Nvidia Opens Its Racks to a Rival Chipmaker

Nvidia has invited a rival chipmaker into its own racks, according to a 247wallst report that sets out why the company would make such a move.

Rack-scale systems are the delivery unit for modern AI data centers, and Nvidia's own racks have long been tightly coupled to its accelerators, giving customers a single-vendor path from chip to cabinet.

Putting a competitor's silicon inside those racks opens a seam in that model. For Nvidia, such arrangements typically answer customer demand for supply-chain diversity: large cloud providers want to mix accelerators from different sources within one rack architecture to reduce cost and vendor lock-in risk.

It is also a response to competitive pressure. AMD and the custom silicon programs run by major cloud providers are competing for the same AI compute budgets, and keeping a rival chip inside Nvidia's rack keeps that customer inside Nvidia's platform.

The 247wallst report does not name the chipmaker involved, nor does it offer shipment timing or volumes. The practical effect of the change therefore still depends on product-level confirmation and real customer deployments.

What to watch is whether the opening extends to interconnect standards and software stacks. If rival chips can run inside Nvidia racks, competition in AI servers shifts further toward whole-cabinet systems and ecosystems rather than raw accelerator performance.