The dollars being spent on AI infrastructure — data centers, servers, fiber, and networking — are already staggering, yet the figures seem to get bigger by the day (looking at you, Stargate). Now Alphabet has bumped up its 2025 capex guidance by another $10 billion to $85 billion; most of it will go to servers. Meanwhile, Nokia prepares to reap the benefits of a “supercycle” of global AI infrastructure spend, with the NEM’s Q2 optical networks revenues already slipping past mobile networks — a sign of the times, for sure. Let’s get into the numbers.
Kelly Hill
Executive Editor
RCRTech
AI Infrastructure Top 3
Alphabet’s AI momentum: Alphabet reported Q2 results that saw Google Cloud revenues jump more than 30% YoY, and more billion-dollar deals signed in the first half of 2025 than in all of 2024. It also raised capex guidance.
Nokia preps for AI ‘supercycle’: Nokia’s network infra unit overtook mobile networks in Q2, reflecting a shift toward AI-driven IP and optical demand. Enterprise sales hit records; traditional telco sales declined. Restructuring looms.
TM builds AI backbone: TM Global is expanding AI-ready infrastructure across Southeast Asia, with flexible GPU services, scalable power, and secure data centers designed to meet rising enterprise and hyperscaler demands across the region.
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
AI fox in the henhouse?: UK ministers are under pressure to be more transparent about the public data that could be shared with U.S. tech giant OpenAI, following a $300B deal with the firm to put AI to work in the public sector.
Supermicro scales AI infra: At the RAISE Summit, Supermicro CEO Charles Liang detailed the company’s strategy to expand globally, lead in energy-efficient liquid cooling, and embrace AI to power the next wave of infrastructure.
Tech comes at you fast: Data center technologies are changing so quickly that by the time a facility gets built, the planned tech is often already obsolete. Is there a case for slowing down builds until the hype shakes out?
Too hot to handle: Power-intense AI infrastructure runs hot. So hot, in fact, that fire risk is no small matter. A recent fire at a data center used by Elon Musk’s X started in a power cabinet.
Quantum conversations: Amid all the talk about data centers and AI infrastructure, why aren’t we hearing more about quantum’s requirements? Turns out, quantum has a business case problem.
Powering data centers: Vertiv and nuclear energy company Okla are collaborating on solving power and heat issues for data centers and colocation facilities. They’re developing designs that would enable data centers to use fission-based power.
American AI Action plan: What’s in the White House’s sweeping AI Action Plan? Federal support for American AI tech and data centers. What’s out? Limitations on AI development, and LLMs that factor DEI into responses.