AI computing power chain stepping into the 2nm era! AMD fires the first shot of 2nm data center CPU, Wall Street sees the rising tide even more surging.

Leader in high-performance chips for PC and data centers in the United States, AMD, announced on Thursday that the next generation AMD EPYC processor "Venice" based on its long-standing foundry partner Taiwan Semiconductor Manufacturing Co., Ltd. Sponsored ADR 2nm advanced process technology has started mass production. According to AMD officials, EPYC "Venice" CPU is the 6th generation EPYC data center CPU, the world's first 2nm process level data center CPU, and also the "world's first 2nm process level HPC/data center CPU to enter mass production stage", marking the 2nm level data center CPU about to move towards large-scale commercial mass production. As AMD fires the first shot in the 2nm advanced process level data center CPU, at the same time Anthropic's groundbreaking release of Claude Cowork and OpenClaw, AI super agent tools capable of autonomously executing tasks, has exploded in 2026, driving the data center CPU into a serious supply shortage. Against the backdrop of the explosive demand for data center CPUs, Wall Street financial giants are increasingly optimistic about the two major x86 architecture CPU super giants - Intel Corporation (INTC.US) and AMD (AMD.US), as well as the ARM architecture owner, Arm Ltd. (ARM.US), all three CPU giants have seen a significant rise in their stocks this year, with Intel Corporation's stock price soaring by as much as 222% as of 2026. For the past two years, AI narratives have been dominated by GPUs, with CPUs serving as a "supporting role" in the AI arms race; but as open-source agents like OpenClaw, which execute AI workflows, dominate inferencing workloads, data arrangements, task scheduling, memory access, network communication, and multi-tool invocations, the market has come to realize: without a powerful CPU at the helm, GPU clusters cannot operate efficiently. This fundamentally brings the CPU back from being "underestimated infrastructure" to the center stage of the chip industry, with a very clear "Renaissance" retro trend. As open-source AI agent tools like OpenClaw explode in popularity, the wave of AI agents is rapidly sweeping across the globe, and AI compute architecture bottlenecks are shifting from GPU-centric matrix multiplication to data center CPUs at the core of control flow, task scheduling, memory/IO coordination, facing a severe supply shortage situation for high-performance CPUs in ultra-large-scale AI data centers. At a time when AI agents are sweeping the globe, the mainstream AI compute investment narrative is transitioning from a "GPU-centric single-point compute competition" to an "AI agent-driven full-stack compute system", with the next round of excess alpha returns no longer exclusive to the AI GPU/AI ASIC field, but rather systemically diffusing into the full-stack AI compute infrastructure layer, including data center CPUs, DRAM/NAND/HBM storage, AI PCBs, liquid cooling systems, data center optical interconnect systems, ABF substrates/glass substrates and extensive wafer foundry services, and in this narrative shift towards AI, CPUs, optical interconnects, and storage chips may be the biggest winners. AMD joins forces with Taiwan Semiconductor Manufacturing Co., Ltd. Sponsored ADR in the era of 2nm AI compute infrastructure "Venice" EPYC is the industry's first HPC product to enter mass production using Taiwan Semiconductor Manufacturing Co., Ltd. Sponsored ADR's advanced 2nm process technology, and it is the 6th generation EPYC data center CPU. However, it is important to note that "mass production climb" does not mean "already fully shipped". It typically means that the product has progressed from wafer fabrication, bring-up stage, validation stage to the production climb and commercial production preparation stage, closer to formal large-scale customer supply, but still needs to go through yield, packaging, platform certification, customer validation, and shipping pace climb. Official announcements from AMD on May 21st showed that their next-generation AMD EPYC processor, codenamed "Venice", is starting mass production climb in China's Taiwan using Taiwan Semiconductor Manufacturing Co., Ltd. Sponsored ADR's most advanced 2nm chip process technology, with plans for future mass production climb in Taiwan Semiconductor Manufacturing Co., Ltd. Sponsored ADR's Arizona wafer factory. This milestone in AMD's data center CPU roadmap demonstrates the company's continuous progress in delivering leading performance and efficiency required for next-generation cloud computing, enterprise, and AI compute infrastructure. AMD emphasized in their announcement that "Venice" is the industry's first HPC product to enter mass production phase using Taiwan Semiconductor Manufacturing Co., Ltd. Sponsored ADR's most advanced 2nm process technology. Dr. Lisa Su, Chairman and CEO of AMD, stated: "Advancing 'Venice' mass production climb on Taiwan Semiconductor Manufacturing Co., Ltd. Sponsored ADR's 2nm advanced chip process technology signifies an important step forward in accelerating the next generation of AI infrastructure. As AI and agent-based workloads rapidly expand, customers need platforms that can quickly move from innovation to production. Our deep collaboration with Taiwan Semiconductor Manufacturing Co., Ltd. Sponsored ADR is helping AMD deliver the speed and scale required for this era, bringing leading computing technology to market." With AI applications expanding from training to inference and increasingly complex agent-based workloads, CPUs are becoming more critical in scaling AI infrastructure, responsible for coordinating data flow, networking, storage, security, and system orchestration in the data center. As AMD continues to strengthen its position in the server market, the mass production climb of "Venice" signifies growing customer demand for EPYC processors to support modern large-scale cloud computing inference endpoints, enterprise HPC, and ultra-large-scale AI data center deployments. Based on Zen 6/Zen 6c and Taiwan Semiconductor Manufacturing Co., Ltd. Sponsored ADR 2nm, the Venice processor can scale up to 256 cores, approximately a 33% increase compared to the current EPYC Turin with a maximum of 192 cores; AMD also claims that Venice can provide approximately 1.7x performance improvement, approximately 1.3x thread density improvement, and support higher memory bandwidth, PCIe Gen6, and other platform capabilities. Compared to existing data center CPUs, Venice is positioned higher than the current AMD EPYC Turin and Intel Xeon 6 Granite Rapids/Sierra Forest generations. The 5th generation EPYC has shown strong competition against Xeon 6 in many multi-threaded server workloads, with AMD's own comparison showing the performance comparison between 5th generation EPYC and Intel Xeon 6 in the multi-core range; if Venice delivers on its promised 256 cores, 2nm efficiency, and Zen 6 IPC/platform upgrades, it could further amplify AMD's advantages in high-density cloud, HPC, databases, virtualization, AI head-nodes, and inference orchestration servers. AMD also plans to further expand Taiwan Semiconductor Manufacturing Co., Ltd. Sponsored ADR 2nm process technology in its data center CPU roadmap with "Verano". "Verano" is also a 6th generation EPYC processor optimized for leading performance per watt. The processor aims to support cloud and hyperscale AI compute workloads and is expected to introduce advanced memory innovations, including LPDDR, to provide the CPU performance, bandwidth, and efficiency required for increasingly power-constrained workloads and applications. Venice is the mainline/leading 2nm data center CPU for the 6th generation EPYC, while Verano is an expanded version that emphasizes memory innovations such as LPDDR/SOCAMM2 to accommodate Agentic AI and future Instinct GPU platform requirements. Two major x86 chip giants embark on a wild bull market together Wall Street analysts are expanding the AI compute infrastructure narrative from a "GPU monopoly/single-core-centric model" to a "AI GPU/ASIC+CPU+HBM/DRAM/NAND storage chips+ optoelectronic interconnect-driven high-speed data center connectivity system" full-stack compute reassessment. Against a backdrop of explosive demand for data center CPUs, Citigroup, a Wall Street financial giant, has significantly raised its 12-month target stock prices for the two major x86 architecture CPU super giants - Intel Corporation (INTC.US) and AMD (AMD.US) this week. This financial giant has also significantly raised its expectations for the size of the data center CPU market and the overall CPU market in its latest report. At the 54th Annual Morgan Stanley Global Technology, Media, and Communications Conference, Intel Corporation CEO Pat Gelsinger announced that Intel 18A (i.e., sub-2nm advanced chip process below 1.8nm) has started supporting Panther Lake production, with a monthly yield increase of about 7%, surpassing Intel Corporation's internal expectations. Gelsinger also stated that as the focus of AI compute infrastructure transitions from training to inference, CPUs are becoming increasingly important and indispensable in the AI era, with the CPU-to-GPU ratio accelerating from 1:8 towards 1:1, and even up to 4:1. Additionally, Intel Corporation's business plans indicate they are actively pursuing ASIC business, offering customized AI CPU or AI GPU chip solutions. The shift of AI from training to inference and Agentic AI will significantly increase the strategic importance of CPUs in AI data centers. While GPUs are responsible for large-scale matrix computations, CPUs handle scheduling, I/O, memory management, task orchestration, security isolation, database access, network stack, and multi-agent workflow execution; as AI applications transition from "single inference" to "continuous software labor", the demand for CPUs shifts from traditional server refresh cycles to AI infrastructure expansion cycles. Citi's latest model also echoes this point: they anticipate the total potential market size of data center server CPUs [i.e., CPU TAM] to expand from $29.3 billion in 2025 to $131.5 billion in 2030, at a compound annual growth rate of approximately 35%, and they have significantly raised their target price for Intel Corporation from $95 to $130 and for AMD from $358 to $460. A report from a senior analyst team led by Atif Malik at Citigroup adds, "AI agent-led data center CPU market are expected to experience a stunning 185% compound annual growth rate, reaching $59.4 billion by 2030, and are expected to account for 45% of the overall market share by 2030, becoming the fastest-growing segment." Data compiled by TipRanks show that the highest target price for AMD is up to $625, corresponding to about 40% potential upside, with AMD's stock price already rising by 110% this year, making it one of the "super bull stocks" in the global stock market. The highest target price of $625 on Wall Street comes from Baird analyst Tristan Gerra, who recently raised AMD's target price from $300 to $625 and maintained an "Outperform" rating, with his bullish logic primarily based on the unprecedented surge in demand for CPU/GPU due to Agentic AI, and the revaluation of AMD's data center platform value. Analyst Tristan Gerra stated that AMD's opportunities in AI server CPUs, GPUs, and accelerated computing platforms have been repriced, especially in the face of the AI inference surge, Agentic AI, EPYC data center CPUs, Instinct GPUs, and the shared drive for the next generation of AI infrastructure expansion, the market is beginning to see AMD as a key beneficiary of the "CPU+GPU+super platform solution" in the AI compute industry chain.