AMD Shares Gain Pre-Market With Wall St Increasingly Bullish: The Latest Calls

Advanced Micro Devices shares (NASDAQ:AMD) are 2% higher in pre-market trading, as markets continue to pick through a wave of analyst upgrades clustered around the $500 level. After a rally that has seen AMD shares add 260% over the past year, the Street points to further potential gains, driven by artificial intelligence infrastructure demand.

AMD’s share price currently sits at $420 in pre-market hours, extending gains that have seen the shares more than triple from levels around $113 just twelve months ago. The current price sits comfortably within a newly established analyst consensus, with the average target now reaching $472.17, a high-end forecast of $625, and a low target of $225.

The recent raft of upgrades this month has seen more than ten firms raise their targets, many by 50% or more. The concentration of targets in the $500-$580 range suggests Wall Street has found a new valuation framework that incorporates both GPU scarcity dynamics and an emerging “CPU renaissance” tied to agentic AI workloads.

Fundamentals Driving AMD Shares Higher

The upgrade cycle accelerated after AMD’s first-quarter 2026 earnings release on May 5, which showed server compute revenues jumping approximately 50% year-over-year and management guiding to roughly 70% server CPU growth in the second quarter. Those figures provided tangible evidence that AI-driven demand is flowing through the profit-and-loss statement, not merely appearing in long-term TAM presentations.

The most significant shift came from Citi this week, raising AMD’s price target to $460 from $358, albeit maintaining a Neutral rating. Citi’s note introduced a new server CPU total addressable market model projecting the segment will grow at a 35% compound annual rate from $29.3 billion in 2025 to approximately $132 billion by 2030. Within that forecast, Citi carved out a new category: agentic CPUs, expected to grow at a staggering 185% CAGR to reach $59.4 billion and represent 45% of the total CPU market by decade’s end. General-purpose CPUs are forecast to expand at 20% annually to $50.9 billion, while AI head-node CPUs grow 21% to $21.1 billion.

Citi’s TAM framework assigns Intel roughly 47% market share by 2030, AMD about 34%, and ARM-based and other architectures the remaining 19%. While the model still favours Intel in absolute terms, it validates AMD’s trajectory toward a $45 billion annual CPU revenue opportunity within four years, a figure that underpins the elevated price targets now dominating sell-side research.

Price Targets

Evercore ISI delivered one of the most bullish calls following earnings, lifting its target to $579 from $358 and maintaining an Outperform rating. The firm’s methodology is clear, with Evercore valuing AMD at 25 times 2028 earnings per share of $28, then discounting that figure back to present value. The approach explicitly ties AMD’s valuation to a long-dated AI compute thesis, emphasizing the shift from training-intensive workloads to inference-led deployments by late 2026. Evercore argues that inference economics focused on cost-per-token, return on investment, and total cost of ownership, will push hyperscalers to diversify beyond Nvidia toward alternative accelerators and more capable CPUs, a dynamic that benefits AMD’s product portfolio.

Goldman Sachs upgraded AMD to Buy from Neutral on the same day, raising its target to $450 from $240. The bank cited agentic AI tailwinds and stronger data center GPU growth, positioning AMD as a likely winner as enterprise and consumer workloads become more AI-intensive. Bernstein also upgraded AMD to Outperform from Market Perform, increasing its target to $525 from $265 and noting that the company’s story is becoming “increasingly fundamental rather than momentum-driven,” a significant shift from a house historically cautious on valuation and cycle risk.

Bank of America Securities lifted its target to $500 from $450 last week, highlighting a larger AI data center opportunity and a bigger server CPU market that AMD can capture materially over time.

KeyBanc moved to $530 from $330, pointing to the expected 70% year-over-year server CPU growth in the second quarter as evidence of near-term momentum.

Barclays raised its target to $500 from $300, calling AMD one of the more compelling AI plays and citing XPU compute shortages and growing CPU relevance in AI infrastructure.

The highest target in the current cohort came from China Renaissance yesterday, which nearly doubled its price objective to $562 from $291 while reiterating a Buy rating.

Arete followed with a $540 target, up from $325, emphasizing confidence in AMD’s AI accelerator roadmap and upside to server CPU and GPU mix over multiple years.

The Agentic AI Thesis

A recurring theme across the upgrade notes is the emergence of agentic AI workloads,  autonomous software agents that perform complex, multi-step tasks requiring sustained compute across both CPUs and accelerators. Unlike traditional AI training, which is GPU-intensive and bursty, agentic inference demands high CPU throughput for orchestration, memory bandwidth for context management, and cost-efficient accelerators for token generation.

This architecture shift is forcing hyperscalers to rethink their data center designs, moving away from Nvidia-dominated GPU farms toward heterogeneous compute clusters that blend CPUs, GPUs, and custom ASICs.

AMD is seen as well positioned to benefit from this transition on multiple fronts. Its EPYC server CPU family has gained share from Intel over the past three years, reaching an estimated 25–30% of the x86 server market. The company’s MI300 series GPU accelerators, while trailing Nvidia’s H100 and H200 in raw performance, offer competitive pricing and are increasingly adopted by hyperscalers seeking supply diversification and cost optimization.

The combination of strong CPU share and credible GPU alternatives gives AMD exposure to both the “head node” CPU opportunity and the accelerator market, a dual revenue stream that few competitors can match.

What Lay Ahead

The reliance on long-dated earnings frameworks introduces some risk in price targets, and expectations. Evercore’s $579 target, for example, depends on AMD achieving $28 in earnings per share by 2028, implying a roughly 40% compound annual EPS growth rate from current levels. Any slippage in AI revenue ramps, margin progression, or market share gains could trigger significant de-rating, particularly given the premium multiples now embedded in sell-side models.

Competitive dynamics also warrant scrutiny. Citi’s TAM model still assigns Intel the largest share of the 2030 CPU market, reflecting the incumbent’s installed base, software ecosystem advantages, and ongoing manufacturing investments. Hyperscaler custom ASICs from Amazon, Google, and Microsoft represent another structural headwind, as these chips are optimized for specific workloads and bypass merchant silicon suppliers entirely. ARM-based server CPUs, led by Ampere and emerging players like Nvidia’s Grace, add further fragmentation risk.

AMD shares’ move to $420 and the clustering of analyst targets around $500 reflect a fundamental shift of the company’s role in AI infrastructure. The convergence of bullish views from previously cautious houses like Bernstein and Goldman Sachs lends credibility to the thesis that AMD’s earnings power has reached a scale that justifies premium valuations. However, the magnitude of recent price-target increases also raises classic late-cycle sentiment risks, where expectations become stretched and any execution miss or macro wobble could trigger sharp corrections.