AMD Stock Is Up 320% Over the Past Year. Should Investors Shift Their Attention Away From Nvidia?

But the large disparity betwixt AMD's and Nvidia's gains surely has galore investors wondering whether they should displacement their attraction distant from Nvidia and toward AMD arsenic it rises. If you're looking to diversify your AI hardware investments, past buying AMD could beryllium a bully idea. But wholly ditching Nvidia astir apt isn't the champion move.

AMD is surging higher acknowledgment to agentic AI and inference

There person been respective cycles of the AI roar already, and astatine slightest for AMD, the 2 aspects of the exertion driving the company's income of cardinal processing units (CPUs) close present are inference (training AI) and agentic AI. Here's what CEO Lisa Su said connected the first-quarter net telephone astir these opportunities: "As inferencing and Agentic AI deployment standard [up], they are fundamentally expanding compute requirements, driving some larger-scale accelerator deployments and importantly much CPU compute. AMD is uniquely positioned to pb successful this adjacent signifier of AI with enactment products crossed high-performance work CPUs and AI accelerators."

That's benignant of a mouthful, but fundamentally AMD believes it's successful a large presumption to connection unsocial hardware for inference and AI agents -- and customers are already lining up. Total gross successful the archetypal 4th roseate 38% to $10.3 billion, and its information halfway income jumped 57% to $5.8 billion.

All of this caller maturation spurred absorption to revise its server CPU full addressable market, and it present estimates it beryllium $120 cardinal by 2030 -- double its erstwhile estimates. What's more, the institution raised its second-quarter gross guidance to $11.2 billion, 46% higher than its erstwhile estimate.

The vantage for AMD is that the institution offers some CPUs and graphics processing units (GPUs), and the premix of AI computing needs is shifting. For example, the institution says the ratio of CPUs to GPUs for agentic AI and inference utilized to beryllium 1 CPU to 4 GPUs (or adjacent 1 to 8). Now, it says the ratio is person to 1 to 1.