From "Stockpiling Cards" to "Efficiency Is King": Dell Technologies, Inc. Class C (DELL.US) Expands Its "AI Factory" the Computing Power Business of New Cloud Providers Is About to Change

Notice that Dell Technologies, Inc. Class C (DELL.US) shares surged nearly 6% in premarket trading on Wednesday after the company announced the launch of a full suite of AI Factory solutions with NVIDIA Corporation (NVDA.US), covering AI solutions and ecosystem innovation, a new desktop agent solution, and dedicated AI infrastructure. Dell Technologies, Inc. Class C said that more than 6,500 enterprises have adopted the Dell Technologies, Inc. Class C AI Factory built in partnership with NVIDIA Corporation, up from 5,000 in the previous quarter. New additions include support for Reflection's Beam model, AI observability in partnership with Dynatrace, and Mistral, Fogsphere, and SUSE Linux, among others. Strategically, Dell Technologies, Inc. Class C is betting on "agentic AI": institutions forecast that agentic AI will become the largest workload by 2028, with inference costs falling 90% by 2030 while token consumption rises 24-fold. Dell Technologies, Inc. Class C is seizing this wave of computing power dividends with "owned infrastructure + end-to-end managed services," competing head-on with the pure public cloud rental route. The Computing Power Business of New Cloud Providers Is About to Change Dell Technologies, Inc. Class C and NVIDIA Corporation are joining forces to expand their "AI Factory" combination punch, and it is rewriting the competitive landscape for enterprise-grade AI computing powerthose feeling the most direct pressure are the new cloud providers that built their businesses over the past two years by "renting out GPUs." The core narrative Dell Technologies, Inc. Class C is putting forward is that "owning is more cost-effective than renting": owned AI infrastructure can be 90% more cost-efficient than public cloud within two years, and with the DSX POD containerized AI factory compressing deployment cycles from six months to 14 days, the economics of enterprises "building their own computing power" have been completely reshaped. Previously, the core selling points of new cloud providers were eliminating heavy assets, on-demand elasticity, and rapid scaling; now Dell Technologies, Inc. Class C, with factory prefabrication + end-to-end managed services, has lowered both the threshold and the payment cycle for "self-building," effectively taking away part of the rental demand that would otherwise have flowed to new cloud providers. But this is not one-sidedly negative. DSX POD is essentially NVIDIA's reference design, and new cloud providers such as CoreWeave (CRWV.US), Nebius (NBIS.US), Lambda, Crusoe, and IREN (IREN.US) were already its early usersthey too can use this modular solution to reduce their own expansion costs and shorten delivery cycles. In other words, the "manufacturing" and "operation" of computing power are accelerating their separation, and the core competitiveness of new cloud providers is shifting from "how many GPUs you have on hand" to "operational efficiency, depth of the software stack, and customer acquisition capability." The real point of differentiation lies in software and ecosystem. Dell Technologies, Inc. Class C has woven an end-to-end network with open models, observability, knowledge assistants, and security governance. Those small and medium-sized new cloud providers that only rent out bare-metal computing power and lack upper-layer software capabilities will face being squeezed in the middleunable to match hyperscale clouds above, and unable to match Dell Technologies, Inc. Class C's "full stack + owned" below. The industry logic has moved from "computing power scarcity" to "efficiency is king." When inference costs are expected to fall 90% by 2030 while token consumption rises 24-fold, what determines victory is no longer stockpiling cards, but who can deliver ROI at a lower cost per token. Dell Technologies, Inc. Class C is betting on "enterprise-owned + agentic AI," while new cloud providers need to prove they are irreplaceable in specialized chips, elastic scale, and data gravity. A reshuffle over "who controls the last mile of enterprise AI deployment" has only just begun.