Anthropic is Rum Group’s $13.7bn compute customer, The Information reports

Anthropic was reported to be the unnamed customer in Rum Group's $13.7 billion, six-year computing contract, according to The Information. The agreement would represent a major long-term infrastructure commitment for Anthropic and a potentially material revenue source for Rum Group, owner of video platform Rumble. Neither company had publicly confirmed the report.

Analysis

If the reported counterparty is verified, RUM’s valuation framework shifts from advertising-platform optionality to AI-infrastructure execution. The relevant question is not the headline contract value but whether the arrangement includes minimum-revenue commitments, customer-funded capacity, termination protections, and pass-through power costs. At roughly $2.3bn of annualized contractual value, even a modest 15-25% infrastructure EBITDA margin would be transformational; conversely, a take-or-pay commitment without adequate pre-funded GPU and data-center financing could create substantial dilution and balance-sheet strain.

Anthropic concentration would improve perceived demand quality but materially increase single-customer risk. Anthropic’s compute needs are large, yet its own funding and model economics remain dependent on continued capital-market access and its strategic relationships with Amazon and Google; a shift in those relationships, model-training efficiency gains, or a customer cancellation would leave RUM with specialized capacity and potentially lower resale economics. The market should also discount the possibility that reported contract value includes contingent capacity expansion rather than committed near-term revenue.

Near-term upside is driven by formal confirmation and disclosure of contract terms; over 1-3 months, financing announcements, GPU procurement commitments, and construction milestones determine whether the revenue narrative is investable. Over 6-18 months, the key issue is whether RUM can earn infrastructure-like returns while preserving its differentiated media-platform economics, rather than becoming a capital-intensive subscale cloud provider. CoreWeave (CRWV), Nebius (NBIS), and crypto-to-HPC conversion operators such as Core Scientific (CORZ) are the more direct competitive read-throughs: verification would reinforce the scarcity value of deployable AI capacity, but also raise competition for GPUs, power interconnects, and project financing.

Key Decisions for Investors

• Do not chase RUM solely on the reported attribution. Establish an event-driven watch position only after company confirmation or an SEC filing identifies minimum commitments, contract duration/termination rights, and required capex; absent those disclosures, the equity is being priced on gross backlog rather than risk-adjusted cash flow.
• For a 1-3 month catalyst trade, consider a small long RUM position only if financing is predominantly customer-backed or non-recourse and management indicates first revenue within 12 months. Size for binary disclosure risk; exit if incremental equity issuance, unrestricted termination rights, or materially delayed power/GPU delivery emerges.
• Use a relative-value screen rather than a broad AI-infrastructure basket: long RUM versus short CRWV is attractive only if RUM’s confirmed contracted-revenue multiple remains materially below CRWV after adjusting for capex and customer concentration. Do not initiate until RUM publishes enough economics to calculate EV/backlog and expected EBITDA conversion.
• Set downside alerts around an Anthropic funding setback, revised cloud commitments with AMZN/GOOGL, or evidence of lower training-compute intensity. Any of these would impair the implied utilization assumptions and could compress AI-capacity peers before RUM’s own financial impact becomes visible.