In today’s lead stories, we look at some of the big questions around AI and AI infrastructure. First — where are all these GPUs going, anyway? Nvidia’s disclosure that just two customers account for nearly 40% of its revenues sheds some light on why GPUs are so expensive and hard to come by. Next question: What’s the value of AI? Or perhaps we can reframe it, as my colleague Sean Kinney does, this way: Even if most AI projects fail, what can we learn from the ones that succeed? Now, questions are great, but we round out today with a potential answer to the big question around how the heck to satisfy data center power demand: The Tennessee Valley Authority has struck a deal to make Small Modular Reactors a reality, by developing six of them across its operating area. Read on for more questions, and more potential answers to think about.
Kelly Hill
Executive Editor
RCRTech
AI Infrastructure Top 3
Nvidia’s mystery buyers: In Q2, Nvidia’s two unnamed direct customers—dubbed “Customer A” and “Customer B”—accounted for a staggering 39% of revenue, leaving investors guessing who’s powering its record AI chip sales.
Gen AI learnings from the 5%: According to a recent MIT Media Lab report, 95% of enterprise gen AI pilots fail to reach production. What do the 5% that get gen AI right know, and what does it mean for everyone else?
Six SMRs for TVA: The Tennessee Valley Authority (TVA) has signed a deal with Entra1 Energy to develop six of NuScale’s SMRs in TVA’s seven-state operating area, which will provide up to 6 GW of nuclear power for data centers.
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
Large telco model for Germany: O2 Telefónica, Tech Mahindra, and Nvidia are building a Large Telco Model to speed fault detection, optimize technician coordination, and push German telecom networks toward autonomous operations.
DT appoints sovereignty chief: T-Systems, the integration division of Deutsche Telekom, has created a new senior-level management position to direct its strategy around data sovereignty.
SKT-Schneider formalize AI infra deal: SK Telecom and Schneider Electric have formalized their partnership to build AI data center infrastructure in Ulsan, combining digital twin software, MEP systems, and group-wide cooperation.
Nationalizing data centers?: The Trump administration has taken stakes in Intel and U.S. Steel. Data centers could be potential targets, as the administration seeks to beat China in AI — and is willing to nationalize enterprises to do it.
Too much, too fast: Does the crunch in data center capacity constitute a crisis? Is it possible for a trend to get so big that it drives innovation to the point of failure? Some big thoughts on the scale of, and scaling of, AI infrastructure.
Alibaba tests AI chip: Alibaba is testing a new AI chip built in China to reduce reliance on Nvidia and foreign manufacturers, part of Beijing’s wider push to strengthen local semiconductor supply chains.
OpenAI eyes India expansion: OpenAI is considering a gigawatt-scale data center in India as part of its Stargate AI expansion, a move that could reshape Asia’s AI infrastructure landscape and boost U.S.-aligned partnerships.
Marvell sinks on AIDC slowdown: Marvell shares plunged after weak data center demand outlook, raising concerns over AI chip sales to cloud giants, uneven orders and competition.
Dell’s DC biz outruns PC biz: Dell’s Infrastructure Solutions Group, which sells servers and services into the data center market, is larger than its PC business for the first time in company history — and is likely to stay that way.
AWS launches NZ region: AWS has opened a New Zealand Region, backed by a $4.4 billion investment, delivering lower-latency cloud services, thousands of jobs, and a digital infrastructure boost.
