Susana Schwartz
Is the AI-RAN frenzy based on fantasy or truth?
The AI-RAN Frenzy in the global telecom infrastructure market is going to push revenues to $6.18 billion by 2032, but can AI-RAN deliver technical and financial outcomes, at scale? Big companies like T-Mobile, Bell Canada, Ericsson, SoftBank, Red Hat, US Cellular, Samsung, and Nokia are testing it out, with some reporting a 10% improvement in spectral efficiency and 20% higher downlink throughput.
The AI-RAN Alliance, TM Forum and other consortia and organizations are starting to look more closely at how AI-RAN fits into the bigger Open RAN picture, and what the real costs will be in terms of energy consumption, security concerns, and data privacy. The proof’s in the pudding, so hopefully more pilots, benchmarks and real-world examples (that include rural and urban environments) will emerge.
Check out Vish Nandlall’s opinion about it in the “Top Stories” below, and scroll down to see today’s headlines.
Susana Schwartz
Technology Editor
RCRTech
AI Infrastructure Top Stories
Is AI-RAN an engineering and economic trap?: Analyst Vish Nandlall questions the hype around AI-RAN, pondering if the promise of a centralized brain that understands the network the way GPT-4 understands language is just a “fantasy.” Will network foundation models and AI-RAN save telecom?
Neoclouds are attracting new attention: Why are enterprises turning to neoclouds, and what are the strategic hurdles they face? ABI Research principal analyst Reece Hayden examines key drivers, like sovereignty, lower GPU costs, and power, and the possible hurdles.
AI agents, not humans, will soon choose networks: Sophisticated AI agents are becoming digital proxies tasked with managing their owners’ digital lives, which represents a monumental shift from B2C to B2Ai (Business-to-Agent), a transition that redefines customer acquisition, retention, and network strategy.
AI Today: What You Need to Know
Chinese photonic quantum chip marks huge leap in computing power: A new photonic Quantum AI chip designed by a Chinese university-affiliated research institute and a Shanghai start-up is purported to outstrip leading NVIDIA graphics processors by a factor of 1,000. If the performance gains are true, it could become one of the most significant advances in next-gen computing.
Omdia’s take on DT multi-agent RAN optimization: Deutsche Telekom is proving agentic AI is not just hype, becoming the first to go live with multi-agentic RAN optimization and network management. Through an event agent, monitoring agent, and remediation agent, it has achieved 75% autonomy. Here’s Omdia’s take on the news.
Proposed filing fee for semiconductor patents raises eyebrows: US semiconductor patent proposal is another front in the China trade war, but is it “daft”? A proposal by the US Patent and Trademark Office to charge for semiconductor patents on the ‘value’ basis rather than a flat fee, with a proposed filing fee between 1% to 5% of the ‘value’ of the patent.
Record-breaking quantum semiconductor: Scientists at the University of Warwick and the National Research Council of Canada say they have achieved the highest electrical conductivity ever recorded in a silicon-compatible material, marking what they describe as a major step toward faster, more efficient electronic and quantum devices.
Cooling-as-a-service delivers end-to-end cooling management: Ecolab has introduced a Cooling as a Service (CaaS) program to optimize cooling performance in data centers and high-performance computing environments worldwide. The program combines hardware, chemistry, digital monitoring, and field expertise into a single offering designed for AI-intensive workloads.
Large, asset-backed GPU purchase: AlphaTON Capital announced an $82.5 million GPU infrastructure investment to acquire and deploy a 1,000+ Nvidia B200 GPU cluster. It will power Telegram’s Cocoon AI and provide external GPU rental services.
