UBS said in a note on Wednesday that it has raised price targets on two semiconductor companies it views as key beneficiaries of accelerating demand for standalone central processing units in agentic AI infrastructure.
Analyst Timothy Arcuri lifted the bank's price target for AMD to $670 from $455 and increased Arm's target to $470 from $260. UBS applied an unchanged 35x earnings multiple for AMD and used a 108x multiple for Arm in deriving the new targets.
Market-size caution
UBS warned that investor enthusiasm around the potential size of the CPU market has driven some forecasts toward $300 billion by 2030. The bank said it favors restraint on such large estimates, noting that DRAM supply will remain a constraint and therefore believes those very bullish outcomes should be considered long-term scenarios for memory suppliers.
Thesis on AMD
On AMD, Arcuri wrote that the company possesses "a distinct advantage" in standalone agentic AI racks, which he described as workloads that are more core- and throughput-driven. UBS now assumes the standalone segment will be split roughly 60/40 between x86 and Arm architectures, and it expects AMD to take an outsized share of incremental demand given what the bank characterizes as Intel's roadmap and supply challenges.
UBS's forecast for AMD's CPU server revenues projects $23 billion in 2027 and $29 billion in 2028, expanding to $50 billion by 2030.
Outlook for Arm
For Arm, UBS maintained its view that Arm-based architectures will account for approximately 70% of a head-node market that the bank sizes at roughly 20 million units by 2030. Arcuri raised his estimate for Arm's internal CPU-related revenue to about $14 billion in 2030, which reflects roughly an 8% share of UBS's $170 billion total addressable market estimate.
However, UBS noted that current core count limitations on Arm designs will constrain the architecture's standalone rack prospects until future product generations can address those limitations.
Implications
The research note frames both AMD and Arm as positioned to benefit from a shift toward standalone CPU deployments in AI infrastructure, while also reminding investors that broader market outcomes hinge on factors such as memory supply dynamics and architectural performance improvements.