On July 30 Reddit reported revenue of $804.9 million, up 61%, its eighth consecutive quarter above 60%. Net income rose 183% to $252.8 million. Free cash flow was $260.7 million on capex of only $1.1 million (I had to make sure my eyes didn’t deceive me there) which is 0.1% of revenue. Third-quarter guidance came in above consensus on both lines. The next day the stock closed at $140.67, down 21%, its steepest intraday decline since the 2024 IPO. One thing did it. Steve Huffman wrote that search referrals were choppy. And the shareholder letter said nothing about data licensing beyond one clause reporting Other revenue of $43 million.
Look at the top line and the bottom line. Revenue up 61%, net income up 183%, guidance above consensus on both. Nothing in the quarter broke. What did not happen is a renewal: the January 2024 licensing contracts ran two to three years, the Google agreement inside them is reported at about $60 million a year, and the quarter closed with the expiry in view and nothing new announced. The stock lost a fifth of its value the next day.
The fear that produces is not about $60 million. It is that Reddit is Google’s pseudo-social network. The users arrive because a search engine points at them. The revenue arrives because Google pays for the data. Neither half belongs to Reddit. On that reading there is no organic business underneath, just a distribution deal with an expiry date.
The bullish view though is that Reddit can replace the money. OpenAI and Anthropic need the same data, and the suit in San Francisco Superior Court is how Reddit establishes the right to charge anyone for access at all. What it cannot replace is the other half of what Google was doing, which is sending the traffic. A model builder pays for the data and sends nobody. Whether that substitution is the same thing is the question, and the honest answer is no.
There is also a version of this where the renewal has not happened yet rather than failed. The January 2024 deals were signed before Reddit was public, and before anyone knew what this data was worth. Since then a court approved the $1.5 billion Bartz settlement, and Reddit has reportedly discussed cutting Google off entirely. If Google signs again, it does not sign at $60 million a year. The contract running out is not the same as the money going away. Huffman said as much on the call. He called the range of outcomes wide, said these deals are not binary, and described the Google relationship as several separable pieces, training data, search links and AI Overviews, with no resolution imminent.
I personally am a part-time Reddit user. I peruse the Boston sports subreddits, but am in no way an expert on the interaction side of the company the way I would consider myself with Meta’s apps or X (formerly Twitter). What I do understand is that Reddit is a forum site with a bunch of different subdomains thematically called subreddits. Advertisers can advertise on popular subreddits for more money (guessing) and voilà, another great business is made.
Which leaves something the quarter can actually settle. If Reddit’s growth were really Google’s growth, it would show up in the composition of that growth and I am not seeing it that way.
The quarter itself
| Q2 2026 | Year-over-year | |
|---|---|---|
| Revenue | $804.9M | +61% |
| Advertising revenue | $762M | +64% |
| Other revenue | $43M | — |
| Net income | $252.8M | +183% |
| Adjusted EBITDA | $342.8M | +106% |
| Adjusted EBITDA margin | 42.6% | vs 33% |
| Gross margin | 91.3% | vs 90.8% |
| Free cash flow | $260.7M | +135% |
| Capital expenditures | $1.1M | 0.1% of revenue |
Trailing twelve month operating cash flow crossed $1 billion for the first time. The balance sheet is $2.79 billion of cash and marketable securities against $350.9 million of total liabilities, none of it debt. Management bought back 1.5 million shares at $157.57 average, now underwater, which tells you how fast the narrative moved and nothing about the business.
One arithmetic note, because the rounded figures do not tie: operating cash flow of $261.9 million less capital expenditures of $1.1 million is $260.8 million, not the $260.7 million Reddit reports. The unrounded line items give $260.742 million. I use the reconciliation, not the subtraction of rounded inputs.
One thing the table does not show. Third-quarter guidance of $860 million to $870 million reads as another beat, but against $584.9 million in the same quarter last year it implies growth of roughly 48%. Reddit’s own guidance has revenue decelerating from 61%.
None of this was in dispute on Friday.
The funnel, and a disclosure that got worse
| U.S. users, Q2 2026 | ||
|---|---|---|
| Daily uniques | 53.2M | vs 53.5M in Q1 — first sequential decline |
| Daily uniques, year-over-year | +6% | international +28% |
| Logged-in dailies | +1% | the Americans who came back on purpose |
| Logged-out dailies | +10% | roughly, arrivals from a search result |
| Weekly uniques | 197.2M | +9%, against +24% globally |
The American business is converting a very large weekly audience into a barely growing daily one, and the marginal American user is a stranger who may not come back.
Huffman’s response was to reframe the metric. “While our visibility in referral traffic remains low, we’re not building for drive-by traffic. We’re building a daily destination.” That is the correct strategy and a genuinely weak thing to say in the same quarter your U.S. dailies went backward. On the call he was blunter, contrasting Google’s ten blue links, which built the funnel, with AI Overviews, which he said have “yet to make a similar level of positive impact.”
Then Drew Vollero: “Q2 will be the last quarter we report logged in and logged out user metrics.” Reddit keeps disclosing total U.S. and international dailies and weeklies. Not the split.
To be fair, management pre-announced the change earlier in the year, so this is not a reaction to a bad quarter. It is still the wrong quarter for it to land. The split is the only public instrument for measuring how much of Reddit’s growth depends on a search engine rebuilding itself to send fewer clicks. Retire it in the same call where you concede referral volatility, and from the third quarter onward nobody outside the company can track the thing the equity now trades on.
Which leaves Reddit’s own disclosure as the clean read. And Reddit is retiring the useful half of it.
The ARPU engine, which is the actual business
U.S. ARPU was $11.85, up 51% year-over-year, on a logged-in user base that grew 1%.
Revenue grew 56% on a daily base that grew 6%. Reddit is not growing by adding Americans. The growth is price, and the rate of that repricing is the entire domestic thesis.
Reddit discloses the split, and it ties out. Ad prices rose about 40% in the quarter and impressions rose 17%, which compounds to the 64% advertising revenue growth almost exactly. So it is mostly price, but not only price. Reddit is still finding new places to put ads.
For scale: Meta earned $16.86 from each daily user worldwide in the same quarter. Reddit earned $6.18, which is 37% of that. Meta no longer breaks its number out by geography, so there is no clean U.S.-to-U.S. comparison from primary sources. But the direction is clear enough. What limits this business is not the size of the audience.
The international gap
Reddit has more users abroad than at home and earns roughly one-fifth as much from each of them.
International ARPU is $2.26 against $11.85 in the U.S., a 5.2 times spread, on 77.1 million international daily uniques versus 53.2 million domestic. Snap’s Europe ARPU is $3.34 and its Rest of World ARPU is $1.20, so Reddit sits inside the normal range and nothing about the gap is anomalous.
One caveat governs every number here. Reddit’s segment disclosures do not multiply together — ARPU times daily uniques misses reported segment revenue by roughly $7.7 million, because Reddit apportions advertising revenue by where impressions are delivered rather than where the revenue table puts it. The full reconciliation is in the sources. Every calculation below runs on reported segment revenue and never multiplies an ARPU against a user count, which also holds the user base flat by construction.
At half the current U.S. level, $5.93 is 2.62 times today’s $2.26. Applied to reported international revenue with no user growth at all, $166.8 million a quarter becomes roughly $438 million — about $271 million a quarter of incremental revenue, $1.08 billion annualized, at gross margins above 90%.
Other revenue, which carries the licensing contracts, runs at about $172 million a year. The international gap, on the existing user base, is worth roughly six times that — and it does not depend on signing anyone.
The lawsuit is a pricing instrument, not a payday
Reddit sued Anthropic on June 4, 2025 in San Francisco Superior Court, pleading five California state-law claims: breach of contract, unjust enrichment, trespass to chattels, tortious interference, and unfair competition under Section 17200. There is no copyright count, and the omission is deliberate.
Anthropic removed the case arguing the claims were copyright claims in disguise and therefore preempted. Judge Trina Thompson disagreed and sent it back, finding the conduct went beyond copying and implicated contractual rights under Reddit’s User Agreement distinct from the rights copyright confers.
That ruling is what matters. If terms of access are enforceable against automated collection independent of who owns the content, the fair use question stops mattering, because you never reach it. You are in contract, and contract is where Reddit wants to be. Reddit ran the same play against Perplexity and three scraping intermediaries, which tells you this is a strategy and not a grievance.
The comparison everyone is reaching for is the Bartz settlement, given final approval on July 20 at $1.5 billion. That number was a function of statutory copyright damages across a certified Works List of 482,460 titles, with a catastrophic exposure tail forcing it. Reddit pled no copyright claims and has no statutory multiplier. Its recovery mechanism is restitution and disgorgement, which is a different order of magnitude and a different question for a court.
And even a large recovery would be a gain contingency: non-recurring, unrecognizable until realized, excluded from adjusted EBITDA by construction. It cannot re-rate a multiple. Reddit sits on $2.79 billion of net cash and over $1 billion of trailing free cash flow. It does not need a check. It needs a precedent that lets it charge rent, before the January 2024 contracts roll.
What you are paying, against five other advertising businesses
| TTM to | EV | TTM EBITDA before SBC | EV/EBITDA | TTM rev growth | Turns per point | |
|---|---|---|---|---|---|---|
| Reddit (like-for-like) | Mar 2026 | $25.7B | $963M | 26.7x | 70.6% | 0.38 |
| Reddit (latest) | Jun 2026 | $25.7B | $1,140M | 22.5x | 66.6% | 0.34 |
| AppLovin | Mar 2026 | $137.5B | $5,133M | 26.8x | 66.4% | 0.40 |
| Meta | Mar 2026 | $1,478B | $131,620M | 11.2x | 26.2% | 0.43 |
| Mar 2026 | $12.2B | $1,228M | 10.0x | 16.3% | 0.61 | |
| Snap | Mar 2026 | $8.9B | $777M | 11.5x | 10.3% | 1.12 |
| The Trade Desk | Mar 2026 | $7.3B | $1,196M | 6.1x | 15.5% | 0.39 |
Adjusted EBITDA is not defined consistently across these companies and Meta does not report it at all, so the measure here is built the same way for all six from filed statements: operating income plus depreciation and amortization plus stock-based compensation, summed over the trailing four quarters. That reproduces what Reddit, Snap and Pinterest call adjusted EBITDA to within a few percent, and it is computable for Meta. Enterprise value is market capitalization plus long-term debt less cash and short-term investments. Every financial line comes from SEC filings; the market capitalizations come from a quote aggregator, because no primary source publishes one, and every multiple inherits that.
Reddit appears twice on purpose. Meta reported its second quarter on July 29. Snap reports on August 3, Pinterest on August 4, and AppLovin and The Trade Desk on August 6. March is the last quarter for which all six can be set side by side, so every comparable is shown on March numbers and the like-for-like Reddit row is its own trailing twelve months to March. The June row is the current state of the business and sits in italics because it is a quarter ahead of everything beside it. That is a real limitation and worth saying plainly: four of the five comparables report within the week, Meta already has, and any of them could move this table.
These six are not all peers, and they are not meant to be. AppLovin is the only name growing at Reddit’s rate, and it sells its own users’ attention through an auction, which is the same machine. Meta is what this business would want to look like at scale and if its wildest dreams came true. Pinterest is the closest match on size and shape: built on what people are interested in rather than who they know, most of its revenue in the U.S., and a big international base it has not learned to monetize, which is Reddit’s exact problem. Snap and The Trade Desk are not comparables at all. They are price markers. Snap is what the market pays when growth stalls at 10%, and The Trade Desk is what it pays for ad technology that is decelerating. Between them they set the floor.
The last column of the table is the one that matters, and it is less flattering than the June row alone would suggest. On the like-for-like March basis Reddit sits at 0.38 turns of enterprise value per point of revenue growth, a hair below The Trade Desk at 0.39 and AppLovin at 0.40, and below Meta at 0.43, Pinterest at 0.61 and Snap at 1.12. Reddit is the cheapest of the six on that measure, but by a whisker rather than a margin. The June row puts it at 0.34 and makes the discount look about twice as wide as the clean comparison supports. Use the March row. Turns per point of growth is my construction, not a standard metric, and it is doing a lot of work here.
On the headline number I was wrong, not half right. Reddit trades at 26.7 times, effectively tied with AppLovin for the most expensive of the six. It is cheap only after adjusting for growth, and only barely. Whether the growth persists is not a footnote to the valuation. It is the valuation.
What the multiple should be, under each case
| Four quarters out | Case one: the funnel argument is right | Case two: the ARPU engine holds |
|---|---|---|
| U.S. revenue growth | +25% (ARPU growth halves) | +51% (matches current ARPU growth) |
| U.S. revenue | $797.6M | $963.5M |
| International growth | +40% | +50% |
| International revenue | $233.5M | $250.2M |
| Quarterly revenue | $1,031.1M (+28%) | $1,213.7M (+51%) |
| EBITDA at 42.6% margin, annualized | $1.71B | $2.01B |
| Multiple | 11–13x | 18–22x |
| Enterprise value | $18.8–22.2B | $36.2–44.3B |
| Per share, fully diluted | $104–121 | $188–227 |
Both cases are built four quarters out, hold the adjusted EBITDA margin flat at the 42.6% Reddit reported in Q2, and convert to the before-SBC basis at 97.3%. That makes the margin the conservative assumption in the bull case and the generous one in the bear. The only thing separating them is the growth path.
Case one’s 28% is Meta territory. Meta prints 26.2% and trades at 11.2 times; Pinterest at 16.3% trades at 10.0 times. A modest premium for a 43% margin and effectively zero capital intensity gets you 11x to 13x. Case two’s 51% sits between Meta’s 26% and AppLovin’s 66%; interpolating on where the other five sit, at roughly 0.40 turns per point, puts it at 18x to 22x. Using the ARPU rate rather than the 56% by which Reddit’s U.S. revenue actually grew is the more conservative of the two available inputs.
Enterprise value is not what anyone owns, so both cases are carried down to a share price. Adding back $2.79 billion of net cash and dividing by the 207.0 million fully diluted share count gives $104 to $121 in case one and $188 to $227 in case two, against a $140.67 close. The fully diluted count is the unflattering one; on the 192.4 million shares the quote services use, both ranges rise about 8%.
Both cases assume the multiple compresses toward the group. It may not. If Reddit holds today’s 26.7 times on $2.01 billion of EBITDA, that is roughly $53.7 billion of enterprise value, or about $273 a share. That case requires 0.53 turns of enterprise value per point of growth, against 0.43 for Meta and less for everything else in the set still growing. Only Pinterest at 0.61 and Snap at 1.12 sit higher, and they sit higher because their growth is slow, not because the market is paying up for it. So it is the least defensible of the three on this framework. It is also the one that happens if nothing about how the market prices this company changes.
Case one midpoints to $20.5 billion, case two to $40.3 billion, and the current $25.7 billion sits 26% of the way from the first to the second. That is a two-point interpolation between my own scenario midpoints and nothing more. It is not a market-implied probability, because the market is not choosing between two outcomes I invented, and it moves if either endpoint moves.
My view is that 26% is too low, and the reason is the composition rather than the headline. U.S. revenue grew 56% on a U.S. daily user base that grew 6%. That is not a business being carried by the funnel in the way the bear case requires, and the international lever has not been pulled at all.
The bear case
The bear case is not that Reddit loses $60 million. It is that logged-out growth was the growth, and it is ending.
U.S. logged-in dailies grew 1%. If search referrals have been feeding the top of the funnel, and AI Overviews structurally reduce that feed, Reddit’s American user base is close to flat and every dollar of domestic growth has to come from price. ARPU carries you a long way at 51%, but ad load and pricing both have ceilings.
And here is the fact that should trouble anyone who found the ARPU section persuasive, including me. Snap’s North America ARPU rose 10% last quarter while its daily actives fell 7%, from 99 million to 92 million, on revenue that grew 2.4%. Decompose it: 1.024 divided by 0.93 is 1.101, which is the entire 10%. Almost none of Snap’s ARPU growth is monetization. It is a shrinking denominator.
That breaks the comparison I made earlier. ARPU is a ratio, and a ratio rises for two opposite reasons. If ARPU growth can be manufactured by losing users, Reddit’s 51% on 1% logged-in growth does not by itself demonstrate that revenue is growing independently of users. It could be the same curve, with Snap further along it, caught a few years earlier, and Reddit’s denominator merely flat rather than falling yet.
The answer is that the decomposition runs the other way for Reddit, and it is not close: 56% revenue over 6% users against Snap’s 2% over minus 7%. There is one asymmetry worth naming, since I am leaning on this comparison. Snap’s decomposition reproduces its reported ARPU exactly, because Snap’s published user count is its own ARPU denominator. Reddit’s does not. Its 56.1% over 5.8% implies 47.6% ARPU growth against the 51% Reddit reports, and the difference is the apportionment gap above. Reddit’s side is approximate where Snap’s is exact. It changes nothing, because 47% and 51% land in the same place and neither is within sight of a 2% revenue line. But you should not have to find that for yourself.
What survives of the challenge is the timing question, and it survives intact. Snap’s revenue growth stalled before its user base did. If Reddit is on the same curve a few years back, the tell shows up in advertising revenue growth first, not in ARPU, which will keep looking healthy right through the turn. That is why the trigger below is an ad revenue number rather than an ARPU one, and it is the single thing most likely to make this piece look naive in a year.
The publisher data supports the idea that AI Overviews are cutting search referrals, though less dramatically than the first round of coverage suggested, and every figure in this paragraph is secondhand: the Wall Street Journal’s article could not be retrieved, so these come from the trade outlet PPC Land, which covered both the original piece and its correction. As corrected on July 22, Semrush figures put U.S. organic Google traffic between June 2025 and June 2026 down 18% at USA Today’s national paper, 20% at Politico, and 31% each at CNN and Business Insider. The corrected trailing twelve-month chart runs from about 3% at People to 44% at the Washington Post, alongside gains of roughly 16% at the Guardian and 15% at the BBC. An 18% to 31% year is a hard year, not a business model ending, and two large publishers grew straight through it.
Add the disclosure Reddit is removing, and the bear has a case that does not require the licensing line to do anything at all.
The honest counter is that for Reddit specifically that link is asserted, not measured. Reddit does not disclose what share of logged-out traffic originates from search, and after this quarter it will not disclose logged-out traffic at all. Anyone claiming to know the size of the referral hit, in either direction, is guessing, and that now includes me.
What would change my mind
| Bullish | Bearish |
|---|---|
| U.S. ARPU growth holding above 40% for two more quarters | U.S. ARPU growth decelerating below 30% while user counts stay flat |
| International ARPU inflecting above $3.00 | Ad revenue growth falling below the mid-forties while ARPU growth stays intact |
| A licensing agreement renewed at a price that scales with usage rather than a flat fee | A licensing renewal signed at the same price as the last one |
| A ruling in the state case that survives a motion to dismiss on the contract theory | Another sequential decline in U.S. dailies |
The first line of each column is the one that decides between the two cases above. The second bearish line is the one I would watch hardest, because it is the tell that arrives before ARPU turns: Reddit running out of ads to sell while prices still look fine. Impressions grew 17% this quarter, and that is the number that has to break first.
The thing I would watch hardest of all is the one Reddit just stopped showing me.
view all sources here
Suggested Citation: Rettke, Sterling. “$RDDT Needs to Prove That it is More Than Just $GOOGL’s Social Network.” sterlingrettke.com, August 2, 2026.
Disclosure: I do not currently hold a position in $RDDT or in any of the comparison companies discussed ($META, $SNAP, $PINS, $APP, $TTD). This piece is for informational and educational purposes only and is not investment advice.
The content on this site is for informational and educational purposes only and does not constitute investment advice, financial advice, or a recommendation to buy or sell any security. Sterling Rettke is not a registered investment adviser. The author may hold positions in securities discussed. Always do your own research and consult a qualified financial advisor before making investment decisions.