M.G. Siegler β€’ β€’

Signal: End of the 2025 Line πŸ“§

TikTok Signs β€’ Waymo $100B β€’ OpenAI $830B β€’ FIFA's Netflix Game β€’ Yann Lecun Raising β€’ Paramount Rejected Again β€’ CoreWeave Scale β€’ Buzzing the Watertower

Twas the week before Christmas... and I'm emptying my notebook below before I'm back on the road for a couple weeks. Writing here and there, as always. Including looking back at last year's predictions – trying to give Apple a few more days to pull off #1! For now, there's a lot below. Happy Holidays. 🍻


The Inner Ring...

Conflicts & Interests in AI
*Of course* Amazon is investing in OpenAI, next up…
The Grand Netflix Hollywood Unification Theory
Warner Bros/HBO is phase one of Netflix’s bigger play here…
The AI Chaos Ladder
Where do the leaders in AI fall?

I Note...

πŸ“Ή TikTok Signs Away US Entity – Technically, the deal still isn't done. It's set to close on January 22, 2026 and presumably China (or the Trump administration) could still kill it in some way. But presumably TikTok wouldn't have signed without China's blessing – or assurances China would give the final sign-off? I mean, who knows, this whole situation has been such a farce. Including the "around $14B" valuation – I often joke about big deals being less than AI seed rounds these days, but this one truly is! It's also a little odd that an Abu Dhabi-based fund is one of the key backers to ensure American control of the entity? It's also still not clear which "new investors" own the mysterious other 5% – is it the Murdochs? The US Government? Oracle is clearly the lead here, so it will be interesting if that in some way gets roped into the Netflix/Warner Bros/Paramount love triangle – does the ownership of the father mean anything for the ownership of the son? What if the father is backstopping said ownership? With shares in said buyer of TikTok? It's yet another wrinkle that could be in Netflix's favor here... But the main question remains: is TikTok really still going to remain the hot consumer product when owned by a nearly 50 year old tech company (one in turbulent waters given their AI reinvention efforts, no less) and private equity funds? Where's 'TikTok 2'? [Axios]

πŸš™ Waymo Worth Way More Than $100B – I mean, at least in my own estimation, just based on our current market environment (Tesla, Uber, etc) and not to mention that AI startups are closing in on such valuations with little to show for it – certainly not traction and/or revenue like Waymo has! Then again, Alphabet is getting others to validate the valuation, which is undoubtedly part of the very reason for the raise (as Google could clearly just fund this themselves). And they sort of have the burden of showing actual revenue – $350M ARR – versus even Tesla, which still has a very much in beta self-driving taxi program. Still, there's a whole future-of-transportation/magical experience narrative that they could spin here (and would, were Elon Musk running the company, you imagine – though he would also say, as he often does, that Waymo's approach isn't scalable in the way Tesla's will be) where $100B is an absolute steal. For now, maybe they're just fine to fly relatively low-key under the lidar. [Bloomberg πŸ”’]

πŸ’° OpenAI Worth Way More Than $800B – By which I mean they're apparently targeting an $830B valuation in a new would-be fundraise. That's a curiously precise number for a famously imprecise business, unless you consider that they really perhaps just want that number to be higher than $800B – the number at which SpaceX is now valued, of course. And when earlier reports suggested the valuation could be a mere $750B – the old pre versus post switcheroo? – that simply couldn't stand. So who is investing? I mean, who is left? Amazon, for one, it seems. Undoubtedly more sovereign wealth funds for another. Would Apple revisit? Would Google dare?! Nothing can be ruled out. Good on Disney for squeezing into the old $500B valuation – with warrants for more, no less. 🎢When you wish upon a paper gain...🎢 Speaking of, all of the SoftBank money from the last fundraise isn't even in the bank yet. This is true perpetual fundraising. And clearly necessary! But it will also undoubtedly push up the targeted IPO valuation (as I predicted a whole month ago). Maybe OpenAI could buy Waymo with this fresh $100B? [WSJ πŸ”’]

⚽️ FIFA's New Netflix Game – There's been a bunch of Netflix gaming news in the past several weeks that's now completely buried under the weight of Warner Bros. But the initiative will clearly be key if Netflix wants to become the "everything app" for entertainment – and the first $1T media company – which is why they keep trying to pivot their way into the right model/experience in gaming. FIFA is an interesting play here because it was a beloved franchise (though not exactly the best brand!) that vanished over a licensing dispute with EA (which obviously kept going with new 'EA Sports FC' branding). The timing of the return, of course, is good with the World Cup next year. But can an unproven developer (that Netflix is partnering with) really nail this on the first shot? (They're also co-developing the first new James Bond game in forever due in March.) If nothing else, the price is right (free, for Netflix members). [BBC]

πŸ’° Yann LeCun is Raising Money – Two key tidbits: 1) the company is called Advanced Machine Intelligence Labs (AMI Labs) – doesn't get more generic than that – 2) they're targeting raising €500M at a €3B valuation. That almost seems quaint and shows some restraint in this day and age! Still, it presumably will make LeCun a billionaire, or close to it, on paper. Sort of interesting/fun to think about in the context of what Zuck has been dishing out to AI talent and what he clearly wasn't dishing out to LeCun. This report reiterates (as LeCun has stated) that Meta won't be invested in the lab but there is still some sort of "partnership", which continues to sound like PR for both sides post-breakup. Also, LeCun has brought on a CEO named LeBrun, that's just LeFun. [FT πŸ”’]

🎬 Warner Bros Discovery Rejects Paramount's Takeover Bid – No surprise, but the two most interesting tidbits are how WBD felt misled by Paramount as their bids kept morphing to make them more attractive, and how they don't believe Paramount is actually good for the money – even though it's being backstopped by Larry Ellison. They don't trust the revocable trust (where Ellison's Oracle shares reside) element. It can't help that those shares are in a free-fall as investor concerns mount over Oracle's AI spend and the debt required... Paramount hasn't raised the offer thus far and instead seems intent on taking it to the shareholders for a vote. And while some of those investors are open to Paramount's offer, it feels like there are an increasing number of things starting to work against it – beyond the Oracle concerns, the YouTube/Oscars deal may oddly help Netflix's regulatory case, the fact that hedge funds are now circling the WBD TV networks (so there's clearly outside interest in those), Jared Kushner oddly dropping out of the consortium may not matter monetarily but seems like it will optically for you-know-who, etc. It's worth listening to RedBird Capital's Gerry Cardinale for the counter-arguments on Matt Belloni's podcast The Town. He truly believes it will come down to how those TV networks are valued, and that the WBD/Netflix would be implicitly wildly overvaluing them (versus their offer). It is interesting that no one is talking about the potential Netflix stock upside in this deal... Feels like they should? All eyes turn to January 8, the deadline for investor tenders... [NYT]


I Quote...

"The bull case is that they’ll scale into it, and that a lot of companies have low margins to start, but this is a company at scale. There is no scaling going on here."

– Gil Luria, an analyst with D.A. Davidson, giving his thoughts on CoreWeave – a company which he says has the "ugliest balance sheet in technology, by far."


I Wrote...

YouTube Hands the Best Acquisition Oscar to Netflix
With YouTube poaching The Academy Awards, it sure feels like Netflix is going to be allowed to buy Warner Bros now…
ChatGPT Starts to Break the Bounds of Chat
A necessary product evolution is underway.…
Congrats Regulators, You Killed Roomba
Actually, you just put a Chinese robot in every home…

Asides...

  • Warner Bros aside, Netflix continues to be fast and furious with their video podcast signings, cutting deals with both iHeartMedia and Barstool this week. The audio will continue to play anywhere, but the video is Netflix-exclusive. Sorry, YouTube... [The Athletic πŸ”’]
  • One would-be podcaster not going anywhere? Howard Stern. After all that, he's sticking with SiriusXM. It's a shorter deal (three years) and he'll undoubtedly be cutting his workload again. But no fun Apple/YouTube/Spotify/Netflix deal. [THR]
  • Meanwhile, also undoubtedly in response to YouTube, Instagram is putting Reels on your TV. First via a Fire TV app. Yes, this seemingly also helps to make the case for why Netflix should be allowed to buy Warner Bros. [THR]
  • Is Blue Owl backing away from the data center project in Michigan a blue-owl-in-the-coalmine situation, or just specifically about Oracle's debt load? Or both? It's seemingly the first sign of prudence in the market – this is the key partner for the first Stargate and Meta's 'Hyperion'. [FT πŸ”’]
    • Speaking of data centers, I've been meaning to link to this fantastic visual reporting piece about the issues with power demand. [FT πŸ”’]
  • Thinking Machines Lab is planning to release their own in-house developed models next year – you'd sort of hope so given the rumored $50B valuation they're seeking, but also what they've done to date, product-wise (helping others create custom models, seemingly distilled). [Information πŸ”’]
  • I have two questions on the news that the parent of Truth Social is merging with a nuclear fusion startup. Both of them are "what?" [NYT]
  • On one hand, the new app directory in ChatGPT is pretty straightforward and simple. On the other, OpenAI has tried to launch an "app store" before. Everyone keeps trying to launch app stores, but will users come? [Verge]
  • Meta is readying their "Mango" AI model for image and video generation. Wonder if they'll go with these names outward-facing to try to match/copy "Nano Banana" (and what OpenAI should have done with "Strawberry"). [WSJ πŸ”’]
    • This is, of course, tied to "Avocado", the likely closed new non-Llama flagship model, ironically distilled from other "open" models. Oh, Meta. [Bloomberg πŸ”’]
    • Also in the internal Q&A that was leaked (as always), Alexandr Wang said they're exploring "world models" which is weird because they just parted ways with the high-profile guy leading that charge...
  • Seems like we can add "continual learning" to the AI terms that will be oft-repeated in 2026, alongside "sparse attention", "reinforcement learning" (nothing new, of course, and perhaps the key learning from Deep Seek a year ago, alongside "MoE"), and yes, of course, "world models". [Bloomberg πŸ”’]
  • Do you want to know seemingly every product in Apple's pipeline for the next two years? Well, here you go thanks to yet another massive software leak. [MacRumors]
    • This apparently involved a prototype iPhone that was sold. Which had an internal build of iOS with secret in-the-works device codenames included. Yikes. [MacRumors]
  • The teaser for Disclosure Day, Steven Spielberg's fourth alien-related movie looks... I'm honestly not sure. But I appreciate how vague it is. Is there some connection to Close Encounters of the Third Kind? No one will say... [THR]

I Spy...

Yeah, it's getting harder to see, quite literally, how Paramount Skydance is getting this deal. Certainly neither Netflix or Warner Bros Discovery think that right now!