The CPU chip trade is one of the hottest on Wall Street right now, and UBS is doubling down on it. On Wednesday, analyst Timothy Arcuri raised his 12-month price target on Advanced Micro Devices (AMD) to $670 from $470, implying 29% upside from Tuesday’s close. He simultaneously raised his ARM Holdings (ARM) target to $455 from $260 – a 28% premium to where shares last traded. Both stocks carry Buy ratings at UBS. The catalyst? A fundamental shift in how AI companies are building their infrastructure, driven by the rise of agentic AI systems that require central processing units (CPUs) alongside – and increasingly instead of – traditional graphics processing units (GPUs).
The numbers behind this trade are staggering. AMD stock has surged 153% in just the past three months. ARM has done even better, up 171% in the same span. These are not speculative rallies – they are backed by real demand. Hyperscalers including Microsoft, Google, Amazon, and Meta are rapidly deploying standalone CPU racks to power the next wave of AI agents. AMD benefits from its strong core-count advantage and the vast x86 software ecosystem, making it a natural fit for agentic workloads that blend traditional software with AI inference. ARM benefits from its efficiency-first architecture, which aligns directly with hyperscalers’ need to minimize latency at scale.
Wall Street consensus backs this view strongly. Of 53 analysts covering AMD, 45 have issued a Buy or Strong Buy rating. ARM counts 25 Buy or Strong Buy ratings from the 40 analysts who cover it. For investors, the key question is whether this momentum has legs after triple-digit gains. UBS believes it does – and the data on agentic AI adoption and hyperscaler capex spending supports that thesis. Investors already holding these names should consider staying in; those looking to enter may find better entry points on any AI-sector pullback. AMD trades at a more reasonable valuation than most AI peers, and ARM’s royalty-based model gives it durable earnings power as CPU adoption accelerates globally.
