On October 19th, TD Cowen reiterated its "buy" rating for Supermicro Semiconductor ( AMD ) and set a target price of $675, stating that the rapid growth of autonomous AI intelligent agents ( Agentic AI ) and AI inference workloads in production environments is driving a recovery in demand for data centers ( CPU ). The firm noted that as artificial intelligence applications shift from model training to actual deployment, the importance of CPU in multi-step task coordination, tool invocation, API management, and continuous decision-making continues to increase. AMD has previously raised its forecast for the potential market size of CPU to $220 billion and expects server CPU business revenue to grow by 70% year-on-year by 2027.
The AI intelligent entity drives the demand for CPU, and AMD expands the potential market size to 220 billion US dollars.
TD Cowen believes that as AI intelligents start to handle complex business processes, CPU is becoming an important beneficiary of AI infrastructure investments. Unlike GPU, which is mainly responsible for large-scale parallel computing, CPU is better suited for handling multi-step workflow coordination, external tool invocation, API request management, and decision-making tasks that need to be executed in sequence. This means that the more AI applications are integrated into the actual production environment of enterprises, the greater the demand for general-purpose computing resources may also be.
Earlier this year, AMD raised its forecast for the total potential market size ( TAM ) of its CPU business by nearly half to $220 billion, and expects server CPU business revenue to grow by 70% year-on-year in 2027. TD Cowen pointed out that the proliferation of AI intelligents and production-grade inference applications is creating clearer commercial demands, providing new growth momentum for the server CPU market.
Due to the continued strong demand for data center CPU, TD Cowen has also slightly raised its performance forecasts for AMD and Intel. AMD's revenue has increased by 40% year-on-year over the past 12 months, and market analysts expect its revenue to grow by 47% in the fiscal year 2026, reflecting that the company's server processors and AI related businesses are still in an expansion phase.
Wall Street is optimistic about the server business of AMD; Citibank raises its target price to $800
Recently, several Wall Street institutions have upgraded or maintained their positive ratings for AMD. The main reasons include the growth of AI intelligent applications, the expanding demand for server processors, and the commercialization prospects of the new generation of AI computing systems.
Citi has raised its target price for AMD to $800, believing that the continuous growth in demand for AI intelligent applications will drive the company's business expansion. Stifel has increased its target price to $700 and maintained a "buy" rating, mainly due to the strong performance of its server business. Cantor Fitzgerald has reiterated its "overweight" rating and $700 target price, focusing on the potential contribution of AMD Helios system shipments to future performance.
TD Cowen This time, the target price is maintained at $675. Based on the reported stock price of $546.68, there is a potential upside of about 23.5%. Although AMD has fallen by approximately 11% in the past week, the stock price has still risen by 147% since 2026, indicating that the market maintains high expectations for its long-term AI business growth.
AMD plans to significantly increase chip supply in 2027 to meet the computing demands of AI.
AMD CEO Su Zifeng recently confirmed that the company plans to significantly expand its chip supply in 2027 to meet the growing demand for artificial intelligence computing. As cloud computing companies and large institutions accelerate the construction of AI infrastructure, AMD is also expanding its server CPU and AI chip business, aiming to gain a larger share in both general-purpose computing and AI dedicated computing markets.
The core conclusion of this research report is that infrastructure demands are shifting from training clusters to a broader range of computing tasks. As intelligent entities gradually enter actual production environments, enterprises will not only need more powerful chips but also require additional systems for scheduling complex tasks and system management. This change is expected to create new growth opportunities for server businesses and support the industry's positive investment outlook.
