Nvidia announced a new partnership model for Neoclouds (specialized GPU cloud providers): Nvidia acts as a financial backstop and commits to leasing back unused GPU capacity at a fixed price. In return, Nvidia receives a share of its partners’ cloud revenue in addition to hardware revenue.
- First partners: Firmus (170,000 GPUs in Indonesia) and Sharon AI (40,000 GB300 GPUs)
- Predecessors: similar deals with $CRWV (-2,71 %) (6.3 billion $, 2025) and Lambda (1.5 billion $)
$NBIS (-12,85 %)$USCTF (-4,42 %)$CIFR (-4,08 %)$WULF (-3,25 %)$APLD (-3,45 %)$KEEL (-0,58 %)$CORZ (-2,56 %)$HIVE (-0,77 %)$BTDR (+7,15 %)$CLSK (-2,91 %)$MARA (+4,67 %)$HUT (-0,69 %)$RIOT (-1,1 %)
$IREN (-0,7 %) has so far not named as a participant—but has had a strategic partnership with Nvidia since May covering up to 5 GW of AI infrastructure, a $3.4 billion cloud contract , and Nvidia holds the right to purchase up to 2.1 billion IREN shares .
Adopting this model seems like a very logical step, since IREN will require extremely high investment costs (>$100 billion) to expand its entire pipeline.
In my view, this would be very positive in the short to medium term.
The biggest risk with $IREN (-0,7 %) and other Neocloud providers is not demand, but rather the financing of their expansion plans, which run into the billions. A backstop from Nvidia would mitigate precisely this risk: Banks are much more willing to grant loans when the world’s largest chipmaker guarantees capacity utilization—this would provide significant security and more favorable, virtually lower-risk financing. However, the revenue share would directly impact the profit margin . In the long term, companies would thereby cede part of their overall potential to Nvidia—essentially trading long-term profitability for medium-term security.