Beyond the simple “NVIDIA alternative” or “Intel alternative” narrative, AMD is building a story around the enterprise market, balancing, or perhaps, combining, its CPU story with its growing GPU story. Yesterday’s news that Dell PowerEdge XE7745 and R7725 servers will support AMD’s new Instinct MI350P PCIe GPUs highlighted an immediate, tangible deployment of what AMD has boasted is an enabler of production-level generative AI and agentic AI in the enterprise.

This dovetails with what AMD’s Suresh Andani said last week about “inference becoming foundational to enterprise computing,” with AMD’s air-cooled, dual-slot PCIe card designed to fit standard 19-inch enterprise server racks. While some hype that this means “zero-Capex redesign for enterprise data centers” might go too far, it isn’t a stretch to believe Andani’s contention that this is “an open, low, and no-cost software ecosystem” for infrastructure that enterprises “already own.”

Plugging a component-level card into standard servers may, indeed, help enterprises more cost-effectively deploy and scale generative and agentic AI, especially since most cannot afford the custom liquid-cooled systems hyperscalers and cloud giants can. With HPC at a lower cost, enterprises can more readily pivot from computation to inference and agentic AI.

As explained by Futurum Group analyst Brendan Burke, “AMD used Q1 FY 2026 to tighten the narrative around two connected growth sectors: server CPUs as AI infrastructure control points, and accelerators as inference deployments move from pilots to production.”

In its Q1 2026 reporting, AMD showed that data center revenue was up 57% YoY, and it has high expectations for Q2, based on this and other announcements, like yesterday’s with Dell. RCRTech will continue to follow the story.

The inference explosion: F5’s 2026 State of Application Strategy (SOAS) Report, shows that 77% of enterprises now prioritize inference over training, with the average organization managing or evaluating 7 different AI models, simultaneously.

ISAC physical AI: Factory-floor ‘physical AI’ runs on lightweight (non-GPU) edge compute, says NTT Data, but 5G sensing, in the form of ISAC, could unlock a new wave of demand for private 5G, tying edge IoT and AI into a single enterprise stack.

Europe’s race for sovereign AI: Vodafone, BT, Telenor and other European telcos are aggressively building sovereign AI and cloud infrastructure as a way to reduce reliance on U.S. hyperscalers, and to a lesser (practical) extent, on Chinese tech.

Water Sustainability: Gradiant claims a single 100MW hyperscale campus consumes as much water as a city of 80k people, but its closed-loop cooling system minimizes discharge through high-recovery concentration and reuse.