Open in interactive viewer → charts, metric popovers & call review

Rumble's First Profit Lands the Week the Market Stopped Financing AI

Quake AI is now 70% of revenue and adjusted EBITDA just turned positive — but Q4 guidance steps backwards and the 250-megawatt anchor project is sliding into 2028.
RUM · Earnings Call · 2026-11-11

A milestone nobody wants to pay for

Rumble picked an awkward week to become a profitable company. Revenue for Q3 2026 came in at $96.6 million against a guidance range of $87–93 million, roughly four times the $24.8 million of a year ago, and adjusted EBITDA printed positive at $4.4 million — the first time as a public company, against a $15.1 million loss a year earlier and a $16.6 million loss just last quarter. Chris Pavlovski's framing was blunt: “We have real assets running real AI workloads for real customers, generating real cash.” — Christopher Pavlovski, Founder, Chairman, and CEO · 2026-11-11 And yet the tape has spent three months voting the other way. The global AI data center theme is down roughly 16% over 30 days and 21% over 90, with a staggering 63 decliners against 8 advancers in the last month; frontier AI labs sit near the top of the 30-day decliner list. Rumble's own shares peaked at $10.23 on 2026-08-25 and have given back 41.7% since. Valuation has compressed with it — price to revenue sits at 26.7x, down 29% year over year. That gap — record results into a de-rating theme — is the entire story of this call. Management spent most of its prepared remarks not celebrating, but explaining why the financing weather changed.

Quake AI is the company now; video is the moat

This was the first quarter reported in two segments, and the split is stark. Quake AI delivered $68.2 million of revenue at roughly a 34% adjusted EBITDA margin. The GPU estate — 22,000 GPUs, described as one of the largest in Europe — ran at 85% utilization, which management characterized as effectively sold out. The video side did $28.4 million, up 15%, but lost $12.1 million at the adjusted EBITDA line. So the tab has been flipped. What was a decade-long advertising and creator-platform narrative is now a minority segment undergoing a hard cost reset — more than 20% run-rate operating expense reduction, a reduction in force, a 90% creator revenue share on video ads, and a new WWE distribution deal. The video business is no longer pitched as a growth engine but as strategic ballast: “it's almost like a moat for the Quake AI business going into the future.” — Christopher Pavlovski, Founder, Chairman, and CEO · 2026-11-11 The stated payoff is a spatiotemporal data marketplace for AI and robotics companies, targeted for late 2027 — a genuinely novel idea, and one that is eighteen months from mattering.

Guidance that goes backwards

Here is the number that matters more than the beat: Q4 guidance is $88–94 million, below the $96.6 million just reported. Masci was unusually candid about why, flagging that Q3 carried “items that we don't expect to repeat in the fourth quarter” — Michael Masci, Chief Financial Officer · 2026-11-11 — specifically a barter agreement with one customer that pulled forward a little over $3 million with an outsized cliff in Q4. Strip it out, he said, and Q3 was merely at the top of the range. The bigger swing factor is the H100s. The vast majority of Rumble's Hopper contracts roll off by the end of Q1 2027, and those were signed in 2025 at stale rates. Management implied roughly $1.40 an hour in the quarter. Where they expect to renew:

The prevailing rates of H100s today by any market metric that you see are firmly with a 2-handle on them, and we definitely are seeing market rates that are similar in our discussions with our customers.

Michael Masci, Chief Financial Officer · 2026-11-11
That is the entire bull case in one sentence — a doubling of unit price on an existing, fully depreciating asset base, with no new capital required. It is also why management will not commit to duration: they want a blend of short contracts with spot upside, not five-year locks at today's prices. Given that supply is constrained and demand is not, that is a defensible trade. It is also a bet on 2027, priced against a market that is currently unwilling to underwrite 2027.

The financing wall at Maysville

The tell is in what happened to the 250 megawatts. Last quarter, that capacity was the headline opportunity: “we're really focused on that 250 megawatts of unmonetized capacity” — Michael Masci, CFO · 2026-08-10 was the standing answer to nearly every growth question. This quarter, the same asset has a caveat — a portion will now slip beyond 2027, and Maysville is under discussion with its customer for an "alternative path forward." Masci reasserted the AI compute market thesis — “I don't see any point in which we're going to have capacity meet that demand in the next 1 to 2 years” — Christopher Pavlovski, Founder, Chairman and CEO · 2026-08-10 — but the unit of growth has shrunk from a gigawatt ambition to getting 250 megawatts monetized at all. The funding plan is the interesting part, and it is why the project equity line item matters. Masci's answer to a $11–12 billion buildout question was not debt: “customer prepayments are really going to be a large portion of the financing for projects that is necessary” — Michael Masci, Chief Financial Officer · 2026-11-11. That is prepayment as a substitute for equity, which works only if the customer is desperate — which, per Rumble, it is. Balance sheet-wise the claim of being among the least levered NeoClouds holds up, though the direction is worth watching: liabilities to assets jumped to 32.9% with the acquisition. Liquidity stood at $142.3 million ($134.4 million cash plus 94.68 Bitcoin worth $7.9 million), against negative $99 million of free cash flow and a capex line that spiked to $42 million in a single quarter. That combination does not fund 250 megawatts without either customer checks or the debt markets reopening — and the call's repeated theme, in the form of debt markets and financing markets keywords that did not exist in Rumble's vocabulary a quarter ago, is that those markets have gotten pickier. One quiet flag: receivables to revenue has climbed to 125.8% from roughly 49% two quarters ago. Combined with a barter contract inflating Q3, revenue quality deserves a skeptical eye, not just revenue quantity.

What stopped being said

The most telling thing about this call is its silence. For the better part of two years, Rumble's earnings calls were crypto calls: the Rumble Wallet, the Tether advertising commitment, the Bitcoin treasury, ARPU as headline. A year ago there was 2025-08-11's admission that “materially moving towards adjusted EBITDA breakeven is still important that we've spoken about before, but it's become a lesser relative priority as we evaluate investing into aggressive growth.” — Brandon Alexandroff, Chief Financial Officer · 2025-08-11 That promise was finally kept — but the crypto scaffolding has been dismantled so thoroughly it barely earns a mention. The 94.68 Bitcoin is disclosed as a number, not a strategy. What replaced it is narrower and more binary. Rumble is now a small, unlevered GPU landlord with an aging Hopper fleet, an H100 repricing event twelve weeks out, a stalled megaproject, and a video asset kept alive as a data moat for a marketplace that does not launch until late 2027. The upside case is genuinely option-like: if unmonetized capacity reaches market in 2027 with Vera Rubin-generation economics, the revenue per megawatt math that management quoted last quarter — nearly $11 million per megawatt per year — would dwarf today's run rate. The risk is that the market has already decided it will not finance the wait, and this quarter's guidance gives it another data point to be nervous about.