2025-12-03 21:27:06
At this point last year, OpenAI was firing on all cylinders which culminated in the "12 Days of OpenAI" – better known as the "12 Days of Shipmas": a dozen consecutive days of product updates and announcements. A real shock & awe campaign of execution. This year, it's unclear if we're going to get the same level of holiday spirit as the company is the one under assault on multiple fronts, challenging their long-held leadership position in AI. If anything, it may be more like the "12 Days of Oh Shitmas"...
2025-12-03 06:00:08
In case it wasn’t already obvious that 2026 is going to be a wild year for AI, just look at the end of 2025. Currently the biggest story in tech is that Sam Altman has issued a “Code Red” at OpenAI, aiming to light a fire under the company as they’re seemingly under attack for their leadership position in AI for the first time. And that hasn't even been the leading story atop Techmeme most of the day. Because with said AI war in full swing, Apple’s AI chief, John Giannandrea, is out.
He’s “retiring” which is optically the right thing for him and Apple to announce. But come on. It’s one thing for the CEO of Apple to be thinking about retirement with a brief lull in the action (for Apple, at least), while the company is near their peak in terms of sales and market cap, it’s another for the leader of what many would assume is the single most important technology for any company going forward to decide it’s a good time to retire. He’s retiring because Apple has been operating like the entire company is retired, at least when it comes to AI.
He’s retiring after a complete shake-up of his group both above him and below him as the company aims to reorient themselves for the coming AI battle. And he’s doing so ahead of any larger shake-ups so it’s not total chaos with so many people likely leaving in 2026. Including, perhaps, that aforementioned CEO.
Anyway, the writing was clearly on the wall with Giannandrea — just as it was with Yann LeCun, Meta’s former head of AI… Meta, um, decided to go in another direction, completely leaning in to LLMs as LeCun keeps leaning away towards "World Models". Which may or may not be the future...
Meanwhile, over at Amazon, they're releasing not only their own, new frontier LLMs, they're also releasing their new Trainium chips to help build said models. As well as the models from Anthropic, which makes models they both now compete with and use. Because the models from Anthropic are also going to be trained with TPUs from Google. And GPUs from NVIDIA too, it seems. Which is interesting since Amazon owns 20-something percent of Anthropic. And Google owns another double-digit percentage.
They're also partnering with Microsoft, which is also working on their own, new frontier models, now free from the constraints of the previous deal with OpenAI which prevented them from going after AGI. They can now fully compete with those models, but also use them, since they own 27 percent of said company. While Microsoft is also working on their own chips, both they and OpenAI currently train on NVIDIA GPUs. At least until OpenAI also builds their own chips with Broadcom. Also, they're going to work with AMD's chips in some capacity too. But at least for now, they're not training on Google's TPUs, as a $100B commitment from NVIDIA may have snuffed that possibility out.
For now.
And that alignment makes sense because the main reason for the actual "Code Red" at OpenAI and the de-facto "Code Red" at NVIDIA in recent weeks is seemingly mostly about Google. Gemini has not only gotten good enough as a model, but is clearly getting good enough as an actual product to eat into ChatGPT. And there's real fear that the models are improving in ways that have surprised OpenAI on the pre-training front, and that may have something to do with said TPUs. Which is also potentially a problem for NVIDIA. Is "Garlic" a remedy? Nobody knows.
Regardless, the fact that Google is able to train cutting-edge frontier models without GPUs is clearly not a narrative NVIDIA likes floating around out there. Especially the talk of others using the TPUs. But even worse than Anthropic using them in Google's Cloud would be the notion of Google selling their chips to Meta to use in their own infrastructure. In some ways, this is worse for NVIDIA than the debate around chip depreciation and even circular financing. Because margins tend to get less fat and profits tend to get skinnier as competition enters the picture. And everyone that NVIDIA partners with is also seemingly trying to enter that picture.
As they continue to hammer money into NVIDIA...
And so we're entering 2026 with the broader AI landscape perhaps even more unsettled than it was entering 2025. That was, of course, right before DeepSeek entered the picture – well, technically right after, as that bomb landed and yet didn't explode for a few weeks. Once it did, it sent all of the players above scrambling. But they also soon realized that practically speaking, not much had changed. Still, it caused a mentality shift simply around the notion that things could quickly change. (And maybe to some extent about "open source" models – unless you're Meta.) And that has continued throughout the year.
So in a way, it's fitting that we're ending the year with things more chaotic than ever. Google, which had been depressed much of the year – at least from a stock price perspective – seemingly woke up at Google I/O and slammed their foot on the gas. The ramification of which we're just seeing now with Gemini 3's release.
And it's the first time that OpenAI's general lead in the space has felt threatened, perhaps outside of the DeepSeek blip. They started out strong from a product perspective and kept getting stronger with viral hit after viral hit spurring more usage. But Code Reds don't lie, and even Colonel Jessup can't lie about them. Such an order worked for Google a few years ago. Will it work for OpenAI?
It's weird timing for such an order, given that the company had just brought on a new "CEO" of their product division. Now seemingly everything that Fidji Simo was brought on board to oversee is taking a backseat to getting back to the fundamentals. And the actual CEO is seemingly back to overseeing those, and issuing Code Reds.
From the outside, it has long felt like OpenAI may be trying to do too much. This is seemingly an indication that the company also now agrees. The shopping agent stuff takes a backseat, as does the daily briefing work, as do the ads. The latter wasn't actually rolled out yet, but code leaks suggested it was coming imminently – and that was undoubtedly another key element of Simo's involvement, given her previous work with both Facebook and Instacart. Awkwardly, these product initiatives were likely three of the key pillars of ramping revenue. If those are paused, is burn about to get even hotter?
And none of this even speaks to the ongoing issues with AI safety, and China, and the myriad other things in motion and in play...
Things feel... tense at the moment, to put it lightly. I mean, NVIDIA is just blurting out "ENRON!" at this point, coprolalia-style. If the markets start turning against AI spend, that's going to be a big problem, to say the least. Even though OpenAI isn't – yet – public, the trickle-down effects will be very real here. Talk of spending "trillions" could quickly morph back into mere "billions". An Age of AI Austerity.
If that comes in 2026 and the AI Bubble starts to deflate – it doesn't even have to burst – we might be seeing a lot more Code Reds.
Can the tech industry handle this truth?
2025-12-02 21:19:29
Yesterday, Alex Kantrowitz and I sat down to record our monthly chat for his Big Technology Podcast, and we kicked off talking about the possibility that Tim Cook could be contemplating retiring from his CEO role at Apple. My thought remains that the timing seemingly is perfect right now, and assuming he's thinking about it, he might not get a better opportunity...
Speaking of timing, about 20 minutes into our chat, talking through the situation at Apple that Cook may be handing to John Ternus, his would-be successor, I was trying to think through the main things to focus on with that transition:
Say Ternus is elevated to be CEO. Who’s in charge of AI? They need someone to be in charge. Right now it’s been John Giannandrea. I think everything you read and hear, similar to the Yann LeCun situation at Meta, it sort of feels like the writing is on the wall there. That he will probably not be there for very much longer.
Well, "very much longer" turned out to be a few hours. Right after the podcast went live, Apple announced that Giannandrea would be retiring from Apple.
The rest of the discussion centers around Cook's tenure and how any succession might parallel the ones that took place at Amazon and/or Disney. Is Cook waiting for that "one more thing" to unveil hardware-wise in the form of Smart Glasses? And speaking of hardware, with the seeming pivot back to that being Apple's main narrative (in the face of the AI struggles), does that make Ternus the right guy at the right time, or does his hardware focus hold him back in some way given that Apple is increasingly becoming a Services company?
Continuing on the Giannandrea thread, I noted that it felt like they still needed to bring in someone from the outside to give new life to that group, and that M&A could be the framework for how they refresh their talent (which continues to bleed over to Meta). Well, they're not doing M&A yet, but it seems like they poached AI researcher Amar Subramanya from Microsoft to lead the group. Granted he was at Microsoft less than six months because most of his time was actually spent at Google, much like Giannandrea before him. But JG came from a Google focused on ML – the element of AI that everyone talked about before LLMs came along – Subramanya was seemingly working in the heart of Gemini.
Is is also interesting that he's reporting to Craig Federighi and not Cook (as Giannandrea was). This makes Federighi the de-facto true head of AI, which I put out there as a possibility given the post-Siri shakeup which saw him take over those groups alongside Mike Rockwell.
Alex wanted me to sell him on the new iPhone 17 Pro, which I think I successfully did. Certainly over the iPhone Air, which was just positioned oddly by Apple in the market.
We go on to talk through the Great Big Tech Market Cap Race taking place between NVIDIA, Apple, Google, and Microsoft. And how the ever-shifting narratives around AI are altering that and seemingly bringing those four companies together at the top. Google's "comeback" has impacted not just OpenAI, but NVIDIA and perhaps Microsoft as a result. And, if Google and Apple are truly about to partner to bring Gemini in to help power Siri, the resurgence could help Cupertino's narrative too.
Next up in AI narrative shifts – well, once NVIDIA stops putting their foot in their own mouth – world models? Is that a window for Tesla to come back into the equation? Clearly Elon Musk is angling for that with Optimus...
Who would I bet on? I would do what these companies are doing and bet on all of them – bet on each other, to hedge. But if I had to pick one, probably Google for the reasons I laid out previously before the current stock run. Though I remain intrigued by OpenAI's "AGI or Bust" bet – if they can find enough places to get enough capital to keep going... It's obviously the biggest wildcard in all of this and will be until the capital wells start to dry up.
But I do think it's interesting/worth noting that Google seems to be doing some great product work on top of AI, long OpenAI's strong suit. The new "Dynamic View" mode of Gemini – awful name aside – is incredibly impressive.
Finally, if the worm really is turning with regard to AI spend, is Anthropic's move to focus on profitability in the face of OpenAI's continued growing spend an especially savvy play? Sure, they still need to invest a lot of money to work towards the future of AI, but perhaps mere tens of billions is enough and not hundreds of billions – or trillions.
2025-12-01 17:54:10
Where there's a will, there's a way. Here's Ming-Chi Kuo:
Intel expected to begin shipping Apple’s lowest-end M processor as early as 2027
There have long been market rumors that Intel could become an advanced-node foundry supplier to Apple, but visibility around this had remained low. My latest industry surveys, however, indicate that visibility on Intel becoming an advanced-node supplier to Apple has recently improved significantly.
Apple previously signed an NDA with Intel and obtained the advanced-node 18AP PDK 0.9.1GA. The key simulation and research projects (such as PPA) are tracking in line with expectations, and Apple is now waiting for Intel to release PDK 1.0/1.1, currently scheduled for 1Q26. Apple's plan is for Intel to begin shipping its lowest-end M processor, utilizing the 18AP advanced node, as early as 2Q–3Q27, but the actual timeline remains contingent on development progress following the receipt of PDK 1.0/1.1.
While Kuo's track record is hit-or-miss, that's usually only when he tries to veer outside of his wheelhouse in Apple's supply-chain. He seemingly has good details for this shift here. And, of course, this follows other reports from earlier this year that Apple might through Intel that foundry bone they so desperately need.
To be clear, this wouldn't be Apple handing a big chunk of their chip manufacturing to Intel – it's a tiny sliver of the work TSMC does for the company. That's in line with how I guessed such a partnership might play out back in September:
I say "tall" because it also feels unlikely given the tight relationship between Apple and TSMC – with the former often first in line for the newest processes coming out of the latter. But again, maybe if Trump specifically asks Apple, Tim Cook will figure out a way to use Intel to fab some of their chips. Apple has a lot of them these days, beyond just their CPUs. If you squint, you can almost see a world in which Intel helps Apple make their 'C' line of chips, which would be an extra nice narrative since their modem chips were born out of Apple buying Intel's modem business...
Well, per this report, it's not the 'C' chips – at least not yet – but the lowest end of the 'M' lineup probably makes more sense given both the relative low volumes (and thus, lower stakes – whereas 'C' chips are increasingly going into Apple's cutting edge devices) and the fact that Apple has been making such chips for years.
Back to Kuo:
Apple’s lowest-end M processor is currently used in the MacBook Air and iPad Pro mainly, with combined shipments of roughly 20 million units for 2025. As MacBook Air shipments in 2026 may be impacted by a new more-affordable MacBook model using an iPhone-class processor, shipments of lowest-end M processor in both 2026 and 2027 are expected to be 15–20 million units.
I think he means iPad Air here, not iPad Pro, as the Pro models use the top-of-the-line M chips, of course. If Apple + Intel could hit this second half of 2027 timetable, presumably the "low-end" M chip at the time would be an M4 or M5 chip – yes, the top-of-the-line right now. Currently, the iPad Air ships with an M3 chip and presuming that gets updated early next year to an M4 chip, there's a world in which the 2027 model gets an M5 chip (though who knows about exact product timing by then, obviously).
Meanwhile, the current Mac lineup has the last-generation M4 as the lowest-end chip (well, technically, the Mac Studio still has the M3 Ultra, while the Mac Pro still uses the M2 Ultra (!), but those are obviously higher-end chips). But then there's that interesting Walmart MacBook Air, which is still very much for sale (and, in fact, down to $549 for the holidays) using the M1. You have to believe that this chip, as great as it has been for Apple, will be end-of-life'd sometime soon and it will be interesting if Apple gives Walmart access to say, the M2 or M3 chips to continue selling the product. If so, it's possible that this is the lowest end M chip that Apple intends to make with Intel.
But it's perhaps just as likely that Apple lets that partnership run out with the rumors that they're about to launch their own affordable MacBook early next year. Interestingly enough, this new product apparently wouldn't run on an M chip, but instead on Apple's A-series, built to date for iPhones (and the more affordable iPad). Kuo's report specifically notes the M chip, so presumably this would-be Intel partnership isn't about these chips – and certainly not the ones due soon, of course.
If Apple is about to move towards only using their lowest-end M chips in the iPad Air (with the lowest-end of the MacBook lineup shifting to the A chips), this might make even more sense volume-wise to try Intel's hand here.
But why is Apple throwing Intel, their old CPU partner who they famously parted with on not the best terms when they spanked them with the introduction of the M-series itself several years back, a bone here? Well, presumably they like the notion of continuing to diversify their manufacturing away from single points of failure – which TSMC very much is right now. Intel clearly isn't ready or able to take over all of Apple's chip business – but they likely never will, as in an ideal world, Apple would have multiple fab partners, all around the world, for making their chips.
Speaking of the world, this Intel partnership would obviously also give Apple a nice "made in America" narrative once again – and give Tim Cook something else to hand to President Trump on a golden platter. You'll recall that the current largest owner of Intel is... the US Government.
But the biggest question mark here remains if Intel can use such a deal to truly ramp their fab business into an actual rival to TSMC. This deal would be a nice, marquee name, of course, but it's mainly for the optics right now. Can Apple help them whip the actual chip-making processes into shape? Will it give others the confidence to go with Intel? Amazon? Microsoft? AMD?!
One more thing: sort of humorous to think of this news in the context of Intel's old "Intel Inside" marketing campaigns. If Intel starts producing Apple chips, there would technically be Intel inside, though not the chip itself, but the process.
M.G. Siegler
M.G. Siegler
M.G. Siegler
2025-11-30 07:33:13
There I was with a dinner to eat and a couple hours to kill. Oh look, Flight Risk freshly available on HBO Max. I hadn't heard great things – I hadn't heard much of anything. But Mark Wahlberg, Michelle Dockery, and Topher Grace directed by Mel Gibson, how bad can it possibly be?
Bad.
I honestly can't think of a worse movie I've seen in recent memory. Granted, I tend to avoid overtly bad movies and had I just looked at the Metacritic score ahead of time, I probably would have avoided this one. Instead, I just went for it. My god.
The premise is fine, I guess. The person who is the reason for the flight is a flight risk, but also the flight itself turns into the risk. Get it? And I sort of like the notion of having a small group of actors naturally contained in one space for an entire movie. And I even like all of the actors involved here individually!
But put together, they just don't work. Part of it is the accents – more Dockery's American accent than Wahlberg's faux redneck one (because it was part of the plot) – but a bigger part was Wahlberg's hair situation. I mean, did he insist on shaving his head so there would be stubble so people wouldn't dare think he's actually bald or something? It just looks ridiculous.
From there, it only gets worse. Topher Grace tries to do Topher Grace stuff but it doesn't work here. Comically obvious plot points are revealed with zero subtlety – "how did you know we were going to Seattle?" The whole fear-of-flying angle, which is actually the one thing that might make sense for the plot, quickly flies out the window, though sadly no one else does until the end. I quite enjoyed how the autopilot on the tiny prop plane flying over mountains in Alaska was super smooth and flew perfect for most of the flight, but when the pilots were in control, that's when the shit hit the fan.
I also loved the warning not to fire the gun in the plane lest it take the whole thing down, followed by endless firing in the plane, including of a flare gun.
I'm sorry, but nothing can possibly top when the nice man over the radio helping the person who has never flown before (yet keeps reading off altitude measurements as if she's an experienced pilot?) tells her that she should do a trial run at the landing, knowing she's running out of fuel – and not to mention that one of the characters is dying from blood loss. They only don't because she refuses and sure enough, when she goes to do the actual landing, there's no more fuel. In other words, had she done the trial run of the landing, they'd all be dead. This is never addressed.
Oh look, a perfectly prepared syringe of morphine!
But at least they addressed the fact that the flight risk guy knew one random payment he helped orchestrate for the mob or whomever was going to a specific area of New York where, as it turns out, the head of the FBI or whomever also happens to live, as he discloses while telling his agent to make sure to call him at home, and very specifically says to call him at his home in that very specific area of New York. What a twist!
The same FBI director or whomever takes about five seconds to admit to murdering his subordinates when the accusation is made. Oh yes, and then he's of course on the phone with the would-be assassin at the (mercifully short) end.1 AND I WOULD HAVE GOTTEN AWAY FOR IT TOO, IF IT WEREN'T FOR YOU PESKY KIDS.
Honestly, I'm embarrassed for everyone involved in this movie. I'm also embarrassed that anyone would give this film a rating above zero. Mel Gibson, for all his faults, has directed some great movies. Some truly great ones, even. What the fuck was this?
2025-11-28 22:23:20
Happy belated Thanksgiving to those in the US (though we celebrated here in London as well). Catching up the day after with a lot packed in below. While also, of course, gearing up for The Game tomorrow. Five in a row? Seems improbable though certainly not impossible... #GoBlue 〽️
M.G. Siegler
M.G. Siegler
🗣️ Ilya Sutskever on Dwarkesh Podcast – There's a lot in here, and yes, some great memes. A few main takeaways include his notion that Safe Superintelligence can compete with his former colleagues at OpenAI (and elsewhere) despite having less capital in the bank because those guys are bogged down in the "rat race" – not only having models and products out there in the world, but also solely (or mainly) focused on LLMs (where "scaling has sucked the oxygen out of the room" when it comes to new ideas). SSI has "returned to research", trying to come up with other models and methods but at the same time, he's backing off the notion of going straight for Superintelligence and hinting they might get some work out there sooner ("gradual roll out"). Because the work might actually be around "a mind that can learn" rather than be pre-loaded with everything – which isn't natural. As always, Sutskever loves going back to human evolution and that's clearly guiding him here. Will it work? No one knows, and he's pretty honest about that – which also may be why his co-founder Daniel Gross bailed for Meta when Zuck came with a "Godfather" offer (which Sutskever directly addressed too – he didn't want to sell, but Gross clearly did). He thinks we're still 5 to 20 years away from what we would consider AGI/Superintelligence/Whatever. And perhaps more notably, that the LLM-focused folks may "stall out" on their progress, perhaps in the next couple of years. That doesn't mean some won't be good businesses, but they'll be even more squarely competing in that "rat race". That's, um, something to watch! [Dwarkesh]
😮 OpenAI's First "Jaw-Droppingly Good" Prototype – While it has long felt like you could almost triangulate what they've been working towards, the news here is that they finally have a prototype they're excited about. From what Sam Altman and Jony Ive are saying, it's not clear even if the device is actually working yet – all we have is the "beautiful cabin by the lake, and in the mountains" vibe reveal from Altman – or if they're just pleased with the form factor they've landed upon. But it's apparently less than two years away, per Ive. "I hope when people see it, they say 'that's it," says Altman. To which Ive quickly jumps in, "Yeah, they will." His short laugh appended to the end of that gives one a lot of hope. He clearly likes whatever they have! Now for the fun part: figuring out how to manufacturer it at scale. It doesn't seem like a coincidence that OpenAI is ramping up their poaching from Apple's hardware teams... [Emerson Collective]
🎞️ Could Netflix Win Warner Bros? – This report would sure like us to believe it's a two-horse race between the streaming giant and Paramount (with Comcast seemingly in a distant third). The playbook seems to be all about convincing WBD that a deal with them would go through regulatory easier because streaming is a more dynamic and competitive market than traditional Hollywood. But Paramount has their trump card, literally, so it's just hard to see how they don't win this. Unless the WBD board feels like they really should just sell the studio + streamer and let the networks spin-out in to their own entity to be sold at a later point – i.e. can Netflix make a partial deal that's better than what Paramount is offering for the whole company right now? The fact that Netflix's own investors don't seem to love the prospects of this deal is interesting too... Updated "sweetened" bids are due on Monday. [New York Post]
📺 Warner Bros Discovery Wants to Make HBO, HBO Again – Speaking of, after years of trying (and failing) to turn HBO into Netflix – doing basically everything to destroy the brand in the process (to be fair, this was mostly AT&T), including burying it in the most MAXimum way possible, WBD now seems set on reverting everything, from the branding on down. Of course, the world is quite different than when HBO rose to power – least of which because Apple and others have stepped in to fill some of the HBO-shaped hole – and it's not just streaming, but the fact that movies themselves play a different role with the brand (i.e. HBO is no longer the place you go to see movies at home first after theatrical runs and video store rentals). "Prestige TV", which HBO more or less started, has now mainly subsumed everything. All of this sounds good – including using "Max Originals" for the more regular cadence shows – but it's sort of humorous that it's happening just as someone new is going to buy the brand yet again. Will they monkey with that swing? Probably! Will it ironically be Netflix? Maybe! [THR]
📲 The iOS 'Snow Leopard' Moment – Mark Gurman says the next iteration of iOS – iOS 27 – will be more about cleaning up and streamlining the OS, just as Apple has done from time to time with macOS in the past. Feels like a good thing, but also a necessary thing as operating systems age and naturally gather cruft. It also should give Apple a way to refocus around AI, as many such features and work will presumably be under-the-hood as well (and will start rolling out with iOS 26.4 in the Spring). And though it's not mentioned, one presumes Apple also has to do some interesting work to get the OS ready for the forthcoming 'iPhone Fold'. No one is really talking about the software, but presumably it will run "straight up" iOS on the front screen and some sort of new, expanded version when unfolded. Or could it unfold into iPadOS? Would that be too jarring for people? One more thing: Gurman also refutes the FT report that Tim Cook could retire in the first half of next year. We'll see. I still think it was a trial balloon floated by someone. [Bloomberg 🔒]
M.G. Siegler
M.G. Siegler
"Let’s be very clear: What Sam does, I cannot do. There’s so much to do just on my scope that I think I have a decade or more of things that I can do just right there. And I’m telling you, we need all of us. We need Sam so badly. We need me."
– Fidji Simo, when asked directly by Zoë Schiffer for a Wired profile if she would ever "consider becoming CEO of the whole company?" It's a good answer, in that I believe it to be true and sincere. As has been proven time and again, no one can cut deals and raise money quite like Sam Altman. But it's also not a "no."
My suspicion, from the get go, is that this remains the long-term plan. If and when OpenAI goes public, who is CEO of the company?
M.G. Siegler
M.G. Siegler
While OpenAI itself may not carry a ton of debt (they just have a relatively small credit line, which they haven't drawn), everyone is well aware that their partners are loading up to the hilt. Why? To build data centers for OpenAI.
As we've discussed, as a (highly) unprofitable startup, OpenAI would have trouble getting access to such debt – certainly not with the ease that Amazon and Meta and Google can – so they're leveraging their partners, coming up with "new ideas" for financing. That includes, by the way, the data centers OpenAI intends to operate themselves. Who is paying for that? Likely NVIDIA.
Humorously, this likely came together to combat Google, but also to prevent OpenAI from going down the TPU path with them...
Anyway, it's rather remarkable to see the partners' debt for OpenAI data centers so cleanly laid out in visual form. As is this:
The $100bn of bonds, bank loans and private credit deals tied to OpenAI are equivalent to the net debt directly held by the six largest corporate borrowers in the world — including carmakers Volkswagen and Toyota and telecoms groups AT&T and Comcast — according to a 2024 report by asset manager Janus Henderson.
Private Credit. SPVs. The race is on to get this all working – at least somewhat economically – before the inevitable burst... Also to just get them working, period. The shortages – from power on down – are starting to mount. As such, so is the pressure.