RUM Group’s AI Bet Hinges on Monetizing 250 Megawatts
RUM Group’s AI strategy hinges on turning more than 250 megawatts of unmonetized power capacity, acquired through Northern Data, into a compute-as-a-service business. Chief executive Chris Pavlovski says 150 megawatts in Georgia could generate $3 billion in annual revenue, but acknowledges the company must prove it can deploy the right customers and hardware. He also argues that Rumble’s video platform could eventually supply data for robotics training, though he characterizes that opportunity as early-stage.
The AI infrastructure wager rests on power that is not yet producing revenue
? chris-pavlovski says RUM Group’s central task is to monetize more than 250 megawatts of power that is not yet generating revenue. He singled out 150 megawatts in Georgia as representing what he calls a $3 billion annual revenue opportunity, while describing the broader 250-megawatt portfolio as capacity his team is working to bring online.
That distinction defines the company’s AI case. RUM Group is not simply adding an AI business to its video platform; it is trying to build a larger compute-as-a-service operation around assets acquired through Northern Data. Pavlovski said the market is “missing” the Quake AI story because the company must still demonstrate that it can execute on those assets.
We have to execute as a team and we have to show that we can take this business and capitalize on the opportunity that we have.
The existing monetized estate is Pavlovski’s operating reference point. He put it at roughly 22,000 GPUs, generating about $6 million to $7 million per megawatt. That figure applies to capacity already in service, rather than the unmonetized portfolio. The economics of the remaining power, he said, will depend on the customer and chipset deployed; he named Nvidia Blackwell and Rubin systems as configurations that could raise revenue per megawatt substantially.
Pavlovski said utilization on the current estate had reached roughly 80% to 85% about a month and a half after the Northern Data acquisition closed. His team’s stated priority is to identify the appropriate customers and hardware for the remaining sites, then convert that capacity into revenue.
Pavlovski also pointed to Elon Musk’s estimate of $30 billion to $50 billion per gigawatt as an indication of how rapidly AI-infrastructure economics could scale. RUM Group’s own stated opportunity is narrower and more concrete: $3 billion annually from 150 megawatts in Georgia, against a current benchmark of $6 million to $7 million per monetized megawatt.
Northern Data has changed both the revenue base and the outlook
Ed Ludlow noted that RUM Group reported $40.4 million in second-quarter revenue, up 61% from a year earlier. Northern Data, acquired on June 17, contributed $10.1 million during that quarter. ? chris-pavlovski said first-quarter revenue was $25.5 million, making the second quarter a record since the company went public.
| Metric | Reported figure |
|---|---|
| First-quarter revenue | $25.5M |
| Second-quarter revenue | $40.4M |
| Year-over-year revenue growth | 61% |
| Northern Data contribution in the quarter | $10.1M |
| Third-quarter revenue guidance | $87M–$93M |
The $10.1 million contribution arrived after a mid-June close, within a quarter that totaled $40.4 million. Management’s $87 million to $93 million third-quarter forecast therefore points to a markedly larger revenue base than the one reflected in the first partial period of ownership. Pavlovski said the company issued formal guidance for the first time because it now believes it can better predict its direction and communicate that outlook to the market.
The acquisition also supplied the operating structure for the pivot. Pavlovski said the company has rebranded as RUM Group and now has two divisions: Rumble, its video platform, and Quake AI, its compute-as-a-service business.
Rumble remained the larger contributor in the reported quarter, Ludlow said, generating more than $30 million in revenue. The near-term financial picture is therefore still a combination of the established video platform and a newly acquired compute operation, even as management’s guidance reflects the latter’s growing role.
Rumble’s video community is the proposed advantage in robotics
? chris-pavlovski does not present Quake AI as a reason to leave Rumble behind. He described AI compute as the immediate business, with high demand and constrained supply, and said RUM Group has power, land, and GPUs to serve that demand. But he also sees the video platform as a potential strategic asset if robotics becomes a major next phase of AI.
His premise is that video data will be compelling for training robotic systems. Rumble has more than 57 million monthly active users, Pavlovski said, and a community that could contribute video data and monetize it for a robotics-focused AI market. In his framing, that gives RUM Group an asset that typical “neo-clouds” lack.
We have what the neo-clouds have, and that’s the power, the land, the GPUs, but we also have something they don’t have, and that’s the video data.
The distinction matters because it separates two parts of Pavlovski’s strategy. Quake AI is intended to sell infrastructure capacity: power, land, and GPUs. The longer-range proposition is that Rumble’s audience and video library could provide data relevant to robotics training, potentially allowing users to contribute and monetize that data.
That is Pavlovski’s strategic thesis rather than an established moat. He described robotics as an opportunity that is still early, while positioning Rumble’s existing community as a possible advantage should demand for robotics-training video data develop.