Happy Friday. QQQs +68bps continued its relentless march higher hitting more ATHs, led as much by large caps like GOOGL ORCL APP and AAPL today, but also as speculative names continue to moon higher, with lots of 20%+ gainers in “undiscovered” small cap “AI plays.” Hard to stay on top of everything as seems like a new $1B+ mkt cap AI name pops up every day - things like Pony AI, which I found out out was public today. You know things are frothy when company logos look like this:
Speculative areas like Nuclear and Quantum leading the charge. TMT Momentum continues to lead the way up. Trading environment feels unlike even some of the more speculative environments we’ve experienced in the 20+ year we’ve been doing this….
Let’s get to the good stuff…
INTERNET
META -25bps as Bloomberg reported they’re in talks with ORCL for a cloud computing deal worth about $20B. Bloomberg: Under the multiyear deal, Oracle would provide the social media giant with computing power for training and deploying artificial intelligence models, according to people familiar with the talks. The total commitment amount may increase and other deal terms could still change before a final agreement…The HOT AI Summer continues (equinox is on Monday which means we transition to HOT AI Fall shortly…)
GOOGL +1% continues to new high as 3p data continues to show a 2-3 ppts beat for search
RDDT -2.3%: The main pushback I got on my “taking profits” call yesterday centered around the licensing deals. One investor told me: Even a small DAU miss this asset has become so strategic I can see it getting the cloudflare treatment on multiple. Hmm, maybe but I’m not convinced although the potential upside to licensing deals is the main reason why we’ve liked the stock so much recently, and a potential structure laid out in the Bloomberg article a few days ago could help alleviate DAU worries: It’s proposing a new kind of partnership that would encourage users to become active contributors to Reddit’s popular online forums, so Google traffic could help the company grow and generate content for future training…This is the main reason we continues to have a small position here and think path of least resistance is higher.
Still - the GOOGL deal was signed in January 2004 / OAI in May 2024 and Reddit said in their S-1 that “terms range from two to 3 years.” That’s not very specific so I did some digging with ChatGPT to try to nail down when exactly contracts come up for renewal, and it seems its more likely the deals will be renewed closer to ‘27 rather than early ‘26, which jives with CFO comment on last earnings call in late July that said they were “mid-flight” in deals. Here’s more detail from ChatGPT for all of us investing nerds reading this on a Friday evening:
Reddit’s 2024 10-K shows remaining performance obligations running into 2027 (recognize $114.6M in 2025, $113.2M in 2026, $25.1M in 2027), consistent with some deals extending ~3 years from early 2024.
In Jan 2024 Reddit signed data-licensing deals with aggregate value $203M and terms “ranging from two to three years.” They expected ≥$66.4M recognized in 2024 (which matches Q3+Q4 “Other revenue” of $33.2M each).
RPO at 12/31/24 (mostly data-licensing): $252.9M to be recognized $114.6M in 2025, $113.2M in 2026, $25.1M in 2027
The 2025–2026 RPO split (~$114M/yr) implies the full-year licensing run-rate on the longer contracts is ≈ $114M/yr. That’s very close to Google $60M/yr + “everyone else” ≈ $54M/yr (OpenAI + smaller licenses)
2027 RPO = $25.1M. If revenue is recognized ratably, that’s about $25.1M ÷ ($114M/12) ≈ 2.6 months of remaining recognition after Dec 31, 2026.
Base case: Both Google and OpenAI contracts likely expire in Q1 2027 (most plausibly Feb–Mar 2027).
Alternate (less likely): If one deal is 2-yr (to Jan 2026), then Reddit must have signed enough late-2024 adds to keep 2026 ≈ 2025 in RPO—possible, but the cleaner fit to the filings is both at ~3 years ending Q1’27.
Sanity Check: Reddit’s 424B4 said “two to three years” for the Jan-2024 cohort; the near-equal 2025/2026 RPO makes a pair of 3-year contracts more consistent than one 2-year/one 3-year (unless new 2024H2 licenses perfectly back-filled 2026).
The Anthropic suit jurisdiction is still being decided so unlikely we get a settlement (or licensing deal) there in the very near-term….The point of mentioning all of this?
Keep reading with a 7-day free trial
Subscribe to TMT Breakout to keep reading this post and get 7 days of free access to the full post archives.