What it is
NVIDIA Cloud Partner (NCP) is a certification and partnership tier within the NVIDIA Partner Network (NPN), NVIDIA’s broader structure for companies that build businesses around its hardware and software. The NCP designation is specifically for cloud and data-centre operators — not resellers or system integrators, but companies that stand up and run physical AI infrastructure at scale. To qualify, a provider must build to NVIDIA’s reference architecture: a published blueprint for how to design data centres capable of handling generative AI training and inference, covering compute, networking, storage and software layers so that NVIDIA GPUs and systems work predictably across different operators’ facilities.
In practice, becoming an NVIDIA Cloud Partner means a company has committed to deploying NVIDIA’s accelerated computing hardware — GPUs such as the Blackwell Ultra or, in future roadmaps, Vera Rubin systems — at production scale, and has been validated by NVIDIA as capable of operating that hardware to the vendor’s operational standards. NVIDIA describes the NCP network as having cumulatively deployed more than 1 million GPUs worldwide, representing over 1.7 gigawatts of AI computing capacity, roughly double the prior year’s footprint. Members range from large, established cloud and colocation firms to newer entrants building what NVIDIA calls “AI factories” — facilities whose output is not a traditional cloud service but continuous AI model training and inference capacity.
The program functions as both a sales channel and a quality signal. NVIDIA gains predictable, large-volume outlets for its highest-margin hardware and a network of operators who can be pointed to as proof that its chips are being deployed credibly around the world. Partners gain early access to reference designs, NVIDIA’s technical support, and — critically for newer or regional players — the marketing value of NVIDIA’s name attached to their infrastructure.
Why it matters for AI governance and narratives
The NCP program is a clear example of how AI infrastructure build-out is a sourced claim, not a neutral fact. When a company announces it has become an NVIDIA Cloud Partner, the announcement originates from NVIDIA’s commercial ecosystem — NVIDIA itself, its partner, or local government and investment-promotion bodies — not from independent verification of capacity, timelines, or actual compute delivered. Designations like “first NVIDIA Cloud Partner in [country]” function as both infrastructure fact and geopolitical signalling: they let national governments claim entry into frontier AI compute capacity, let regional operators attract capital and policy attention, and let NVIDIA demonstrate the global reach of its hardware ecosystem, all through a single press announcement.
For an observatory tracking how AI capability and governance narratives are framed, NCP designations are a recurring device in a specific genre: the data-centre investment announcement. These items typically bundle a hardware commitment, a national or regional “first,” and often an energy or workforce development angle (as with Fermaca’s accompanying university AI-training initiative), and they rarely specify which claims are binding commitments versus aspirational roadmaps. Readers encountering an NCP designation in coverage should treat it as evidence of a commercial and promotional relationship with NVIDIA — real, but not independently verified as to scale, timeline, or completion.
Key facts and dates
The NVIDIA Cloud Partner program operates as a specialization within the long-standing NVIDIA Partner Network. NVIDIA’s own materials describe NCP growth in year-over-year terms tied to its GTC developer conference cycle: at GTC 2026, NVIDIA stated that cloud partners had doubled their AI factory footprint over the prior year, growing from roughly 400,000 GPUs and 550 megawatts of capacity to more than 1 million GPUs and 1.7 gigawatts. Known NCP members span large, established operators (Rackspace) and smaller or specialized entrants (Groq, Atlantic.Net, SWI Group).
In the Latin American case referenced in the editorial, Mexican energy and infrastructure company Fermaca was announced on November 4, 2026, as the first NVIDIA Cloud Partner based in Mexico, unveiling a phased “AI factory platform” beginning with a facility called Fermaca Digital Origins in Mexico City, built around NVIDIA Blackwell Ultra systems with a stated future roadmap toward NVIDIA’s Vera Rubin architecture. The announcement included an associated educational initiative, “México Aprende IA con Fermaca y NVIDIA,” aimed at university instructors. As with similar announcements, the Fermaca designation is best treated as a commercial and promotional milestone — confirmed by NVIDIA and the company’s own statements — rather than independently verified operational capacity.
Where to learn more
- NVIDIA Cloud Partners Deliver Scalable AI Solutions — NVIDIA’s own description of the NCP program and reference architecture
- Fermaca Unveils NVIDIA-Powered AI Factory Platform to Power Mexico and Latin America — primary press release on the Fermaca designation
- Fermaca Becomes First NVIDIA Cloud Partner in Mexico — independent trade-press coverage of the designation
- NVIDIA GTC 2026: Live Updates on What’s Next in AI — NVIDIA’s figures on cumulative NCP GPU deployment and capacity growth