Citi upgrades AMD to Buy, citing Meta-driven GPU upside and rapid growth in agentic AI server CPUs

Citi upgraded Advanced Micro Devices (AMD) to a “Buy” rating and raised its price target to $575, according to a report carried by Barchart and syndicated by Yahoo Finance. The move centers on Citi’s view that AMD has meaningful upside from accelerated GPU deployments connected to Meta and from growth in server central processing units (CPUs) designed for agentic artificial intelligence workloads.

In the bank’s framing, the near-term catalyst is Meta’s expanding use of GPUs for AI. Citi pointed to what it described as a large, likely demand pool for AI compute equipment, with the implication that AMD could benefit as customers build out or scale their AI training and inference capacity. The report does not specify the number of systems, timelines, or whether Meta deployments involve specific AMD GPU product lines.

Citi also highlighted a broader demand theme: agentic AI server CPUs. Agentic AI refers to systems that can take actions, follow plans, and carry out tasks with a degree of autonomy rather than only returning a static answer. In the report, Citi described an agentic AI server CPU market it valued at more than $137 billion, positioning AMD to gain share as data centers upgrade hardware for these workloads.

The $575 price target and the “Buy” rating were presented as reflecting both expected revenue opportunity and improving confidence in AMD’s ability to participate in AI infrastructure spending. However, the report does not provide additional detail on which financial model inputs changed, such as adjustments to earnings estimates, gross margin assumptions, or specific data-center customer wins.

From a company perspective, AMD has been working across both accelerator hardware (GPUs) and data-center CPUs, which are used to run workloads that may be paired with GPU acceleration. That combination matters because AI systems often require both general-purpose compute and highly parallel processing. AMD’s opportunity, as Citi portrays it, is tied to the expansion of data-center capacity for new classes of AI applications, including agentic AI.

Citi’s emphasis on Meta is also notable because large cloud and technology companies can influence the rest of the market through their buying patterns and deployment cadence. Still, the report stops short of detailing any new Meta contract, purchase order, or update to AMD’s disclosed customer pipeline.

As with many bank-initiated rating changes reported through market syndication, the key limitation is disclosure. The company itself did not accompany the upgrade with new guidance in the information provided here, and the report does not break down the assumptions behind the $575 target or identify specific product SKUs and contractual terms.

Looking ahead, investors and analysts will likely focus on whether AMD’s results and forward commentary substantiate the “GPU sales potential” thesis and whether AMD continues to show momentum in data-center CPUs for AI-driven, agentic workloads. Any subsequent supply announcements, major customer design wins, or detailed financial updates would be central to validating the bank’s narrative.