AMD将在韩国设立AI研究中心，打造开放式AI基础设施生态
Related
AMD (AMD.O) is collaborating with TSMC (TSM.N) to increase production capacity for its next-generation CPUs.
According to Odaily Odaily, market sources indicate that AMD (AMD.O) is collaborating with TSMC (TSM.N) to increase production capacity for its next-generation CPUs. (Golden Ten)
Bank of America raised its CPU market size forecast due to AI demand and also raised its price targets for AMD, ARM, and Intel.
According to Mars Finance, on June 11, Bank of America raised its forecast for the server CPU market size, citing AI demand as the primary driver. Bank of America analyst Vivek Arya stated that driven by demand for agentic AI, Bank of America increased its forecast for the total potential market size of server CPUs by 2030 from $125 billion to over $170 billion. He believes that AI systems capable of planning and executing multi-step tasks will further increase CPU usage beyond accelerators. Bank of America also raised its target prices for AMD, ARM, and Intel, citing enhanced long-term semiconductor growth prospects and continued expansion in chip demand.
Intel's Chen Liwu: As AI shifts from training to inference, the CPU to GPU ratio may reach 4:1.
Mars Finance reported on May 20th that at JPMorgan Chase's 54th Annual Global Technology, Media & Communications Conference, Intel CEO Chen Liwu stated that the Intel 18A has supported mass production of Panther Lake, with yield improving by approximately 7% per month, exceeding internal expectations. The Intel 14A 0.5 PDK has been released, and the 0.9 PDK is planned for release to external customers in October. The team has already begun long-term planning for the 10A and 7A process nodes. As AI shifts from training to inference, CPUs are becoming increasingly important and indispensable in the AI era. The CPU to GPU configuration ratio is approaching 1:1 from 1:8, and even reaching 4:1. Currently, Intel is actively pursuing ASIC business, providing customized solutions. (Wide Angle Observation)
Key Wall Street Ratings Summary: UBS raises AMD's price target to $670, Seaport maintains "Sell" rating on Nvidia.
According to Mars Finance, on Wednesday, June 24th, several Wall Street institutions released significant rating adjustments, involving companies such as Nvidia, Tesla, AMD, Reddit, and Arm. Seaport reiterated its "sell" rating on Nvidia, stating that Nvidia's on- and off-balance sheet burdens continue to increase, and Broadcom's entry into the revolving financing trend indicates that the AI industry is facing increasing financing constraints. Stifel initiated coverage of BlackBerry with a "buy" rating, stating that the market still misunderstands BlackBerry; the company is no longer just an automotive supplier, but a key software layer in the physical AI stack, and it collaborates with chip companies such as Nvidia, Qualcomm, and AMD. UBS reiterated its "buy" rating on AMD and raised its target price from $470 to $670, stating that it is more positive on AMD as discrete CPU racks begin to gain market acceptance. TD Cowen reiterated its "Buy" rating on Arm and raised its price target from $265 to $475, stating that Arm's $15 billion AGI CPU target for fiscal year 2031 is reasonable, but GPU-to-CPU pairing and single-core pricing remain key variables. RBC reiterated its "Outperform" rating on Tesla, projecting Q2 deliveries of 405,000 vehicles, higher than the Visible Alpha consensus forecast of 401,000. Needham reiterated its "Buy" rating on Reddit, noting its rare competitive position and minimal impact from the proxy AI storm. Additionally, Citizens initiated coverage on Bitdeer, CleanSpark, Keel Infrastructure, and MARA Holdings, all with "Outperform" ratings, citing these Bitcoin mining companies' efforts to repurpose existing power capacity to provide high-performance computing capabilities to hyperscale customers and large enterprises.
Sources indicate that Nvidia plans to market its Vera AI CPU to Chinese customers, and some cloud vendors are reportedly planning to begin testing and deployment.
According to sources familiar with the matter, Nvidia has begun marketing its first standalone central processing unit (CPU) product, Vera, to Chinese customers. Designed specifically for Agentic AI (Autonomous Intelligence) systems, the chip is now in mass production, marking Nvidia's attempt to further expand its market in China through its CPU products. Sources say that some Chinese customers have expressed interest in Vera. One major Chinese cloud computing company plans to purchase more than 300 servers equipped with dual Vera CPUs for testing, and will decide whether to expand the purchase after the testing is completed. Vera, built on the Arm Holdings architecture, is NVIDIA's first standalone CPU product. NVIDIA previously stated that Vera's performance in AI agent-related computing tasks is 1.8 times that of competing products, and expects the product to contribute approximately $20 billion in revenue by the end of this fiscal year (ending January next year). The report points out that as the AI industry's focus gradually shifts from model training to inference, CPUs and custom chips are gaining more attention. Vera also puts Nvidia in direct competition with Intel and Advanced Micro Devices (AMD), which have long dominated the server CPU market. Sources familiar with the matter indicate that due to strict U.S. restrictions on high-end GPU exports, CPUs face relatively fewer regulatory hurdles in the Chinese market compared to GPUs. Currently, some Chinese customers plan to deploy Vera chips in overseas data centers for testing. However, software ecosystem compatibility and the existing deployment system of domestically produced AI chips may still affect the subsequent large-scale adoption of Vera. (Reuters)
AMD announces mass production of its next-generation EPYC processor "Venice" using TSMC's 2nm process.
Odaily Odaily reported on May 21 that AMD announced that its next-generation AMD EPYC central processing unit, codenamed "Venice," has officially entered mass production. The chip uses TSMC's most advanced 2nm process technology and is the industry's first 2nm product to enter mass production in the high-performance computing (HPC) field. In addition, AMD plans to extend its 2nm process for data center CPUs to its upcoming "Verano" product. (Jinshi)