The Ethernet for Scale-Up Networking (ESUN) Alliance officially launched at the 2025 Open Compute Project (OCP) Global Summit, uniting North America's AI heavyweights, including OpenAI, Meta, AMD, Nvidia, Microsoft, and Broadcom. Its goal is to create an open Ethernet interconnect standard to address AI chip connectivity fragmentation and redefine global AI infrastructure. The collaboration marks a rare alignment across major rivals in pursuit of interoperability.
The global AI wave is driving a "golden decade" for the printed circuit board (PCB) industry, sparking revolutionary innovations in design, manufacturing, and materials that accelerate its integration with advanced semiconductor sectors. This trend has not only exposed upstream capacity pressures but also intensified geopolitical risks, profoundly reshaping global PCB investment strategies.
Micron Technology has started customer sampling of its new 192GB SOCAMM2 (Small Outline Compression Attached Memory Module), a low-power DRAM module designed for AI data centers.
As global demand for AI computing power skyrockets, the data center landscape is undergoing a significant transformation — shifting away from standalone server performance and toward system-level, rack-based upgrades. At the OCP Global Summit 2025, the industry spotlight moved beyond individual servers to full racks designed with 800V direct current (DC) power, liquid cooling, and high-bandwidth network integration.
Taiwan's offshore wind maritime engineering firm Dong Fang Offshore (DFO) is set to transition from the over-the-counter (OTC) market to the Taiwan Stock Exchange mainboard in mid-November 2025. The company is rapidly expanding its fleet to capture strong growth opportunities in AI-fueled undersea communication cable demand.
Samsung Electronics' Device Solutions (DS) division CTO Song Jae-hyuk expressed strong confidence in the revival of Samsung's foundry business and reported smooth progress on the company's 2nm process during a semiconductor industry briefing hosted by South Korea's chief presidential secretary for policy, Kim Yong-beom.
Samsung Electronics is reportedly set to start production of its AI-powered home assistant robot, Ballie, at its facility in Ho Chi Minh City, Vietnam, suggesting that the long-delayed device is finally nearing mass production.
Compal Electronics Inc., one of Taiwan's leading contract manufacturers, has reportedly secured orders from Dell Technologies to produce AI servers beginning in 2026 — a milestone that underscores the company's growing foothold in the booming artificial intelligence hardware market.
Demand for AI servers remains robust, prompting industry players to actively develop and continuously update product architectures. Supply chain observers note that as AI endpoint requirements grow more complex, future server platforms will move away from one-size-fits-all standards toward customized configurations tailored to specific applications. This shift also increases market demand for rack designs with flexible, adaptable configurations, a key focus for server vendors going forward.
As Taiwan's semiconductor and AI industries continue to surge, universities are facing an unexpected casualty of economic success: a talent drain from academia. Graduate students, lured by soaring salaries in the tech sector, are abandoning plans for doctoral studies and, in some cases, even leaving master's programs early after securing job offers from chipmakers.
Quanta Senior Vice President Mike Yang stated that Quanta will continue expanding production capacity in the US and Thailand. Although Taiwan is no longer considering factory expansion, he expects 2025 to once again be an excellent year. Yang shared that Quanta aims to become a collaborator in intelligent production systems.
More coverage
