America is optimistic about the continuous surge in CPU demand! UBS raises target prices for AMD and ARM, while Bank of America is bullish on TSMC and Sunlight. The latest report from UBS Group points out that Agentic AI will drive server CPU demand significantly beyond traditional cycles. In this context, AMD continues to expand its server market share with strong product competitiveness, while Arm Holdings ARM architecture is increasingly favored by cloud service providers and AI infrastructure vendors due to its energy efficiency advantages. At the same time, Bank of America's latest report believes that the rising demand for CPUs will simultaneously drive the demand for advanced process nodes and advanced packaging technologies, thereby benefiting leading manufacturers in these fields such as TSMC and ASE ASX. The common basis for the judgment of the two institutions lies in the evolution of AI infrastructure - from simply pursuing computing power scale to building more complete computing systems. Although GPUs are still responsible for core training and inference tasks, CPUs are playing an increasingly critical role in data management, agent execution, multi model coordination, and system scheduling. Bank of America predicts that the global server CPU market will exceed $170 billion by 2030. As AMD and ARM continue to capture market share, more and more high-end CPU orders will flow to advanced process foundries such as TSMC. In addition, the popularity and increasing complexity of chiplet architecture are expected to benefit advanced packaging manufacturers such as Sun and Moon. The rise of Agentic AI brings CPUs back to the core In the past few years, the development of AI infrastructure has mainly focused on GPUs, which play auxiliary roles such as data scheduling, storage management, and task coordination to a large extent. As AI applications shift from training to inference and further evolve towards Agenetic AI, the importance of CPUs is rapidly increasing. In the Agentic AI scenario, the model must plan tasks, invoke tools, access databases, manage states, and execute complex workflows. These tasks do not rely solely on matrix operations, but involve a large amount of sequential processing, low latency requirements, and high I/O demands - all of which are more suitable workloads for CPUs. The CPU is evolving from the traditional "main processor" to the "control center" of AI systems. If the CPU cannot efficiently handle data scheduling, memory management, and task coordination, even if GPU performance remains strong, its utilization may be compromised. Driven by this trend, the global server CPU market is expected to enter a new growth cycle, expanding to over $170 billion by 2030- nearly five times its projected level for 2025 (approximately $35 billion). The independent AI server market is favorable for x86, with AMD taking the lead Against the backdrop of a surge in CPU demand, UBS Group is increasingly optimistic about AMD's long-term growth prospects. The organization believes that standalone Agentic AI servers are rapidly emerging, and such applications place greater emphasis on the number of cores and multi-threaded performance - which are precisely AMD's strengths. According to the latest estimate from UBS Group, x86 and ARM architectures will occupy approximately 60% and 40% of the market share, respectively, in the independent AI server market, with AMD expected to capture the vast majority of x86 incremental demand. Based on higher market share expectations, UBS Group has raised its AMD server CPU revenue forecast from $21 billion in 2027 to $23 billion, and from $27 billion to $29 billion in 2028. The agency further predicts that AMD's server CPU revenue may reach $50 billion by 2030, higher than the previous estimate of $41 billion. Accordingly, UBS Group has significantly increased AMD's target price from $455 to $670. Targeting 70% of the head node segment, ARM ecosystem accelerates expansion Besides AMD, ARM is another key area that UBS Group has expressed strong optimism about. The report points out that future AI infrastructure will gradually converge to two types of CPU architectures: one provides computing power for head node CPUs connected to GPUs, emphasizing low latency and high energy efficiency; The other type supports independent Agentic AI servers, with priority given to the number of cores and throughput capacity. UBS Group believes that ARM architecture has significant advantages in the previous market. Products such as NVIDIA Grace and Vera, Google Axion, and AWS Graviton are rapidly expanding. The organization expects that by 2030, ARM based architectures may capture about 70% of the head node CPU market and account for nearly 50% of the total server CPU revenue. Based on an optimistic outlook for AI CPUs, UBS Group has raised its target price for ARM from $260 to $470. TSMC and Sunlight enter a new round of capacity expansion The surge in CPU demand is becoming the new growth engine for the semiconductor industry after GPUs. Bank of America pointed out that this trend not only benefits chip design companies, but will also directly translate into profits in the manufacturing and testing processes. According to Bank of America's estimates, the global semiconductor manufacturing market related to server CPUs is expected to expand from $15 billion in 2025 to $49 billion in 2028. During this period, the proportion of outsourced production is expected to increase from 52% to 71%, reflecting the continued strength of pure OEM factories such as TSMC in the high-end CPU field. The scarcity of advanced process production capacity, coupled with the synchronous increase in volume from multiple customers and architectures, has made the outsourcing process the most certain beneficiary in this upward cycle. At the same time, AI CPUs are driving the demand for heterogeneous integration using chiplets, advanced CoWoS packaging, and system level testing, rapidly expanding the backend packaging and testing market. Bank of America predicts that the server CPU related testing market will grow from $1.9 billion in 2025 to $9.6 billion in 2028, with its share in the advanced packaging market increasing from 11% to 24%. The upgrade of the packaging value chain not only enhances the packaging value of individual chips, but also extends the processing time required for backend processes such as testing, creating a dual opportunity for leading packaging and testing manufacturers to increase both quantity and price.