Together AI Taps Rumble for Dedicated Blackwell Cloud Capacity

Together AI signed a multi-year cloud capacity agreement with Rumble Inc. for dedicated Nvidia HGX B300 systems. Rumble did not disclose the deal value, GPU count or deployment date, while Northern Data assets add more than 22,000 GPUs and approximately 250MW of capacity context. The practical test is whether the agreement turns into delivered Blackwell-class capacity for Together AI customers.

Together AI Turns to Rumble for Dedicated Blackwell Cloud Capacity

Together AI signed a multi-year cloud capacity agreement with Rumble Inc. for dedicated Nvidia HGX B300 systems, giving the AI cloud company another source of Blackwell-class compute outside the hyperscaler stack.

Rumble did not disclose the deal value, GPU count or deployment date.

The agreement gives Rumble a named AI-cloud customer for its compute strategy and gives Together AI additional capacity for customers training, shaping and deploying demanding AI models.

Rumble said the contract could be extended and increased in value depending on its success, but no specific competing cloud platform or benchmark was named.

The practical test is whether the agreement turns into delivered GPU capacity at a scale large enough to matter for Together AI customers.

Customer Proof Without Deployment Scale

Chris Pavlovski, chairman and CEO of Rumble, called the agreement “a major milestone for Rumble.” He said the contract gives Rumble “long-duration revenue visibility” while accelerating its cloud buildout.

Vipul Ved Prakash, founder and CEO of Together AI, said access to reliable Blackwell-class capacity is critical for customers running demanding AI models.

He said the partnership expands Together AI’s global GPU footprint and gives customers more choice in where and how they run workloads.

The commercial signal is narrower than a full infrastructure rollout.

Together AI is a named customer, but pricing, shipment volume, data center location and a start date for the dedicated capacity were not provided.

That leaves delivery timing and capacity scale as the main open items for the customer agreement.

Northern Data Assets Add Infrastructure Context

Rumble’s AI cloud plan is tied to its acquisition of Northern Data’s colocation and cloud businesses, which closed in Q2 2026.

Northern Data operates GPU clusters in Europe through Taiga Cloud, totaling more than 22,000 GPUs.

Ardent Data Centers is tied to approximately 250MW of capacity that is either operating or under development across eight global data centers in the US and Europe by 2027, including both colocation and owned sites.

Together AI offers AI cloud services and was founded in 2022.

Details about its own data centers remain limited, with company documentation saying they are located in North America.

The next signal is whether Rumble discloses deployment timing, GPU volume or customer usage tied to the Together AI agreement.