Rumble Earnings Preview (August 10): The Consensus Number Is For A Company That No Longer Exists

TL;DR

• Updated August 10, after the close: it all came down to the guide, exactly as this piece said it would. RUM Group reported $40.4 million of revenue, up 61% year over year, clear of the single-analyst $30.7 million placeholder this piece warned not to trade. Rumble's video business did $30.3 million; Quake AI, consolidating Northern Data from its June 17 close, added $10.1 million for a partial quarter. The net loss widened to $79.1 million from $30.3 million in Q1, mostly $28.3 million of acquisition transaction costs and higher depreciation on the newly consolidated GPU fleet, working out to roughly a $0.28 loss per share against a Street estimate near $0.09. Management delivered the first formal guide: Q3 revenue of $87-93 million, more than double Q2's reported total once Northern Data is in for a whole quarter. The stock closed the regular session at $6.21, down 2.74% from Friday's $6.39, before the release landed; in thin after-hours trading following the 5:00pm ET call it was quoted around $6.19, a snapshot minutes after the print, not a verdict.
• Rumble, which renamed itself RUM Group on June 18, 2026, reports Q2 2026 on Monday, August 10 after the close, with the call at 5:00pm ET.
• The published consensus is $30.7 million of revenue and a $0.10 loss per share, from exactly one analyst. Treat it as a placeholder, not a bar: one estimate is not a consensus, and it describes the pre-acquisition company.
• The acquisition changed what this company is. Rumble took 85.2% of Northern Data, folded it together with Rumble Cloud into a unit called Quake AI, and now sits on roughly 22,000 NVIDIA H100 and H200 GPUs and about 250MW of current and planned power across ten data centres.
• Northern Data's own 2026 revenue outlook was raised in June to EUR 170-190 million from EUR 130-150 million, and it did EUR 43 million in Q1 with GPU utilisation up from 62% in December to about 85% in March. On a pro-forma basis the combined group would have done roughly $75 million in Q1, not $25.5 million.
• Management has said it will begin issuing formal guidance with this release. That, not the June-quarter revenue line, is the tradeable event. Options price a 12.1% move against a $6.39 close.

When Does Rumble Report Earnings?

Monday, August 10, after the close, with the earnings call at 5:00pm ET. The rest of the week's slate is in the earnings calendar.

The Board

One analyst, one legacy revenue line, and a company that has changed shape underneath both.

What Landed (August 10)

	Q1 2026	Q2 2026 actual
Total revenue	$25.5m	$40.4m, up 61% YoY
Rumble video revenue	$25.5m	$30.3m, up 21% YoY
Quake AI / Northern Data revenue	n/a (deal not yet closed)	$10.1m (partial quarter from June 17 close)
Net loss	$30.3m	$79.1m
Loss per share	n/a	roughly $0.28 (Street estimate near $0.09)
Adjusted EBITDA loss	n/a	$16.6m, improved from $20.5m a year earlier
Cash and equivalents (June 30)	n/a	$203.3m ($220.5m total liquidity)
Q3 2026 guidance	none issued	$87-93m revenue, first formal guide

The consensus this piece warned about turned out to understate the quarter by a wide margin: $40.4 million against the single-analyst $30.7 million placeholder, and against the roughly $34.9 million a broader Street estimate had settled on by report time. Rumble's own video business grew to $30.3 million, ahead of a simple extrapolation of Q1's $25.5 million, and Quake AI contributed $10.1 million for roughly two weeks of consolidation, a run rate that scales toward the low end of Northern Data's own raised EUR 170-190 million full-year outlook once a whole quarter is in.

The loss is where the quarter got expensive. Net loss widened to $79.1 million from $30.3 million in Q1, and management attributes most of the gap to $28.3 million of Northern Data acquisition and transaction costs plus higher non-cash depreciation and amortisation on the newly consolidated GPU fleet, both one-time-adjacent items rather than a deterioration in the underlying businesses. Adjusted EBITDA loss actually improved to $16.6 million from $20.5 million a year earlier, which is the cleaner read on operating trend than the GAAP net loss.

The guide is exactly the event this piece said it would be. Q3 revenue of $87-93 million is more than double Q2's reported $40.4 million, which is arithmetic rather than a surprise: Q2 carried perhaps two weeks of Northern Data, and Q3 carries all thirteen. A full-quarter run rate for the combined group in the high-$80-million to low-$90-million range is the first hard number anyone outside the company has had to underwrite the Together AI contract and the 250MW power build against, and it lands inside what a pro-forma read of Northern Data's own guidance implied.

Read The Consensus Before You Trade It

This is a textbook case of a number that is individually defensible and collectively useless.

$30.7 million is a reasonable extrapolation of Rumble's own video business: Q1 2026 revenue was $25.5 million, up 7% year on year, with global monthly active users up 8% sequentially to 56 million. Grow that a bit, add a stub of something else, and you land near $30 million.

But Rumble closed its takeover of Northern Data in mid-June, inside the June quarter. Northern Data is not a rounding error next to Rumble: it turned over EUR 43 million in Q1 alone, more than Rumble's entire video business. So Monday's revenue line contains an unknown number of days of a business roughly twice the size of the reporting company, bolted on at an unannounced date, against a consensus built by one analyst who may or may not have modelled it.

The practical consequence: any headline that says Rumble "beat" or "missed" revenue on Monday is close to meaningless. Ask instead what the pro-forma run rate is, and what the first formal guide says.

What The Combined Company Actually Owns

Strip the ticker change and the branding, and RUM Group is now two businesses that have nothing to do with each other.

The video platform. 56 million monthly active users, growing at single digits, monetised through advertising and subscriptions, structurally tied to a politically-defined audience. Q1 revenue $25.5 million.

Quake AI, the infrastructure business. Roughly 22,000 NVIDIA H100 and H200 GPUs across nine to ten data centres, about 250MW of energised and contracted power (of which the company describes over 200MW as unmonetised), and a multi-year contract with Together AI carrying $270 million of total contract value on NVIDIA Blackwell B300 systems. Northern Data raised its 2026 revenue outlook to EUR 170-190 million, roughly a 30% increase on the prior range, and lifted GPU utilisation from 62% in December 2025 to about 85% in March 2026.

That second business is the reason the stock is worth analysing at all, and it is the reason the comparison set is now CoreWeave rather than any media company. It is also, on the numbers disclosed, a fraction of CoreWeave's scale, with a far weaker balance sheet behind it.

The Cost Of Getting Here

Q1 tells you what this transformation costs. Against $25.5 million of revenue, expenses ran $64.6 million, including $4.8 million of acquisition-related transaction costs, producing a net loss of $30.3 million against $2.7 million a year earlier. That is a loss larger than revenue, before the acquired business and its capital intensity land on the balance sheet.

Two hundred megawatts of unmonetised power is a genuine asset and it is also a genuine liability until it is contracted, because energised capacity costs money whether or not a customer is renting it. The bull case is that the Together AI contract is the template and that utilisation keeps climbing. The bear case is that a company losing more than it earns is now financing data centre buildout, and that the equity is the funding source.

Why The Guide Is The Whole Event

Management has signalled that formal financial guidance begins with this release. For a company that has never guided, that first number does more than any quarterly revenue line can:

• It tells you whether Northern Data's EUR 170-190 million outlook is being carried through, raised or trimmed under new ownership.
• It sets a denominator for the $270 million Together AI contract, which is otherwise a headline without a period attached. A total contract value spread over several years is a very different thing from an annual run rate, and the piece of information that makes it interpretable is the term.
• It forces a statement on capital expenditure, which is the only honest way to size the funding gap.

A guide that consolidates Quake AI and puts a real revenue range on it re-rates the stock regardless of the June quarter. A release that reports a messy stub quarter and defers guidance again is the bad outcome, and at a 12.1% implied move the option market is pricing meaningful odds of exactly that.

The Options Angle

A 12.1% implied move on a $6.39 stock is about 77 cents of expected range. That is not expensive for a company reporting its first quarter in a new corporate shape, with a first-ever guide attached, against a one-analyst consensus.

• This is the rare setup where buying volatility is defensible. The distribution is genuinely two-humped: a real guide with consolidated numbers, or another quarter of "transition". A straddle at 12.1% pays on either tail.
• The equity call is a pass. The infrastructure asset is real, the disclosure is not yet good enough to underwrite it, and a company running a net loss larger than revenue while building data centres has an equity-issuance problem that no single quarter resolves.
• Covered calls on a $6 stock are barely worth the commission and cap the only outcome (a guide-driven re-rating) that justifies owning it.

Trade log

#	Stance	Structure	Strikes and expiry	Cost or credit	Spot at writing	Implied move	Breakeven
1	Long volatility	Long straddle	$6.50 straddle, Aug 14	~12.1% of spot; live price not sourced	$6.39, Aug 7 close	±12.1%	Needs a move beyond about $5.62 or $7.16
2	Pass	Long shares into the print	n/a	n/a	$6.39, Aug 7 close	±12.1%	Scored against the Aug 11 close
3	Pass	Short premium of any kind	Aug 14 expiry	Not sourced	$6.39, Aug 7 close	±12.1%	Scored as a trade not taken

Row 1 is logged against the implied move rather than a live debit, because option prices for the August 14 expiry could not be sourced at writing. It is scoreable: it wins if the realised move exceeds 12.1%.

Updated August 10, after the close and the call, against a thin after-hours print of $6.19 (down about 3.1% from Friday's $6.39 close, not the Aug 11 close this trade log is actually scored against): none of the three rows resolve tonight, by this piece's own terms. Row 1 needs a move beyond 12.1% by the August 14 expiry, and the after-hours tape has moved about a quarter of that. Row 2 is explicitly scored against the August 11 close, which has not happened yet. Row 3 runs to the August 14 expiry. I read tonight's after-hours print as a market that has not yet decided what to do with a revenue beat, a wider-than-expected loss, and a first guide that implies the combined business is real: that is exactly the two-humped distribution the straddle was written for, and a 3% after-hours move on those three inputs together looks more like a market still reading the release than a market that has settled on an answer. Full scoring on all three rows follows in the next update.

The One-Line Read

The revenue consensus describes the old Rumble, the company reporting on Monday is a GPU landlord with a video site attached, and the only number worth waiting up for is the first formal guide, which is the first thing management has ever offered that can actually be checked.

Updated August 10. The guide arrived, and it was the whole event, as this piece argued going in. Revenue beat every version of consensus in circulation, the loss widened on acquisition costs rather than on the operating business, and Q3's $87-93 million guide is the first number from this management team that can be held to account next quarter. The stock's muted after-hours move, about 3%, says the market has not finished pricing that in.

Updated August 17. The market finished pricing it in, and then some: from the $6.21 report-day close the stock reached $7.46 by the August 14 close and printed $8.37 intraday on August 17, a four-session gain of roughly 35% with short interest at about a fifth of the float. Whether it is a buy after that run, the squeeze arithmetic, and what the weeklies now cost are taken up in the decision piece published August 17.