AMD AI systems strategy gets theCUBE spotlight as IBM shifts enterprise AI

AMD AI systems strategy, rather than another GPU benchmark cycle, was the central takeaway from the latest episode of theCUBE Pod, where theCUBE Research analysts framed Advanced Micro Devices Inc. as a credible second supplier in AI infrastructure behind Nvidia Corp. The discussion also argued that International Business Machines Corp. is being misread by investors as it positions around enterprise AI operations rather than frontier model development.

John Furrier, executive analyst at theCUBE Research, said after AMD’s Advancing AI event that the company’s pitch should be read as a systems move, not only a chip roadmap. His view is that enterprise AI competition is shifting toward full infrastructure stacks that combine CPUs, GPUs and software agents into systems companies can use to run AI workloads.

That is a more useful frame for operators than the familiar question of which vendor has the fastest accelerator. Nvidia still leads the AI infrastructure market, according to the analysts, but the conversation around AMD is now about whether it can supply enough of the stack to become a durable alternative for enterprise buyers and cloud providers.

What is AMD's AI systems strategy?

In theCUBE Research framing, AMD’s AI systems strategy is the move from selling components toward rack-scale AI infrastructure that joins silicon and software into a working platform. Furrier described the future enterprise computing model as one organized around goals, projects and conversations, with software designed to interpret user intent rather than requiring users to adapt to applications.

The analysts tied that shift to the idea of the AI factory, a purpose-built infrastructure system for producing and running AI. In practice, that means the value is not only in GPUs or capital equipment, but in the applications and software layers that sit on top of the AI infrastructure.

Dave Vellante, chief analyst at theCUBE Research, said the growth of inference workloads could create a larger opening for AMD. He argued that AMD has more parts of the technology stack than large language model vendors, although the discussion did not provide new revenue figures, market-share data or customer commitments tied to that claim.

AMD as the second AI infrastructure supplier

Vellante said AMD does not need to displace Nvidia to become relevant. His argument is more practical: enterprise and cloud customers want a second source for AI infrastructure, both for supply optionality and negotiating leverage.

Furrier said the contest between AMD and Nvidia is expanding from chips to the architecture of the next computing cycle. Both companies are building rack-scale systems that combine software and silicon, according to the discussion. Nvidia remains ahead, but the analysts described AMD as positioned to become the second-leading company in the AI race if it can turn that systems pitch into enterprise adoption.

Why did IBM come up in the AI systems discussion?

The same episode also turned to IBM, where Furrier said Wall Street was focused on the wrong signals. He argued that the AI market is splitting between companies building frontier models and companies helping enterprises put AI into production across secure environments and older systems.

Furrier placed IBM in the second group, describing its strategy as an effort to become an operating layer for AI-driven enterprises. That is consistent with IBM’s long-standing base in large organizations, though the podcast segment did not cite new IBM customer numbers or product revenue to support the scale of the opportunity.

For technology buyers, the through line is that AI infrastructure is becoming a systems procurement decision. Nvidia owns the default position, AMD is trying to make itself the safe second option, and IBM is arguing through its enterprise posture that much of the AI budget will go to deployment, governance and integration rather than model training alone.

This story draws on original reporting from SiliconANGLE.