Bullish on continued CPU boom! UBS raises target prices for AMD and ARM, Bank of America bullish on TSMC and ASE.

```

As the AI industry moves from the training stage to the era of inference and Agentic AI, CPUs, once overshadowed by the brilliance of GPUs, are regaining the spotlight in the market.

According to a recent UBS report, Agentic AI will drive server CPU demand to significantly exceed traditional cycles. Against this trend, AMD continues to expand its server market share with strong product competitiveness, and the ARM architecture is favored by cloud providers and AI infrastructure vendors for its energy efficiency advantages. Meanwhile, a recent Bank of America report suggests that expanding CPU demand will drive simultaneous growth in demand for advanced manufacturing processes and advanced packaging, thus benefiting manufacturing leaders such as TSMC and ASE.

The common basis for the views of both institutions is that AI infrastructure is shifting from simply pursuing computing power to building more complete computing systems. GPUs are still responsible for core training and inference tasks, but the role of the CPU in data management, agent execution, multi-model collaboration, and system scheduling is becoming increasingly critical. Bank of America predicts that by 2030, the global server CPU market size will grow to over $170 billion.

With AMD and ARM continuing to expand market share, more high-end CPU orders will flow to advanced foundries such as TSMC, while the popularity and increasing complexity of chiplet architectures are also expected to benefit advanced packaging companies such as ASE.

Rise of Agentic AI, CPUs Regain Their Core Role

In recent years, AI infrastructure construction has mainly focused on GPUs, with CPUs taking on supporting roles such as data scheduling, storage management, and task coordination. As AI applications gradually shift from training to inference and further evolve toward Agentic AI, the importance of the CPU is rapidly increasing.

In Agentic AI scenarios, models need to plan tasks, call tools, access databases, manage states, and execute complex workflows. Such work does not rely entirely on matrix computation, but involves a large number of sequential executions, low-latency, and high I/O demand tasks, which are more suited to CPU processing.

The CPU is upgrading from the traditional "host processor" to the "control center" of AI systems. If the CPU fails to complete data scheduling, memory management, and task coordination in a timely manner, even if the GPU's performance is strong, it may result in decreased utilization.

Based on this trend, it is expected that the global server CPU market will enter a new growth cycle, and by 2030 the market size will exceed $170 billion, nearly a fivefold increase from about $35 billion in 2025.

Standalone AI Server Market Points to x86, AMD Seizes the Initiative

Against the backdrop of booming CPU demand, UBS is increasingly optimistic about AMD's long-term growth potential.

The institution believes that, standalone Agentic AI servers are rapidly emerging, and these applications place greater emphasis on core count and multithreaded performance—precisely AMD's strengths. According to the latest UBS calculations, in the standalone AI server market, the x86 and ARM architectures will account for 60% and 40% market share respectively in the future, with AMD expected to capture the majority of incremental x86 demand.

Based on higher market share expectations, UBS has raised its forecasts for AMD's server CPU revenue in 2027 from $21 billion to $23 billion, and in 2028 from $27 billion to $29 billion, and expects 2030 server CPU revenue to reach $50 billion, higher than the previous forecast of $41 billion.

Accordingly, UBS has sharply raised AMD's price target from $455 to $670.

Targeting 70% of Head Node Share, ARM Ecosystem Accelerates Expansion

Apart from AMD, ARM is also a key focus of UBS.

The report points out that future AI infrastructure will gradually form two types of CPU architectures: one is the Head Node CPU connected to GPUs, emphasizing low latency and high energy efficiency; the other is standalone Agentic AI servers, which emphasize core count and throughput capability.

UBS believes that ARM architecture has clear advantages in the former market. NVIDIA Grace, Vera, and Google Axion, AWS Graviton and other products are all rapidly expanding. The institution expects that by 2030, ARM architecture will account for about 70% of the Head Node CPU market and win close to 50% of overall server CPU revenue share.

Based on an optimistic outlook for AI CPUs, UBS has raised ARM's price target from $260 to $470.

TSMC and ASE Usher in a New Expansion Cycle

The surge in CPU demand is becoming a new growth pole for the semiconductor industry after GPUs. Bank of America points out that this trend not only benefits chip design companies, but will also directly transmit to manufacturing and packaging/testing segments.

According to Bank of America's calculations, the global server CPU-related semiconductor manufacturing market size is expected to expand from $15 billion in 2025 to $49 billion in 2028. The proportion of outsourced production will rise from 52% to 71%, reflecting the ever-strengthening core position of pure-play foundries such as TSMC in the high-end CPU arena.

The scarcity of advanced process capacity, combined with parallel ramp-up by multiple customers and architectures, makes the foundry segment the most certain beneficiary in this upcycle.

At the same time, AI CPUs' rising demand for chiplet heterogeneous integration, CoWoS advanced packaging, and system-level testing is driving rapid expansion of the backend packaging and testing market. Bank of America expects the server CPU-related packaging and testing market to grow from $1.9 billion in 2025 to $9.6 billion in 2028, with the advanced packaging market share rising from 11% to 24%.

Upgrades to the packaging value chain not only increase the packaging value per chip, but also lengthen the work time for backend processes like testing, creating a scenario where packaging/testing leaders benefit from both higher volumes and prices.

Based on the above supply-demand landscape, Bank of America has simultaneously raised its valuation expectations for supply chain leaders such as TSMC and ASE, believing that advanced manufacturing and advanced packaging remain the most formidable barriers in the value chain. In the new round of computing architecture changes driven by Agentic AI, the parallel evolution of CPUs and GPUs is increasingly apparent. Foundry and packaging/testing companies with technological leadership and scale advantages are expected to continue benefitting in this cycle from both a diversified customer base and increased per-chip value.

Risk Warning and DisclaimerThe market comes with risks; investment should be approached with caution. This article does not constitute personal investment advice, nor does it take into account specific investment objectives, financial situations, or needs of individual users. Users should consider whether any opinions, viewpoints, or conclusions in this article are suitable for their particular circumstances. Investing accordingly is at your own risk. ```