AMD shares surge to record high; Wells Fargo raises PT to $615, sees Venice CPU outshipping Nvidia Vera in 2027

AMD shares rose over 7% on June 30, reaching a record high of $579.73, bolstered by Wells Fargo raising its price target from $505 to $615—a 21% upward adjustment while maintaining an Overweight rating. The price target reflects analyst confidence in AMD's CPU and GPU expansion driving significant revenue growth through 2028. Notably, Wells Fargo predicts AMD's next-generation EPYC CPU platform, Venice (built on TSMC's 2nm process with up to 256 Zen 6 cores), will outship Nvidia's 3nm Vera CPU by unit volume in 2027, citing superior core density and architectural efficiency.

GPU revenues are projected to accelerate from $15.6 billion in 2026 to $40.6 billion in 2027 and $63 billion by 2028, representing a massive expansion in AMD's data center AI accelerator business. For CPUs, Wells Fargo models 68% revenue growth in 2026 over 2025, followed by 28% growth in 2027 and 22% more in 2028, bringing CPU revenue to approximately $25 billion. The rise also signals a shift in analyst sentiment: analysts now see AMD gaining meaningful share in the transition from AI training (building LLMs) to AI inference (running queries), where the competitive dynamics differ from training.

Analyst Rakers sees AMD earning about 3% more than consensus estimates for 2027 ($13.40 per share) and 8% more for 2028 ($18.75 per share), with potential to reach $20 per share in the near term. Strong demand for CPUs drives the thesis, with computer chip prices expected to remain elevated. AMD's ROCm software ecosystem support doubled in 2025 and downloads grew tenfold year-over-year, reducing friction for developers moving beyond CUDA-exclusive deployments.

For practitioners: AMD's gains reflect conviction that inference workloads will diversify away from Nvidia's dominance, and that heterogeneous architectures (Venice's 256 cores) will prove superior for certain HPC and AI workloads. Teams evaluating multi-year capex for inference infrastructure should monitor Venice's actual shipments and customer validation in 2027. AMD's improving software narrative (ROCm ecosystem depth) and architectural positioning on density/efficiency merit serious evaluation alongside Nvidia, especially for workloads where cost-per-inference or power-per-token matters.