OpenAI writes Reddit a check for $70 million per year. Google writes one for $60 million per year. Both checks are for the same thing: access to 25 billion posts and comments that Reddit's users created for free over 20 years.

That is the business I am looking at. Here is what I found.

Reddit reported $663 million in revenue for Q1 2026 — up 69% year-over-year. Adjusted EBITDA margin: 40%. Free cash flow: $312 million in a single quarter. FCF margin: 47%.

To put that in context: this is a business that converted 47 cents of every dollar of revenue directly into cash. The best-run software businesses in the world — Constellation Software, Microsoft at its peak — operate at 30–35% FCF margins.

Reddit got there in one year. A year ago, this was a money-losing company. 2025 was their first-ever profitable year — $530 million in GAAP net income on $2.2 billion in revenue. FCF tripled.

The reason is simple: 91% gross margins and essentially zero capital expenditure. $1 million in CapEx in Q1 2026 on $663 million in revenue. When you do not need factories, trucks, or infrastructure to scale, operating leverage arrives fast.

This passes my compounder test. Reddit generates more cash than it reports in earnings — FCF ($312M) exceeded net income ($204M) in Q1 2026. That gap is partly stock-based compensation add-back (a real dilution cost I will address), but it is mostly driven by near-zero CapEx on a 91% gross margin business.

Most people think Reddit's moat is community. It is not.

Community moats are fragile — Discord, Telegram, and Facebook groups all compete for them. The 2023 API protests showed that Reddit's own communities can weaponise the platform against management.

The actual moat is the archive: 25 billion posts and comments, accumulated over 20 years, spanning every niche topic imaginable. Authentic human thought at a scale no competitor has and no one can replicate.

As language models flood the internet with synthetic text, the signal that trained those models — real people arguing, recommending, confessing, and explaining things to strangers — becomes scarcer. Reddit owns 20 years of it.

Reddit's CEO Steve Huffman said it plainly after Q1 2026 earnings: "There is no artificial intelligence without actual intelligence. And that comes from Reddit."

He also said the current pricing is too low. The Google and OpenAI deals renew in 2027. That renegotiation is the single most important event in Reddit's next three years.

Here is the math: OpenAI generates revenue from Reddit-trained responses that dwarfs the $70M/year it pays for access. If both deals roughly triple at renewal, licensing crosses $400 million annually — an additional $270 million of near-100% gross-margin revenue. At that level, licensing alone adds roughly $250 million to Reddit's annual FCF run-rate: 23% on top of the ~$1.1 billion annualised Q1 figure, before advertising grows a dollar.

That is not a prediction. It is what the renegotiation is worth if Huffman's leverage holds.

International ARPU: $2.31.
US ARPU: $10.79.

That 4.7x gap is not a problem. It is a roadmap.

Reddit currently operates direct ad sales in a handful of countries. In H2 2026, they are expanding to 35 new countries. Machine translation is live in 35 languages. International logged-out DAU grew 42% in 2025.

Here is the arithmetic: Reddit's roughly 500 million weekly users at international ARPU of $2.31 generate approximately $1.2 billion in annual revenue from international users. The same users at US ARPU of $10.79 would generate $5.4 billion. That gap — $4.2 billion in latent annual revenue — closes over years as direct ad sales expand country by country. That is not a prediction either. It is the gap on the table.

1. Google traffic dependency. Sixty percent of Reddit's daily active users arrive via Google search. If Google's AI Overviews start answering queries without sending users to Reddit, that traffic disappears. Google pays Reddit for training data — but that is a separate deal from the one that sends traffic. The interests can diverge.

2. Advertising concentration. Ninety-four percent of revenue is advertising. Reddit is a third-tier ad buy for most brands. In a recession or ad budget freeze, Reddit gets cut before Google and Meta do. This platform has never been tested through a real advertising downturn at scale.

3. Regulatory. In February 2026, the UK fined Reddit £14.47 million — the largest children's privacy penalty in UK history — for inadequate age verification. An FTC investigation into whether licensing user-generated content for AI training is an unfair trade practice is ongoing. EU GDPR exposure is unquantified.

Reddit is genuinely different from most of what I evaluate. The 91% gross margins and near-zero CapEx make it one of the most capital-efficient businesses I have looked at. Going from loss-making to 40% EBITDA margins in twelve months is not normal.

International ARPU convergence is one tailwind. AI licensing repricing is another. Most businesses get one. Reddit has both in front of it.

The caveat: FCF overstates shareholder economics because Reddit has significant stock-based compensation. SBC is a real cost — it dilutes you. Adjust for that before getting too excited about the 47% FCF margin.

The Google traffic risk is the tail risk I keep coming back to. Reddit's logged-out traffic is 60% of its user base, entirely dependent on Google continuing to surface Reddit in search. Google has both the motive and the ability to change that overnight.

At $32 billion market cap and 11.8x EV/Revenue — 21% below its own historical median — this is not obviously cheap or obviously expensive.

This goes on the watchlist. The 2027 proof point: if AI licensing renegotiates above $300 million per year, the thesis is confirmed and the stock is mispriced at current levels. If it stays near $130 million, the "irreplaceable archive" story is less valuable than the CEO is claiming.

If you are evaluating any business right now, run the same test I ran here: find the asset no one else can replicate, check whether management is monetising it honestly, and ask whether the growth runway is longer than the price implies. For Reddit, the archive is the asset. 2027 is when we find out what it is actually worth.

Buying trigger: licensing closes above $300M/year in 2027, or a 25%+ drawdown from current levels (~$160).

I will cover the renegotiation the week it lands. Follow The Investing Show on YouTube — that is where I publish all portfolio and watchlist updates.

Reddit is currently on my watchlist, the 2027 licensing renegotiation is the deciding variable. The gap between $130M and $400M in annual licensing revenue is exactly why I am watching before buying.

wanted to discuss valuations and opportunities live with us? Join 256 investors and entrepreneurs from Los Angeles to Singapore: www.strategysprints.com