No Deal is Safe as Meta Furiously Goes After Superintelligence
Scale's Data Promise • Lakers Deal • More Mindhuner • xAI's Burn • The Naked Gun
The news that Meta had tried to acquire Safe Superintelligence came out after I had already published some thoughts on Meta trying to "hackquire" a VC fund to bring on its two partners, Nat Friedman and Daniel Gross — but it obviously makes the story even crazier.
First and foremost, it’s seemingly a showcase of just how ruthless Mark Zuckerberg truly is. Ilya Sutskever turns him down, so he pivots to going after his co-founder and CEO of the company. Damn.
Second, if you thought the Scale AI deal was wild at just under $15B, what might a deal for SSI have looked like? Presuming the approach was after the fundraise in March which valued SSI at $32B, what could the offer possibly have been? It would obviously have to in the same "hackquistion" format, but what’s the number to try to convince them to sell? $50B? More?! What would Wall Street have thought about that? The stock didn’t move all that much with the Scale deal, but that insanely expensive talent deal has some other actual benefits for Meta — well, providing Scale’s business doesn’t entirely collapse after the deal (more on that below). But a deal for SSI would seemingly really just be for the talent and perhaps some early technology? Certainly not any revenue, let alone prospects for a real business!
Third, assuming Gross jumps ship, where does that leave SSI now? Again, they’re a startup valued at $32B that’s just one year old, now with its head cut off…
One more thing: Meta also apparently approached Perplexity as well before the Scale deal, as a late-breaking report from Bloomberg reveals. To say they’re on a mission seems to be an understatement here. And it’s sort of fascinating to contrast this approach with Apple, the other tech giant that needs to get its AI act together. Meta is going to do a half dozen deals before Apple does one.
💰 Meta Hackquires a VC Fund
"Team Superintelligence" continues to assemble at Meta, but it's getting even more awkward...
• Enjoying a Lucky Saint Lemon Lager (very good) 🍻
• Listening to "Man on a Mission" by The Black Keys 🎶
• Written on an M4 MacBook Air 💻
• Sent from London, England 🏴 Notebook
😬 Scale Makes a 51% Promise
Among the many strange things about the Meta/Scale AI deal was the fact that when it was announced, they didn’t really address the Meta data issue. That is, if Meta was acquiring 49% of the company, what would that mean for Scale’s other customers? And because it wasn’t the first, second, and third topic addressed, everyone assumed that it meant that as a part of the deal, Meta would have some level of proprietary access. That may still very well be the case, but this post aims to make it clear that this will not include other customer data. But everything from their obtuse title to the fact that it wasn’t written by the new CEO (or even the former one) is weird! It’s almost like it was written as a crisis comms response to the fact that not only has it been reported that Google is backing away from using Scale, but now OpenAI is as well — after saying the opposite immediately after the deal was announced. Assuming that those are, in fact, two of Scale’s largest customers, that’s obviously a big problem. Not just for the business, but for the signal it sends to the market. So much for not "icing" it! It now looks like OpenAI just torpedoed it. So these are the counter measures, I guess. But they can’t really counter the fact that Meta is acquiring every last share that they can without tipping into the majority. [Scale AI]
🏀 Los Angeles Lakers Sold for $10 Billion
On one hand, this is a number so large in professional sports that it’s hard to wrap your head around. On the other, it’s about what the two year old startup behind the "vibe coding" product Cursor is worth right now — and that valuation may be about to double (see: below). To stick to sports, this deal tops the previous record sale for a sports franchise set earlier this year, incidentally for the Lakers main rival (certainly in terms of NBA championships), the Boston Celtics for… $6.1B. This deal is 64% larger than the next closest deal. Ever. And that Celtics deal was wild because it topped the previous record which was set by an NFL franchise, which are generally considered to be the most valuable teams in sports. But the Washington Commanders sold for a "mere" $6.05B in 2023. On paper, the Dallas Cowboys had topped the most valuable frachises list for years and currently sit at… just over $10B. But the Lakers were only worth $7.1B on that same list so if we want to use the same multiple, the Cowboys are now worth nearly $15B. Back to tech, remember when Steve Ballmer bought the cross-town Los Angeles Clippers and everyone thought the price he paid was absolutely insane? The year was 2014 and the price? $2B. [WSJ 🔒]
🧠 The Hunt for More 'Mindhunter'
I’m going to chalk up the fact that I’ve seen this story about the possibility for Mindhunter coming back about 50 times in the past few days to the algorithms of the internet knowing me well. But Holt McCallany’s comments have gotten a lot of coverage and so the pump seems primed. …if David Fincher is, that is. Beyond being expensive, it’s clear that they never made a season 3 because Fincher had to salvage season 2 by doing much of everything himself. So if it comes back, McCallany says it would be in the form of three two-hour movies rather than a fully episodic show. This is clearly becoming more of a trend, especially at Netflix, with shows like Stranger Things that more or less morph into staggered-release movies — which is something I have long advocated for, including getting creative with the actual formats. Why does a "show" have to keep being done in the standard show format? Why not make a movie and then maybe go back to a show? Let the story dictate the format. Anyway, Mindhunter was great, so here’s hoping… [CBR]
🔥 xAI Burning $1B a Month
While it’s easy to think that all these AI companies are raising so much money simply because they can — and there’s certainly some truth to that, see: Anysphere below — it’s also in many cases because they need it, badly. Certainly that has been true with OpenAI as their servers continue to get lit on fire from usage. And per these numbers, it may be even more true with xAI, as they’re expecting to burn $13B this year. And if the company has $4B on their books, and they’re raising $9.3B (in equity and debt), you can do the math… They’re going to need to raise again next year and in fact, they’re already previewing that raise to investors looking at this round. The main issue for xAI is that while they’re battling OpenAI, Google, etc in training cutting-edge models, they don’t have the revenue to help offset the spend. This report says they expect to do just $500M this year, versus around $13B for OpenAI. But they also just merged with Xitter, which will obviously boost that number by a lot, but not enough to cover the burn (or is that $13B burn number including the Xitter revenue?!)! They’re also telling investors they aim to be profitable in 2027. How? Unclear. Something, something, Xitter data, or something. Another merger, perhaps? [Bloomberg 🔒]
Spyglass
🤖 Does Emo AI Dream of Mundane Tasks?
Is an emotional Alexa setting our timers really what we want here?
☎️ Dial 'T1' for Triumph
And you can't spell 'Triumph' without 'Trump'...
📺 Netflix Really Is TV Now
A major French network outsources their streaming to Netflix...
💬 You the User
2012 WhatsApp just eviscerated 2025 WhatsApp
Loose Leaf
Sam Altman confirmed that Meta — and Zuck in particular — is making $100M+ offers to key AI talent at OpenAI, but so far "none of our best people have decided to take him up on that." Which suggests that some non-"best" perhaps have? The counter is that his company is better positioned in the race for AGI and as such, may be more valuable one day. Also, maybe they’ll go after social too. Perhaps more smartly than Meta is so far with AI. [TechCrunch]
After 4 nomiations (3 for acting) over the past 35 years, Tom Cruise is finally going to get an Oscar. Albeit an honorary one. Which will be presented to him months before the Academy Awards. Which is stupid. You want to present your biggest stars with the award during that show. Obviously. [NYT]
Apparently not getting an honorary Oscar? Bob Iger, despite the weird, seemingly sycophantic lobbying by Kathleen Kennedy. All sorts of conflicts there with Disney potentially re-upping broadcast rights of the show, not to mention the Kennedy situation itself… [Variety]
The "hide spoilers" feature for Threads is a bit silly. Is anyone actually going to adhere to it? It feels like something that maybe the community could help with, noting things that are spoilers to be put behind the feature in feeds, but that would obviously get abused. So it’s probably just a cute feature that no one ends up using, at least not as intended. [THR]
I’m old enough to remember when Anysphere (makers of the "vibe coding" app Cursor) was valued at $9.9B. Which means I’m at least two weeks old. That price is now looking more like $18B to $20B. Though the interest is unsolicited and they may or may not jump at it. (Obviously they will. Things are too wild out there right now. You gotta make hay while the sun shines…) How long until Meta tries to acquire them? It has probably already happened? [Bloomberg 🔒]
OpenAI is pre-emptively warning that their new models are likely to carry a higher level of risk when it comes to things like um, biological weapons. They’re seemingly getting ahead of this to avoid the situation where Anthropic disclosed higher risk fears only after the release of their latest models, but also to make the case, once again, for why the US needs to lead in AI… [Axios]
Hidden in the iOS code is a reference to an as-yet-unannounced new Apple device, presumably the "HomePad"/"FacePod" thing. But it’s not even in the iOS 26 beta code but instead the iOS 18.6 code, which seemingly suggests the device is either near or was supposed to be near by this point — which suggests that it has in fact been delayed by Apple’s AI debacle. [9to5Mac]
It sounds like Edward Berger has already made his pitch to direct the next Bond movie — the first true Amazon Bond. But also in the running are Edgar Wright, Paul King, and… Denis Villeneuve, which would obviously be amazing. One name Matt Belloni doesn’t bring up here? Christopher Nolan. BUT he does mention his brother, Jonathan Nolan, is making his pitch… [Puck]
Also no word on the actual actor to play Bond yet, but this was an interesting comment from one of the rumored front-runners… [Deadline/Xitter]
It’s not just Oakley that Meta and EssilorLuxottica are planning for their next branded smart glasses, but Prada too, apparently (which has an eyewear deal with Luxottica). It’s not just about the high-end brand but also because their glasses tend to have thick temples, perfect for computing internals… [CNBC]
Midjourney, long my preferred AI image generator of choice, is now going down the video route as well… [TechCrunch]
The Fediverse integration with Threads continues to evolve... [Verge]
Which is good because the Fediverse itself still needs a lot of work — certainly from a UX perspective. [Tim Chambers]
Memo to Amazon employees: get busy getting smart on AI or get out. I mean, Andy Jassy’s letter is obviously a bit nicer than that, but that’s essentially the message, continuing the internal AI memo trend… [About Amazon]
When is a law not a law? When it involves TikTok, apparently, with another 90 day extension from the ban which was set to go into effect in January. [CNN]
I Spy...
When I first heard that Liam Neeson was going to be the lead in The Naked Gun reboot, I was confused as Neeson, as great as he is, is not exactly in the vein of Leslie Nielsen. The closet thing between them are their names! But as it turns out, this looks to be just as much a spoof of Liam Neeson movies — you know the same general "I will find you, and I will kill you" movie that he now makes about once a year — as it is of anything else. That’s a promising premise.