2025-12-30 20:31:09
Between Apple and Meta, I think it was a pretty close call in terms of which Big Tech company had the worse 2025 when it came to AI. Both had to completely overhaul their approaches after embarrassments. But only one has had to spend billions, let alone tens of billions, and perhaps hundreds of billions to correct the mistakes – at least so far. And so Meta clearly wanted to end their no-good-very-bad-year on a high note. As Angel Au-Yeung, Raffaele Huang, and Kate Clark report for WSJ:
Meta has agreed to acquire AI startup Manus, a Singapore-based company with Chinese founders that conducts deep research and performs other tasks for paying users.
Meta is closing the deal at more than $2 billion, according to people familiar with the acquisition. Manus was seeking a fresh round of fundraising with a valuation of $2 billion when Meta approached the startup, some of the people said.
Given that Manus is apparently well past $100M in ARR and still growing fast, such valuations seem almost reasonable. Then again, the Chinese origins of the parent company, Butterfly Effect, have long lingered over the company, presumably even with an official move to Singapore. As such, Meta perhaps smelled an opportunity. One that helps their AI narrative, at least from a business-perspective, immediately.
While their Big Tech peers, Microsoft, Google, and Amazon are all monetizing their AI initiatives both through upselling their cloud offerings or by selling the AI products directly to customers, Meta is doing neither. At best, they're indirectly monetizing their AI by leveraging it to bolster their existing businesses in ads – but those others are doing that as well. So it's sort of a tough sell to Wall Street, if nothing else, especially when your AI spend is on par with those players – or even higher when it comes to things like compensation.
And unlike the nearly $15B Scale deal, where a "hackquisition" saw them acquire only 49% of that business, Meta gets the whole enchilada here. That is to say, they're buying the actual business, which gives them an AI business overnight:
Meta says it plans to continue to operate and sell Manus’s service and integrate it into its suite of social-media products. Meta has previously touted so-called open source models that are largely free to access, modify or distribute.
“We plan to scale this service to many more businesses,” Meta said in its announcement about the deal.
Translation: Meta just bought into the enterprise market! Now, whether or not other businesses will want to buy enterprise offerings from Meta is another matter. But I suspect keeping the separate Manus branding, with the Meta firepower and resources behind the scenes, could help them here. If not, they'll probably face the same challenges that others trying to break into enterprise sales run into, from Google on down. Selling into enterprise is just a different beast which requires different muscles. It took Google bringing on board Thomas Kurian from Oracle for this to truly work (with the necessary growing pains along the way).
With that in mind:
Manus’s co-founder and chief executive, Xiao Hong, who often goes by the nickname “Red,” will report to Javier Olivan, chief operating officer of Meta, some of the people said.
When you hear that Meta is doing another massive AI deal, you might assume it's to bolster the "Superintelligence" group under former Scale chief Alexandr Wang, but again, this deal is different – it's a deal to create a new business arm of AI, so reporting to the COO makes sense.
At the same time, the tech Manus has built can obviously help Meta's other AI efforts, including across their consumer-facing products. And Meta undoubtedly wanted to make that clear by having Wang tweet his excitement about the deal, lest you think this is another sign of internal turmoil within Meta.
So this deal seemingly makes a lot of sense for Meta on a few fronts. And it also may point to the start of a push into enterprise. Again, easier said than done, but don't be shocked if this is a wedge of sorts. If they can keep Manus expanding into businesses, we should see other Meta cloud offerings follow, putting them more in line with those aforementioned Big Tech peers.1
They have, of course, tried this before in ways – from Workplace on down – but it hasn't really worked. But buying up a hot product, team, and tech is what Meta does best! Just ask the FTC...
One more thing: As for those China-ties, consider them completely severed. As Yifan Yu and Cissy Zhou report for Nikkei Asia:
Meta says it will acquire AI startup Manus and that the Singapore-based company will no longer have any Chinese ownership or operations in China after the deal, while a source close to the matter put the value of the transaction at between $2 billion and $3 billion.
What at first looked like a questionable decision by Benchmark to invest in Butterfly Effect given the geopolitical risk, METAmorphosized into a savvy bet and quick exit.
1 Moves which OpenAI is already talking about making as well, clearly as a preview to help justify their own AI spend...
2025-12-29 23:44:24
Last year, while trying to kill time waiting to ring in the new year, I jotted down ten predictions that I thought might come to pass in 2025. Actually, I thought most of them wouldn't, but I thought they could. Here we are with only a few days left in the year, so let's check back in, before jotting down some new ones for 2026...
Apple buys an AI company – As we enter 2025, the consensus view on Apple's initial foray into AI seems to be indifferent at best and underwhelmed at worst. As Siri herself starts to get real upgrades in 2025, that could change a bit. But Apple is running a race quite slowly – too slowly – behind folks who are doing laps much faster. One way to leap ahead would be to buy a player. Apple famously doesn't do a lot of splashy M&A but perhaps an exception could be made for a Mistral or a Perplexity. Who knows if Apple would even be allowed to buy the former given their war with the EU. But the latter could be interesting in particular in light of the impending end of the Google default search deal/payments...
My first prediction – the one I was probably most confident in! – is the one that absolutely did not happen. To be fair, I do believe that Apple bought up some AI talent, but I was also clearly thinking this would be a larger deal. And as we got into 2025, it became clear to many people why Apple might want/need to do such a deal. And Apple apparently thought about it too – hello, Perplexity – but ultimately decided against it; something I eventually talked myself into as well – especially given the prospects of a Google partnership, which should come into play in early 2026, to bolster Siri. Finally. Anyway, smaller deals and simply thinking about doing larger ones don't give me a passing grade here. Grade: D+
Someone buys Warner Bros Discovery – David Zaslav clearly – clearly – wants to do a deal. But the music is slowing down and other players are already finding seats to the point where it's looking more like he won't be able to find a partner to buy as much as he'll be the one selling. Perhaps it's just some assets – CNN and/or HBO? – or maybe it's the whole thing. If the M&A environment truly is about to thaw, Comcast or someone else could make a move. Might Paramount aim to bulk up even more post-Skydance deal? And of course, I'm always trying to come up with ways for Apple to spend money, per above. It does make some sense...
While I guessed that it might be Skydance, fresh off of the Paramount deal – a good guess, if I do say so myself! – it was actually Netflix that bought Warner Bros. Okay, technically the deal hasn't closed. And probably won't for months – if not years. I mean, it obviously may never happen. Or Paramount Skydance may yet get that last laugh. But still, at least right now on paper, I nailed this: Grade: A
Intel gets bailed out – We'll see what ends up happening with the CHIPS Act under the new Trump administration, but regardless, I have a hard time believing they're going to just let Intel wither on the vine. Enabling/bolstering a true American player in chips is a win, no matter the political party. Could a new administration help broker a deal to get, say, Microsoft to support the rejuvenation of Intel? First things, first, Intel needs to put in place a new leadership team. I'll keep my (early) money on Lip-Bu Tan here – if he can overhaul the entire board as well...
We can argue the definition of a "bail out" I suppose. But at the highest level, clearly the government not only investing in Intel, but becoming their largest shareholder, seems like it should qualify for my notion. Further, that support clearly got others on board as well – both SoftBank and NVIDIA with money, and rumors of Apple and others looking at Intel for their manufacturing needs. And I think I get some bonus points for correctly predicting Lip-Bu Tan as the next CEO of Intel. Grade: A-
Elon Musk bails on the White House – I'm a bit torn on which side will sour on the other first here, but I'll go with Elon getting fed up with all the in-fighting – which we're seeing already, of course – and bureaucracy. He'll say the whole thing is fucked and go back to trying to bring about change in his more insurgent manner, by shitposting on Xitter. There are many reasons for him to hang on as long as he can – billions, in fact – but perhaps he'll pre-empt a move by Trump here. There can be only one main character, after all. And he only shares the spotlight for so long...
The only issue here remains the question of who dumped whom in the breakup which was utterly predictable. "Vibes" suggest it was more Trump (or his inner circle) that got sick of Elon, but there was obviously annoyances the other way as well. (And clearly, Tesla shareholders wanted him back and focused on at least one of his companies.) After Elon went nuclear a few times on Trump on Xitter, and Trump responded with reminders of who has the real control, the two seem to have reached some sort of détente. Grade: B
Amazon's Alexa overhaul proves less than "remarkable" – Not exactly shocking given all the reports and the fact that that project has clearly been delayed multiple times already. But I just have this feeling that Amazon is going to have a hell of a time trying to convince the world that a new Alexa is above and beyond the old Alexa – let alone worth paying for. Their previous success with the Echo devices is their undoing here. Might it push them ever closer to Anthropic as a result? The pieces are already in place for that relationship to be the new Microsoft/OpenAI marriage...
Yeah, I mean, after a grandiose unveiling that reeked of over-hype, it took forever for Amazon to actually roll Alexa+ out to a meaningful swath of the user base. Once they did... the silence is sort of deafening. I have it now, it's... fine. A less capable version of ChatGPT or Gemini mixed with a more capable Alexa (though in some ways, the new functionality is predictably less capable, or at least stable). Still, it's nice to have conversational AI scattered in some rooms around my home (because it's still not on every Alexa device). But not the game-changing transformation that had Panos Panay getting emotional. As for that Amazon/Anthropic relationship, it's getting... complicated in ways that really does have echoes of Microsoft/OpenAI! Grade: B+
Microsoft and OpenAI kiss and make up, or break up – It all comes down to the negotiations to turn OpenAI into a for-profit company. Assuming this happens, and Microsoft gets a massive (actual) equity stake as a result, I think that plus the continued floundering of their own consumer AI efforts will lead to a re-kindling of the relationship. Oh yes, and the fact that OpenAI still needs another $20B+ or so to make their AGI dreams come true. SoftBank and the sovereigns can help, but they still need a true American Big Tech™ partner to compete with Google and DeepMind here. If/when the for-profit status comes into effect and the relationship is rekindled, OpenAI should shoot past a $200B valuation in new tender offers. If the rift instead deepens and Microsoft and OpenAI break up in the name of independence, look for the former to be involved in a deal to buy TikTok...
This is a tricky one to grade since they technically did kiss and make up in order to seal OpenAI's deal to transition into a PBC – thus securing Microsoft's 27% stake – but it's not like they're fully back in love. Instead, Microsoft made their moves early in the year to pretty much fully outsource any OpenAI growth commitments to SoftBank. At the same time, this frees up Microsoft to more fully and openly compete with their partner, both with their own internal push for AGI/Superintelligence but also by partnering with other OpenAI rivals like Anthropic. Still, it's not like it's a full break-up of the relationship. Again, Microsoft owns a massive 27% chunk of OpenAI. It remains... complicated. Anyway, having just unwound one insanely expensive deal, it didn't seem like Microsoft was in a hurry to jump into the TikTok deal (again) – though they seemingly did debate it! (I would also just note that it's also not fully clear who the owners of that mysterious 5% equity slug are yet...) Anyway, mainly I'm annoyed for drastically underselling OpenAI's would-be valuation. $200B? They shot past that one month into the year. Try $500B now. With talk of new money coming in at $850B! Grade: C+
NVIDIA comes back to Earth, a bit – Overall, there's almost no way that Blackwell isn't a massive hit for the company. To say there's pent-up demand undersells it, and these will certainly not be undersold! But I also wouldn't be surprised if NVIDIA's overall numbers are more muted in 2025 than people may think due to supply and complexity issues. But more so I'd chalk this prediction up to the law of large numbers simply catching up with NVIDIA. They can't keep growing as they have, so simply awesome growth will have to do instead of bonkers growth. Also, what if they're not as dominant in AI inference as they have been in pre-training? And, of course, everyone – including all of their major partners – are now sprinting full speed to try to diversify away from the reliance. Still, that's probably much further off than 2025. Again, this year already seems pretty signed and sealed...
Another tricky one to grade since it's both right and wrong. From a market cap perspective, NVIDIA has fallen a bit – from the $5T highs down to a "mere" $4.6T to end the year – but they're also well ahead of where they were a year ago (and still ahead of everyone else). The earnings remain great – they even managed to reaccelerate last quarter! – but yes, investors are getting a little more cautious on their numbers given the above (natural) issues. And mainly, there's growing concerns about the "AI Bubble" in general. The shorts are out! But the main "vibe" I think I got correct here is around Google's TPU chips as an up-and-coming challenger. They're not going to displace NVIDIA's place at the top any time soon, but they could eat into some of the business. And yes, their last-minute $20B Groq deal seems to be about both snuffing that out while also ensuring they don't lose out on the growing inference wave. Grade: C+
Threads passes Xitter in active users – This seems to be more a matter of "when" not "if" at this point. But while early models suggested it may take a few years, Meta is seemingly getting more aggressive when it comes to growth. A few more turns of their dials and I bet they pass Xitter this year. Bluesky will continue growing too, but I don't think they will be able to compete with the growth engines and talents (and cloning capabilities) that Meta possesses – even if they're making more of the right moves in the space...
This sort of happened. On mobile, by some metrics, Threads is now bigger than Xitter at least in terms of DAUs. By other counts, it's very close. But on the web, it almost certainly is not. Overall? It's hard to say for sure, but most estimates still have Xitter ahead by around 100M (MAUs) or so. Call it 500M to 400M (maybe as much as 600M to 350M or maybe as little as 450M to 400M). Close, but no cigar. Also, it seems like Xitter may be growing again, so it's a matter of if Threads can keep up the faster growth rate in 2026 to fully overtake them. Will the merger with xAI help or be a distraction? Is this whole thing just gonna be a part of Tesla and/or SpaceX one day anyway? As for Bluesky, that prediction seems accurate: still growing, but not nearly at the rate of Threads. Grade: C
Google starts to feel real pressure on search – This is a tricky prediction to make because everyone makes it every year – and has for years. And yet it's never true. But I do think that the various AI players may finally start to make some inroads here in 2025. And that could also be related to the end of the default search contracts – that will undoubtedly be tied up in courts for years, but just the threat of it may open some players – like Apple – to start to look at other options here. Perhaps this is just anecdotal, but I'm using ChatGPT for a lot more queries these days... Of course, Google could combat this with a bold placement for Gemini...
Even though Google continues to maintain that Search is healthy and there's "nothing to see here", simply watching their actions – or listening to Apple exec Eddy Cue, under oath, no less – would seem to suggest otherwise. 2025 was the year that Google pivoted hard to shove AI into basically every surface of Search. And it seems to be working – as OpenAI's "Code Red" at the end of the year is testament to. While the judge in the antitrust case altered the details of the default search deals a bit, he ultimately allowed such deals to stay in place – a huge win for both Google and Apple (and Mozilla). Why? Largely thanks to the looming threat of AI disrupting everything, of course. So yeah, this one feels mostly correct. Grade: A-
Mark Zuckerberg unchained – While it all felt a bit too calculated for my taste, the "Zuckaissance" of 2024 was nevertheless impressive. But come on, this is PR 101. The media builds someone up to then knock them down. Zuck has lived this cycle a few times now and he's due to be knocked down again after a year of the press eating up the gold chains and wakeboarding. Not even sucking up to the incoming president can stop this inevitability. It's just a question of what causes the backlash. Something with Llama? Instagram? WhatsApp? Good old Facebook? Something political? Maybe flooding all our feeds with AI bot nonsense? Anyway, this knock down needs to happen in 2025 to get ahead of the next Zuck build up as the "Orion" AR glasses get closer to launch...
It took a whole week for the 10th and final prediction for 2025 to start coming true. After a 2024 in which Mark Zuckerberg seemingly got endless love in the press and was riding high, 2025 was indeed the knock-down, wake-up call. Beyond the blowback he got (from both sides) for his political wishy-washy-ness, he also got served a huge piece of humble pie in the form of his AI ambitions for Meta. Once Llama 4 failed to compete with (let alone overtake) their main competitors in AI, Zuck had to spend billions to blow up their efforts and start over. Maybe it works, but I'm skeptical. Regardless, it's a bad look when you spent the previous year talking up the importance of "open source" AI, then you basically throw that out the window to bring on a team aspiring to build behind closed doors – which is also seemingly causing a lot of internal strife, as you might imagine. But hey, could have been worse: they could have lost their own antitrust case and been forced to sell/spin-off Instagram and/or WhatsApp. But the FTC's case was so weak that the judge sort of laughed it off. As a result, Meta's stock, which ended up slightly for the year, could have fared far worse – especially with Wall Street growing skeptical of their CapEx spend in particular. That is, of course, also why we're seeing Meta start to pivot away from the metaverse. Awk. Grade: A
One more thing: I also had a bonus bet...
One more thing: My bet would be that after some strong initial buzz amongst the Apple diehards – including myself – Apple's new 'FacePod' home hub thing fails to catch on in a major way as well. Beyond people who buy everything Apple – again, that's me – it's not clear what it will be used for. Though I'm not sure how much it will matter as it's just an initial attempt to enter the home in a major way (well, beyond HomePod, I guess), but still, if it's not a hit out of the gate, there will be pressure on Apple following the fizzle of the first Vision Pro as well...
It's true that Apple's 'FacePod'/'HomePad' failed to catch on – because they never shipped it. It was seemingly delayed as a result of the AI fiasco and should instead come early next year. I'm still skeptical of the device, so let's call this a "push" for now.
Overall, ChatGPT tells me my GPA would be 2.96. We'll round up slightly for a solid B on the year. I also went ahead and published a page with a running tally of predictions throughout the year. Well, the correct ones, at least 😝. On to 2026 🥂.
M.G. Siegler
M.G. Siegler
2025-12-28 19:21:20
This is the way season one ends. Not with a bang but... the ominous sign of a bang.
I enjoyed season one of Pluribus. I wouldn't say I loved it,1 but I found it to be an interesting change of pace, quite literally, from many of the other shows currently on TV – by which I mainly mean streaming, of course. It's sort of wild to me that it's apparently Apple's most popular show yet, at least by some early measurements, because it certainly doesn't seem like the kind of show that would be wildly popular. Then again, it's all relative and especially with Apple TV – the artist formerly known as 'Apple TV+' – as their most popular shows still tend to not be nearly as popular as the most popular shows on other streamers, simply given Apple TV's relative size.2
Anyway, overall I liked it. There are some mild Breaking Bad vibes as you'd expect from Vince Gilligan,3 but a lot of that is mainly in the setting and subtly comedic tone. A few of the episodes have some major pacing issues in my book, but I understand that this is where people will yell about not appreciating the art. I mean sure, I get that, but I personally found it weird to veer from a science fiction tone to a full-on zombie apocalypse tone to a muted – again, quite literally – quest for groceries to a vision quest in the jungle to a sort of Eat, Pray, Love – a romance with a robot that predictably morphed into Sleeping with the Enemy...
I loved the extremely contemplative Train Dreams – it's probably my second favorite film of the year after One Battle After Another – but that had the same general vibe throughout. I found it hard to find my bearings with Pluribus.
So I went on my own side quest, as it were, trying to look for clues as to how the show could be an allegory for our Age of AI. I realize this may be problematic simply due to the timing of when Gilligan was writing the first season, but still... I swear there are signs.
Earlier today, John Gruber wrote about Tim Cook's seemingly odd and perhaps tone deaf tweet about the season's finale. Noting that, "Tim Cook Posts AI Slop in Christmas Message on Twitter/X, Ostensibly to Promote ‘Pluribus’", Gruber notes the inconsistencies in the image, which seems especially odd when you consider the Apple TV account's retweet of Cook's message:
The Apple TV X account retweeted Cook, and added a credit: “We thought you might like this festive artwork by Keith Thomson, made on MacBook Pro.”
Apple didn’t tag the “Keith Thomson” who supposedly created this artwork for them, but if it’s this Keith Thomson, Apple must have somehow fallen for a scam, because that Keith Thomson’s published paintings are wonderful. The “K Thomson” signature is at least sorta kinda like that Keith Thomson’s signature on Apple’s sloppy illustration — but not really the same. (I like a bunch of the paintings from that Keith Thomson, and love a few of them, but this one in particular feels like it was made just for me. It’s perfect.)
This seems far too obvious to be a mistake/blunder on Apple's part. And the actual full message I think makes it clear this is some sort of commentary. This isn't a message and image for you or me, it's for...
Hi, Carol.
We thought you might like this festive artwork by Keith Thomson, made on MacBook Pro.
It reads as Apple channeling their inner "They" to suggest that Carol might like a new work by a famous artist. Of course, in the universe of the show that artist is presumably one of "Them" so he would be incapable of creating something wholly new and original, but instead would perhaps give an output based on the collective knowledge of Thomson's work. Would They/Them hallucinate some strange inconsistencies as the AI still does in our real world? Who knows – because we don't fully understand what causes them in our actual models because we don't fully understand the outputs.
Of course, this may break down if you consider that They/Them don't refer to themselves as any one individual. So unless Keith Thomson is somehow one of those on Earth not infected/turned (which seems impossible given what we know about the remaining people), it's weird they'd call him out. Then again, we know that Carol prefers that They/Them still refer to themselves as individuals, as she makes clear to Zosia. So perhaps this is just the collective trying to send Carol a Christmas message they think she'd appreciate.
Another interpretation might be that They/Them are actually using AI (on that MacBook Pro) to create the image. And maybe they're even having the body of Keith Thomson create it, so technically it is "by Keith Thomson".
Keith Thomson using AI to produce art that's like Keith Thomson's art because it's trained on Keith Thomson's art. How's that for a mindfuck?
I'm sure I'm reading wayyyy too much into that tweet (and retweet), but given my previous post about Pluribus as an AI allegory, I think it's sort of interesting to think about in that context. The reality is that multiple things may be true, Gilligan could have written the show without knowing it would be a commentary on generative AI, but as that technology came into focus, and as they started making the show, perhaps he's leaning into that theme. There are undoubtedly broader themes too about hive minds and whatnot, but... I'm sticking to my story for now.
We'll see what season 2 brings, but I would also just note the fears that AI can bring about the end of the world, while Carol has the Collective bring her a literal atom bomb...
M.G. Siegler
1 I'm also on the record noting that I don't love Severance, which is far more blasphemous, I know. I've tried to get into four different times now and just can't. Something about it irks me... ↩
2 This is still confusing as all hell to write out because Apple TV, of course, is also a set top box and an app, on top of being a service. Like, which one am I talking about here? They really should rename at least the puck. Even just to 'Apple TV Puck' or the more bland 'Apple TV Box' would be fine. 'Apple TV Set' would be a fun throw-back, but maybe confusing too – in an age where ads and bloatware are increasingly being thrown on to screens by TV manufacturers, I still want an actual Apple television. ↩
3 Like everyone else on the planet, I loved Breaking Bad yet somehow have never gotten around to watching Better Call Saul. I mention this simply because I know this is where most people know Rhea Seehorn from leading up to Pluribus. ↩
2025-12-26 02:49:06
Twas the night before Christmas and all through the house, not a creature was stirring, not even a... wait. What's that? Is that Jensen Huang scurrying around doing deals over the holidays?!
It sure is!
Well, presumably he had been working on the deal with the specialized AI chip startup Groq before the holidays started. But I also wouldn't rule out that this deal came together in a hurry given what we know about how he's operated in the past with some mega deals, such as the um, (up to) $100B investment in OpenAI. A deal which came together quickly, perhaps on a trip with a certain President, and perhaps right after the rumors of OpenAI potentially looking at a deal with Google to use their TPUs – we'll come back to that. It was also a deal which was announced three months ago but apparently still has not been finalized, by the way, so there's that.
Anyway, you can't help but wonder if the real key for the timing here may have been to bury it under the spirit of Christmas. Because boy is it a doozy.
CNBC first broke the news with the headline: "Nvidia buying AI chip startup Groq for about $20 billion in its largest acquisition on record". But as it turns out, that wasn't quite right. And actually, it was NVIDIA that wanted to make that clear. They weren't buying Groq, they were merely paying Groq that $20B for a "non-exclusive licensing agreement".
Oh, you've heard that terminology before? Of course you have. It's the magical incantation one must cite in order to summon the spirits of the "hackquisition".
With this deal, NVIDIA is following their Big Tech brethren down this now well-trodden path. Microsoft, Amazon, Google, Meta, and NVIDIA have done deals which aren't technically acquisitions, but effectively are. Because they allow the acquiring – sorry, non-acquiring – company to sort of pick and choose what they want from the acquired – sorry, non-acquired – company in exchange for considerations that can go to... well, sort of anyone they choose. Certainly the investors to get them to sign off on such deals, and often the key employees at those companies. Sometimes the other employees of the companies too, but that's mainly so they don't feel bad and/or raise a stink about the faux-deal. Because when they do that, all hell tends to break loose, at least from a PR-perspective.
Anyway, the first wave of "hackquisitions" were almost more like "hackquihires" because they were simply a way to get access to key talent at those companies and any licensing agreements were seemingly an excuse to make it all a little less blatant. But as such deals have grown in size, they've grown to look even more like actual acquisitions. For example, Meta's nearly-$15B deal with Scale.ai bought them 49% of that company alongside their key talent, namely CEO Alexandr Wang. Why? Presumably Meta wanted more for that amount of money and they thought Scale, a hot company at the time, might still give them some upside. Of course, they were also effectively gutting said company and it seems decidedly less hot now – especially since many of their customers were Meta competitors who no longer wanted to send their data to Scale – shocking, I know. Still, Meta also likely believed they could use Scale's service to help with their future model training – efforts which were rebooting alongside bringing Wang on board.
NVIDIA's Groq deal seemingly has similarities. Namely, the key was clearly to bring CEO Jonathan Ross on board. And every story also points to company president Sunny Madra being crucial as well. But beyond that, it sounds like NVIDIA may actually care about some of the IP rights here (which they're presumably getting a license to with their "non-exclusive licensing agreement"), to be able to leverage Groq's techniques in would-be future chips.
In that way, "non-exclusive" feels less important here – that's another framing to make it seem less like an acquisition, but is anyone else really getting access to this IP now? Regardless, NVIDIA probably feels confident that with Ross and Madra – not to mention their own in-house prowess – they'll be able to implement it and execute upon it far better than anyone else. And they're undoubtedly not wrong.
And that points to another layer to this as well. Ross is not just a co-founder of Groq, he's also the creator of the TPU – something which he cites in his own bio. You may recall the TPU was last a part of a major news cycle when Jensen Huang was "delighted" about Google's success with their AI chips. How do I know that NVIDIA is actually not so "delighted" about Google's success here and is in fact sweating the rise of the TPU? Well, this deal, for one!
NVIDIA is paying $20B to grab some talent and license some tech. Twenty billion dollars. It's one of the largest deals of any sort in the history of deals. And they're technically acquiring nothing.
Granted, Groq's technology and first batch of chips have been divisive in the industry. Some seem certain they're the future of inference, others aren't sure they're the future of anything. Maybe NVIDIA is fully bought in on the former, or maybe they simply want access to key talent that created the TPU – and, importantly, to keep them away from anyone else who might try to make their own XPUs.
Many are trying, of course. But only Google has really found some level of success thus far. But that level of success has seen them not only train their own state-of-the-art models to rival any of those trained on NVIDIA chips, but also now growing talk about how much more efficient the TPU is versus the GPU. In an era of growing energy fears, this is... potentially a real problem for NVIDIA. To the point where they probably need to have their own TPU-like option – even if they remain fully committed to their GPUs as being the bigger and better option.
Certainly they need to counter this narrative on the inference side. And again, that's where Groq was focused. Maybe their chips were legit, maybe they weren't, or maybe they weren't yet, but even $20B is a relatively small price to pay to effectively lock this team and tech up. NVIDIA made $32B in profit last quarter. This potentially helps them protect those profits.
Speaking of... one can't help but wonder if this isn't the "hackquisition" that breaks the regulators' back, as it were. They've looked into these deals before and largely haven't pursued them. But this is a situation where a "regular" acquisition between NVIDIA and Groq almost certainly would have been blocked simply because NVIDIA controls over 90% of the AI chip market.
So what will regulators do here? Certainly it seems smart of Jensen to try this under the current administration versus any other. It's a group that keeps trying to help him out with China – for a fee, naturally – even if China continues to be a problem on their end.
Regardless – and this remains a key to these "hackqusitions" – NVIDIA will get access to Groq and their team fast with this structure. Will anyone even remember the $20B deal after the holidays? Nothing like some last-minute Christmas shopping!
M.G. Siegler
M.G. Siegler
M.G. Siegler
2025-12-20 01:53:19
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. 🍻
M.G. Siegler
M.G. Siegler
M.G. Siegler
📹 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]
🚙 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 🔒]
💰 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 🔒]
⚽️ 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]
💰 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 🔒]
🎬 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]
"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."
M.G. Siegler
M.G. Siegler
M.G. Siegler
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!
2025-12-19 06:32:26
"Chaos isn't a pit. Chaos is a ladder." For some reason, my mind always drifts to this Petyr Baelish quote when thinking about the current state of AI. And in particular, the constant turmoil inside of the companies making it happen. In building the future, chaos is the only constant, it seems.
And yet it's relative. As we sprint towards the end of 2025, with so much in flux, I thought I'd layout where it seems like the major companies building the AI stand right now – not necessarily in terms of output, but in terms of the situations producing those outputs. At least in terms of outside perception.
Let's count them down from least chaotic to total chaos, shall we?