Bank of America Securities analyst Vivek Arya increased the price target for Advanced Micro Devices (AMD) to $720, up from $620. The new target implies a market capitalisation of roughly $1.2 trillion, a step up from the $1 trillion level AMD reached earlier in the week. AMD shares were trading at $632.99, up 0.6% in early trade and more than 13% for the week.
BofA’s upgrade rests on a projected surge in the server CPU market. The firm estimates the total addressable market (TAM) for server CPUs will triple from $61 billion in 2026 to $211 billion by 2030. Of that, $180 billion is expected to be driven by AI workloads, while on‑premises enterprise CPUs represent an incremental $80 billion opportunity.
The analyst also raised AMD’s CY2027 and CY2028 earnings per share forecasts by 2‑3% and applied a 30‑times multiple to the CY2028 earnings estimate, up from the previous 27‑times multiple. Unit volumes for AI‑focused CPUs are modelled to rise from 16 million in CY2026 to 53 million in CY2030, with blended server CPU average selling prices climbing from about $1,600 to $2,600 over the same period.
Industry commentary highlighted a shift toward ARM‑based server CPUs. Creative Strategies’ Ben Bajarin projected ARM’s share of server CPU units could approach 40% by the end of the decade, potentially reaching 50% if hyperscalers accelerate porting and NVIDIA expands ARM adoption. He also noted that high‑end AI CPU packages for NVIDIA’s Vera chip could command prices of $4,000‑$5,000 or more.
Arya emphasized that CPUs are expanding the AI system TAM rather than replacing accelerators, arguing that CPU latency and task‑completion speed are increasingly critical to overall GPU utilisation. He suggested a new framework for assessing CPU demand that focuses on the number of GPU‑initiated tasks a CPU can handle concurrently and the speed of result delivery, rather than a simple CPU‑to‑GPU ratio.
The analysis positions AMD to benefit from the broader AI‑driven compute expansion, while x86 architectures are expected to remain entrenched in legacy cloud and enterprise software environments.
