Nvidia officially denied reports that it paused its AI cloud commitments initiative, despite allegations that it restricted GPU leasing to approved partners only.
Nvidia denied a Wall Street Journal report claiming the company had placed certain transactions under its newly launched 'take or pay' AI Compute Partnership initiative on hold. The denial came less than two months after the program's early July unveiling and days before Nvidia detailed the effort during its earnings call. The reported pause allegedly stemmed from partner frustration over Nvidia's attempts to influence their operational decisions and internal concerns regarding potential antitrust scrutiny.
"The new business model we introduced in July that opens up compute access to the fast-growing AI ecosystem is still in place and continues to evolve due to high demand," a Nvidia spokesperson told Tom’s Hardware. While the WSJ report did not claim Nvidia abandoned the program—which involves committing to rent capacity and guarantee minimum revenue for newly built AI data centers—it indicated that specific deals were temporarily halted. Nvidia’s statement confirms the initiative remains active but is undergoing structural adjustments.
According to the report, Nvidia attempted to dictate how 'AI Compute Partners' leased their capacity. Cloud providers participating in the program were reportedly told they could only lease Nvidia’s GPUs to customers approved by the company. Additionally, Nvidia preferred distributing available capacity across multiple smaller AI firms rather than allowing a single large customer to secure most or all of it. Some cloud operators pushed back against these restrictions, asserting their right to control client selection, which may have prompted Nvidia to pause certain agreements.
Although Nvidia does not lend money or directly finance AI data center construction—effectively avoiding circular financing—it provides demand commitments and guaranteed revenue levels. This structure likely triggered internal antitrust concerns, leading Nvidia to potentially revise deal terms with partners. The initiative operates alongside Nvidia’s broader financial expansion, including reported revenue topping $96 billion, memory commitments soaring to $160 billion, partnerships with financial giants to create $500 billion in AI infrastructure funds, and the current $36 billion in partnership commitments.
Modern AI data centers require billions in upfront capital for premises, infrastructure, and compute hardware long before operators secure sufficient customer contracts. Lenders and investors require assurance that future capacity will be rented. Under this program, Nvidia leverages its own demand commitment for a portion of a facility’s capacity in exchange for a percentage of the facility’s revenue if demand exceeds expectations. This de-risks financing for lenders by backing a portion of the revenue stream with Nvidia rather than relying solely on the operator’s customer acquisition.
During the earnings call, Nvidia Chief Financial Officer Colette Kress explained: "Nvidia provides a take-or-pay commitment on a portion of the facility's capacity, a minimum revenue guarantee that gives lenders the confidence to underwrite the project, and in exchange, we share in a portion of the NeoCloud's revenue earned above that floor." She noted the dual-revenue model: "In this model, we get paid twice, once on the hardware sale, and again through the share of rental revenue, a highly recurring stream layered on top of a one-time equipment purchase."
While exact percentages and economic terms remain undisclosed, the structure operates straightforwardly. For instance, Nvidia might commit to renting 30% of a new facility’s capacity and guarantee minimum revenue over six years. If demand is strong and the facility rents 80% of its capacity, surpassing the minimum guarantee, Nvidia does not need to absorb the guaranteed capacity. Instead, because revenue exceeds the agreed floor, Nvidia receives a percentage of the excess. Conversely, if demand is weak and only 20% of capacity is rented, falling below the guaranteed level, Nvidia’s take-or-pay obligation would require it to cover the shortfall between actual and contracted minimum revenue per the agreement. Alternatively, Nvidia could lease back unused compute capacity for its own operations to bridge the gap.
Despite reported friction over tenant controls, the program has achieved significant traction. Just weeks after its formal launch, Nvidia had secured $36 billion in new agreements spanning six years. An Nvidia filing with the Securities and Exchange Commission stated: "Our commitments, which are typically six years in duration, totaled $36 billion as of July 26, 2026." Nvidia has not disclosed which portions of monetizable capacity it typically commits, making it impossible to determine the exact value of hardware intended under the $36 billion figure.