News

A Red Sea submarine cable break impacted Microsoft Cloud services: On September 6, 2025, Microsoft published an announcement on its Azure service status page, confirming that multiple submarine cables in the Red Sea region had been severed. This incident directly impacted network connectivity in the Middle East, South Asia, and Europe, causing significant latency for Microsoft Cloud services and forcing international traffic to reroute, increasing latency and congestion.

Broadcom Launches 51.2TB/s Tomahawk Ultra Switch: At the Hot Chips 2025 conference, Broadcom highlighted its newly announced 51.2TB/s Tomahawk Ultra switch, which the company describes as a "low-latency scale-up switch purpose-built for HPC and AI applications." The Tomahawk Ultra is the successor to Broadcom's 102.4TB/s Tomahawk 6 switch and supports the company's strategy to drive Ethernet adoption in both scale-up and scale-out applications.

NVIDIA's Rubin GPU has successfully taped out: NVIDIA has announced the successful tapeout of its Rubin GPU, which is on track for mass production and commercialization. The Rubin GPU utilizes TSMC's 3nm process technology and features HBM4 video memory. It connects to the Vera CPU via the latest sixth-generation NVLink, offering an estimated bandwidth of 3.6TB/s.

H3C Launches Next-Generation 800G AI Computing Switch: Recently, at the 9th Future Network Development Conference, H3C Group, a subsidiary of Tsinghua Unigroup, launched the H3C S9828-128EP, a new-generation 800G AI computing switch. Equipped with the industry-leading 102.4T ultra-high-performance switching chip and housed in a compact 4RU design, this switch offers up to 128 800G OSFP full-speed ports, achieving comprehensive improvements in performance, latency, energy efficiency, and operational maintenance.

IBM and NASA release Surya, an open-source AI model: IBM and NASA have collaborated to release Surya, an open-source AI model focused on predicting solar weather and designed to protect critical infrastructure from solar storms. Based on the Hugging Face platform, the model provides analysis of high-resolution solar observation data and includes a massive heliophysics dataset. This marks NASA's first open-source space weather AI tool.

Digi International Launches Enhanced Digi Connect EZ Series Device Servers: Digi International, a leading global provider of IoT connectivity products and services, announced the Digi Connect® EZ 8/16/32, the latest addition to Digi's next-generation device server family, providing secure, scalable connectivity for enterprise, commercial, and industrial automation applications. The series combines state-of-the-art hardware with the powerful Digi Accelerated Linux (DAL) operating system, offering cellular network support and centralized management with Digi Remote Manager.

The liquid-cooled server industry is poised for a boom: By August 2025, the liquid-cooled server industry was poised for a boom, with numerous companies seeing a surge in orders. Inspur Information can produce 36,000 liquid-cooled servers per month, with orders already booked through next year. A subsidiary of Sugon secured a 61MW liquid cooling project in Malaysia. Invair became a certified liquid cooling equipment supplier for NVIDIA. Shenling Environment's order volume doubled.

Russia's WhatsApp Restrictions Affect Server Connections: The Russian government has confirmed that starting September 1, 2025, it will officially restrict the use of Meta's messaging app, WhatsApp, nationwide. This includes a ban on pre-installation on new mobile devices and gradual restrictions on access, downloads, and core functionality (such as server connections) on existing devices. This will impact foreign traders' ability to communicate with Russian clients via WhatsApp, potentially halting order follow-up and business negotiations. Foreign traders will need to migrate their communication channels, such as using official Russian apps like Max Messenger.

Supermicro Lowers Revenue Target: On August 5th, Supermicro announced its second-quarter 2025 results, showing revenue growth of 7.5% year-over-year to $5.76 billion, missing the market estimate of $5.89 billion. The company also lowered its revenue target, projecting $33 billion for fiscal 2026, significantly lower than its previous target of $40 billion. This is due to Nvidia's AI chip upgrades, which have hindered the process of disposing of older product inventory as customers await products equipped with the new generation of chips.