➀ PCI-SIG finalized PCIe 7.0 spec with 128 GT/s per lane, doubling PCIe 6.0 speeds and enabling up to 512GB/s bidirectional bandwidth for data centers.
➁ PCIe 8.0 pathfinding began, targeting 1TB/s bidirectional bandwidth by 2030 through potential 256 GT/s per lane, though optical interconnects remain uncertain.
➂ PCIe 6.0 compliance tests delayed to 2025, while PCIe 7.0 devices are expected to launch between 2028–2029 after rigorous signal integrity challenges at 32 GHz frequencies.
➀ Qualcomm acquires Alphawave Semi for $2.4B to enhance data center and AI chip capabilities;
➁ Alphawave's high-speed connectivity tech will complement Qualcomm's Oryon CPU and Hexagon NPU cores;
➂ The deal signals Qualcomm's intensified competition with Intel, AMD, and Nvidia in server processor markets.
➀ Researchers at the University of Tartu repurposed old smartphones into compact 'tiny data centers' to reduce e-waste and enable real-time edge computing.
➀ The system, featuring 3D-printed casings and custom software, successfully monitored marine life underwater without human intervention.
➂ Potential applications extend to public transit optimization and other edge computing scenarios, offering a sustainable solution for obsolete devices.
➀ NVIDIA continues to benefit from soaring demand for its data center solutions and achieves record first quarter gaming revenue.
➁ Despite $4.5 billion charge from chip inventory write-down, the company's revenue for Q1 fiscal 2026 reaches $44.1 billion.
➂ Export restrictions to China result in $2.5 billion lost H20 chip sales, but NVIDIA manages to thrive.
➀ X, formerly known as Twitter, is experiencing a major outage due to a fire at a data center in Oregon;
➁ The fire broke out at the PDX11 data center, which is leased by X owner Elon Musk from Digital Realty;
➂ No injuries were reported, and the fire was contained to the data center.
➀ Qualcomm partners with Saudi Arabia's HUMAIN AI to develop AI data centers using its CPU and edge computing solutions;
➁ The move marks Qualcomm's return to the data center CPU market after its failed Centriq project in 2019, leveraging Nuvia's Arm-based technology;
➂ The collaboration includes building a semiconductor design center in Saudi Arabia and follows Nvidia's recent AI GPU deployment in the region.
➀ Intel reported Q1 revenue of $12.7 billion, flat y-o-y with a gross margin of 36.9% down 4.1 points y-o-y.
➁ The company expects Q2 revenue between $11.2-12.4 billion.
➂ CEO Lip-Bu Tan emphasized the need for better execution and operational efficiency to regain market share and drive sustainable growth.
➀ Some AI data centers in China are reportedly holding large stocks of China-specific 48GB Nvidia RTX 4090D GPUs and reselling them as new cards.
➁ Companies are dismantling and refurbishing the GPUs to generate profits.
➂ The report states that an AI data center requires a utilization rate of more than 70% to 75% to turn a profit, but the activation rates remain below 20%.
➀ IQE和X-FAB签署了一项联合开发协议（JDA），旨在创建一个基于欧洲的GaN功率器件平台解决方案；
➁ 该协议为期两年，双方将合作开发650V GaN器件；
➂ 该协议将利用IQE的GaN外延设计和工艺专长以及X-FAB的技术开发能力和器件制造能力，为汽车、数据中心和消费应用提供优化的技术-衬底组合。
At NVIDIA GTC 2025, we got to tour Pegatron's booth, where we saw the Pegatron NVIDIA GB300 NVL72 rack and other GPU systems. The GB300 NVL72 racks are liquid cooled with a busbar for power in the rear. The sleds have four Blackwell modules along with two NVIDIA Grace CPUs and are liquid cooled.
The article also mentions that Pegatron is integrating Delta power supply shelves into their racks and that the NVIDIA GB300 NVL72 is neat, with several other GPU systems on display.
➀ IBM has introduced the z17 mainframe, featuring the Telum II chip with enhanced AI performance.
➁ The Telum II processor offers 70% higher general-purpose performance and 50% improved AI capabilities.
➂ IBM also unveiled the Spyre AI accelerator card for additional AI performance.
➀ Marvell is selling its Automotive Ethernet division to Infineon for $2.5 billion.
➁ The division has a sales forecast of $225-$250 million this year with a 60% gross margin.
➂ The acquisition includes Marvell’s Brightlane Automotive Ethernet portfolio, which supports network data rates up to 10 Gbps.
➀ Micron confirms plans to raise memory prices due to strong demand for DRAM and NAND flash.
➁ The price hike is expected to continue through 2025 and 2026.
➂ Factors include supply constraints and growing demand from AI, data centers, and consumer electronics.
➀ China's AI data center expansion in 2023-2024 has led to significant underutilization of billions of dollars in infrastructure;
➁ Rushed projects resulted in poor planning and technical irrelevance;
➂ Market shift towards inference workloads has reduced demand for high-end GPUs, leading to a decrease in rental prices.
➀ The article discusses the importance of virtualization licensing in server hardware costs.
➁ It focuses on AMD's EPYC processors and their optimization for virtualization applications.
➂ The author explores how newer processors offer performance improvements, which can lead to significant cost savings in licensing.
➀ Keysight Technologies has introduced two oscilloscopes designed for testing 1.6T optical transceivers: a single optical channel DCA-M and a dual optical channel DCA-M Sampling Oscilloscope. These instruments are engineered for high optical measurement sensitivity and integrated clock recovery.
➁ The oscilloscopes feature ultra-low optical channel noise and minimal intrinsic jitter, ensuring accurate high-data-rate analysis with jitter below 90 fs. They simplify test setup and compliance assurance while optimizing production efficiency.
➂ Keysight's advanced oscilloscopes support high-speed optical signal analysis and are expected to accelerate AI-driven data center innovations by providing precise measurements for 1.6T transceivers.
OpenAI's release of GPT-4.5 has been delayed due to a shortage of GPUs. CEO Sam Altman mentioned on X that the rollout will be staggered and will involve adding tens of thousands of GPUs next week. This shortage is prompting OpenAI to develop its own AI silicon in partnership with Broadcom.
The high cost of GPT-4.5, at $75 per million input tokens and $150 per million output tokens, is a concern, but Altman emphasizes its unique intelligence.
NVIDIA's GPUs are in high demand, with Blackwell GPUs sold out until October. The expansion of AI infrastructure, including massive AI supercomputers and data centers, is driving this demand.
➀ Lumotive, a specialist in programmable optical semiconductors, has raised $45 million in a Series B funding round.
➁ The funding will be used to accelerate growth in global market expansion, data center AI infrastructure, and aerospace and defense applications.
➂ Lumotive's LCM technology enables ultra-reliable optical circuit switches for high-performance optical switching solutions.
➀ The SEMI Silicon Manufacturers Group (SMG) reported that 2024 silicon wafer shipments fell 2.7% to 12,266 million square inches and wafer revenue contracted 6.5% to $11.5 billion.
➁ Demand for silicon wafers in the second half of 2024 started to recover from the industry downcycle seen in 2023, and is expected to continue into 2025.
➂ Generative AI and new data-center construction are driving demand for advanced foundry and memory devices like HBM, while most other markets are still recovering from excess inventory.
➀ Justin Hotard, the executive vice president and general manager of Intel's Data Center and AI Group, has left Intel to become the CEO of Nokia.
➁ Intel has appointed an interim head for its datacenter and AI unit and will start searching for a new permanent general manager.
➂ Hotard joined Intel in early 2024, and his tenure was marked by successes and challenges, including the launch of new CPUs and missed sales expectations.
