#infiniband

#infiniband

As businesses contend with ever-increasing data volumes and performance-intensive applications such as AI model training, AI inferencing and high-performance computing, they need infrastructure that delivers speed, scalability and efficiency without added complexity.

The site will eventually comprise five HPE Mod Pods, which are modular datacentre pods capable of delivering 1.5mW of datacentre capability, and will be powered by electricity generated from landfill gas at the site. The Derbyshire site will go live in the second half of 2026, according to Carbon3.ai chief strategy officer Sana Kharegani.

Sorano will be available with up to 84 Zen 5 cores - up from 64 on Siena - in a power envelope of just 225 watts. AMD isn't ready to spill all the beans on its latest Epyc just yet, but based on core count alone, we surmise the chip will either feature six density-optimized Zen 5c chiplets with 14 of 16 cores enabled or 12 of the frequency-optimized Zen 5 variety with one of the eight cores fused off.

Qunnect and Cisco have unveiled what they say is the first entanglement-swapping demonstration of its kind over deployed metro-scale fibre using a commercial quantum networking system. The demonstration combined Qunnect's room-temperature quantum hardware with Cisco's quantum networking software stack. The net result of the project is regarded by the partners as being able to bring practical quantum networks closer to scalable deployment, validating a spoke-and-hub model for scaling quantum networks through commercial datacentres.

The Osaka deployment adds 100 Gbps of edge capacity and is hosted within carrier-neutral facilities operated by Equinix. This increases regional proximity, resilience, and throughput for customers serving users in Japan and nearby markets, while maintaining consistent traffic handling and security enforcement. As organizations scale across regions, maintaining low latency, stable availability, and clear operational control has become increasingly complex.

Power, rather than compute, is fast becoming the limiting factor in scaling AI data centers. That shift has prompted Peak XV Partners to back C2i Semiconductors, an Indian startup building plug-and-play, system-level power solutions designed to cut energy losses and improve the economics of large-scale AI infrastructure. C2i (which stands for control conversion and intelligence) has raised $15 million in a Series A round led by Peak XV Partners, with participation from Yali Deeptech and TDK Ventures, bringing the two-year-old startup's total funding to $19 million.

At that point, backpressure and load shedding are the only things that retain a system that can still operate. If you have ever been in a Starbucks overwhelmed by mobile orders, you know the feeling. The in-store experience breaks down. You no longer know how many orders are ahead of you. There is no clear line, no reliable wait estimate, and often no real cancellation path unless you escalate and make noise.

The British government is investing heavily in the national computing infrastructure. With an additional investment of approximately $49 million, the DAWN supercomputer at the University of Cambridge is being expanded. This is according to Neowin. This expansion will increase the total computing power of the system by a factor of six. The aim is to enable researchers and technology companies to compete more effectively with players from the United States and China.

For any IT department, these four words are the beginning of a familiar, often frustrating, journey. In our modern world, where business success is built on distributed applications and hybrid cloud architectures, the network is the circulatory system. When it fails, everything grinds to a halt. Yet, despite its critical importance, it often remains a black box-a source of blame that is difficult to prove or disprove.

Extreme Networks is investigating the acquisition of Ruckus Networks from CommScope. The deal could be worth more than $1 billion (€857 million). No decision has been made yet, and Extreme may decide not to proceed with the acquisition. This is according to Bloomberg, based on sources. At the close of trading on Monday, Extreme Networks had a market value of approximately $2.1 billion. CommScope shares rose 2.9 percent to $19.04, giving the company a market value of approximately $4.2 billion.

Our goal is to provide customer choice. Nvidia has been the market leader and AMD is the other big platform company. We see a lot of that happening now. People do not necessarily see Broadcom as a long-term partner.

Intel is making a new push into GPUs, this time with a focus on data center workloads, as the chipmaker looks to reestablish itself in a market increasingly shaped by AI-driven demand and dominated by Nvidia. CEO Lip-Bu Tan said that after hiring a senior GPU architect, the company is working directly with customers to define requirements, signaling a more demand-driven approach as enterprises and cloud providers weigh their options for accelerated computing, according to a Reuters report.

Light-based internet provider Taara, which spun out of Alphabet's "moonshot" incubator last year, just launched Taara Beam to provide 25Gbps connectivity within cities over invisible beams of light - line of sight permitting. Unlike last year's Taara Lightbridge, which connects communities separated by water and mountains at distances up to 20km (over 12 miles), the shoebox-sized Beam can be mounted to street poles and roof tops for city-wide connectivity at distances up to 10km. The 8kg (less than 20 pounds) device typically consumes about 90W.

Lenovo has announced a new series of enterprise servers and solutions specifically designed for AI inferencing workloads. These consist of three different ThinkSystem and ThinkEdge servers. "Enterprises today need AI that can turn massive amounts of data into insight the moment it's created," said Ashley Gorakhpurwalla, President of Lenovo Infrastructure Solutions Group. "With Lenovo's new inferencing-optimized infrastructure, we are giving customers that real-time advantage."

AMD clarified those estimates are based on a comparison between an eight-GPU MI300X node and an MI500 rack system with an unspecified number of GPUs. The math works out to eight MI300Xs that are 1000x less powerful than X-number of MI500Xs. And since we know essentially nothing about the chip besides that it'll ship in 2027, pair TSMC's 2nm process tech with AMD's CDNA 6 compute architecture, and use HBM4e memory, we can't even begin to estimate what that 1000x claim actually means.

On paper, Positron's next-gen Asimov accelerators, no doubt named for the beloved science fiction author, don't look like much of a match for Nvidia's Rubin GPUs. Yet, the Arm-backed AI startup boasts its inference chip will churn out five times as many tokens per dollar while using one-fifth the power of Nvidia's latest accelerators to do it. Those are certainly some bold claims, which the company contends are possible because the chip was designed to support large-scale inference workloads.

Wi-Fi 8 introduces a technology known as Seamless Multi-Link Device (SMD) to address this issue. Instead of first disconnecting from the existing connection and then establishing a new one, a device with SMD first connects to the next access point before releasing the old connection. According to Broadcom, this approach results in approximately 25 percent less packet loss and a similar reduction in latency.

When ChatGPT launched in late 2022, I watched something remarkable happen. Within two months, it hit 100 million users, a growth rate that sent shockwaves through Silicon Valley. Today, it has over 800 million weekly active users. That launch sparked an explosion in AI development that has fundamentally changed how we build and operate the infrastructure powering our digital world.

The Xeon 600 lineup spans the gamut between 12 and 86 performance cores (no cut-down efficiency cores here), with support for between four and eight channels of DDR5 and 80 to 128 lanes of PCIe 5.0 connectivity. Compared to its aging W-3500-series chips, Intel is claiming a 9 percent uplift in single threaded workloads and up to 61 percent higher performance in multithreaded jobs, thanks in no small part to an additional 22 processor cores this generation.

RAM prices have surged dramatically in recent months, with manufacturers including Kingston, Micron, and Samsung raising prices by an average of 63 percent between September and December 2025 for the most common capacities - 16 GB, 32 GB, 64 GB, and 128 GB modules sold in Europe, according to distribution market data compiled by analyst Context.

Vertiv has announced new configurations of its MegaMod HDX solution, a prefabricated power and liquid cooling infrastructure designed for environments with very high power densities. The solution is intended for applications such as artificial intelligence and high-performance computing and is available in North America and the EMEA region. According to Vertiv, the new variants respond to the rapidly growing demand for computing power and associated cooling capacity in data centers.