion. At the same time, Meta announced plans to spend at least $600 billion on U.S. data centers and related infrastructure by 2028, a level of financial commitment that signals just how central AI infrastructure has become to the strategies of the largest technology players (and adding even more to the demand side of the equation). Additionally, we get some interesting tidbits from Nokia that reflect its market strategy: It struck a new agreement with Supermicro to deliver AI-ready, automated networking solutions, and is seeing the first fruit from a channel partnership with Kyndryl to expand its reach in AI networking. Together, these moves illustrate the escalating race to secure the hardware, capital and networking foundations of the AI era. Let’s take a closer look.
Kelly Hill
Executive Editor
RCRTech
AI Infrastructure Top 3
Twin Nokia AIDC deals: Nokia has new deals with Supermicro and Kyndryl to deliver AI-ready, automated data center networking – at a global channel level, and a local client level (via a global sales setup).
Nvidia rents back its GPUs: Nvidia is leasing 18,000 of its own GPU servers back from Lambda in a $1.5 billion deal, underscoring chip shortages while boosting the cloud provider’s IPO prospects.
Meta’s spending plans: Meta plans to spend at least $600 billion on U.S. data centers and infrastructure by 2028, marking one of the largest AI-driven capital commitments in the technology sector.
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
AI Today: What You Need to Know
Deloitte builds AI infra expertise: Deloitte has launched a global AI Infrastructure Center of Excellence to help organizations design and operate AI-ready data centers, with a focus on power, cooling, cybersecurity, and scaling AI.
Cable cuts: Microsoft’s Azure cloud services faced disruptions from undersea cable cuts in the Red Sea, causing delays across regions. Traffic was rerouted, but reports suggest wider impacts in the UAE and Asia.
Electric job demand: One electrician in Northern Virginia estimates that 45-70% of the work on a data center data involves electrical systems. AI-driven growth is driving a surge in demand for electricians.
Taking stock of nuclear: Stocks in nuclear companies Vistra and Constellation Energy have surged more than 30% this year, with data centers fueling a “nuclear renaissance” that is attracting Big Tech partnerships.
Extracurricular activities: Pennsylvania lawmakers are scrutinizing data centers’ huge demands for electricity, water, and gas, as companies like Essential Utilities invest directly in hyperscale projects outside of jurisdiction of the state PUC.
New subsea cables: A €40 million, EU-backed project will build 550 kilometers of subsea cables linking Sweden, Estonia, and Finland by 2027, to boost digital capacity, resilience, and secure cross-border connectivity.
Scaling 5G and DCs: Telecom Argentina is scaling its 5G network toward 750 sites while expanding data centers for AI workloads, with growth in B2B services and private networks fueling the operator’s rising investment plans.
Optical AI chip: Engineers at the University of Florida created an optical chip using light for AI “convolutions” of image and pattern recognition. Their work improved efficiency to as much as 100 times that of current chips.
Memory matters: China’s AI chip ambitions face a critical bottleneck: High-bandwidth memory shortages, limiting output despite processor capacity. There are stockpiles, but China’s HBM production likely won’t scale competitively before 2026.
The Bible, by AI: Asking AI to create videos based on the book of Revelations, it turns out, makes for some pretty wild content. The tech is driving debates about shallow content versus fostering engagement.
