NVIDIA Corp (NVDA.US) announced on Wednesday, July 1, that it has entered into partnerships with several artificial intelligence (AI) cloud service providers. These collaborations will deploy "large-scale, multi-tenant AI factories" through a new model involving revenue sharing and credit support.
Under this new framework, AI cloud providers can offer external cloud services based on NVIDIA's DSX AI factory infrastructure. NVIDIA stated that this partnership model "accelerates the adoption of the NVIDIA platform across the customer base, provides cloud providers with a capital-light path to expansion, and creates a new, usage-based recurring revenue stream for NVIDIA."
SharonAI Holdings (SHAZ.US) and Firmus are the first companies to participate in this program. SharonAI plans to deploy up to 40,000 NVIDIA Grace Blackwell GB300 GPUs.
NVIDIA's Chief Financial Officer, Colette Kress, noted that with the company's DSX AI factory, startups aiming to develop AI can access cloud computing resources from providers like Firmus Technologies faster and more flexibly than before.
She wrote in a blog post, "For model builders, inference service providers, agent platforms, and enterprises scaling AI, this solution enables rapid access to full-stack accelerated computing power, bypassing the lengthy processes of site selection, power procurement, campus construction, and hardware deployment."
An AI factory is a new type of infrastructure centered on artificial intelligence technology, designed for the mass "production of intelligence." NVIDIA CEO Jensen Huang first introduced the "AI factory" concept back in 2022. In 2026, he elaborated further, stating that AI data centers are essentially "factories that produce tokens," which can ultimately be transformed into code, answers, designs, actions, and various services, directly generating revenue.
The launch of this new AI infrastructure partnership model comes as market concerns about an "AI computing power surplus" intensify. Previously, reports that Meta Platforms Inc (META.US) was considering selling computing capacity sparked investor fears of an AI compute glut, leading to a broad sell-off in AI hardware sectors including memory chips, semiconductor equipment, and optical communications.
The Philadelphia Semiconductor Index (SOX) plunged over 6% in a single day, with Micron Technology Inc (MU.US) and SanDisk (SNDK.US) falling sharply on Wednesday. The ripple effects quickly spread globally, with South Korea's KOSPI index tumbling and triggering circuit breakers on Thursday, dragging down shares of Samsung Electronics and SK Hynix.
Rich Privorotsky, Head of the 1-Delta Trading Desk at Goldman Sachs, pinpointed the issue: "The market's core premise has always been compute scarcity—if supply increases and rental prices fall, that narrative will be overturned."
Analysts at Wedbush suggested that the recent weakness in large-cap tech stocks is primarily due to growing market concerns about the payback period for record-breaking AI infrastructure investments, rather than a deterioration in the industry's long-term growth logic.
The upcoming second-quarter earnings reports from hyperscale cloud providers will serve as a critical test. If multiple companies signal a reduction in capital expenditures, the AI hardware sector could face a deeper correction.
Comments