Inklings #025 📧

Big week for the frontier AI model race with GPT-6 Astra, Fable/Mythos 5.1, and to less extents, Gemini 3.8 Flash and Must Spark 1.3 all pushed out the door, which is obviously not a coincidence. Next week, all eyes turn to Apple to see if they can turn foldables into a (mass scale) thing. A $2,000 starting point would be rough (though somewhat mitigated), but all other signs (and predictions) point to 'yes'. Happy September, American college football (and English futbol) is very much in the air. 🍂🏈⚽️

I Note...

👋 Phil Schiller Abdicates App Store Throne – On one hand, yes, it's a natural time for him to step down with Cook leaving the CEO perch. On the other, it's a little weird this news leaked at the exact same time of that transition – couldn't everyone have waited, I don't know, a couple weeks? Unless perhaps the move is a key part of that transition, with John Ternus looking to turn a new leaf with the App Store and developers? I know this sounds a bit conspiratorial, and yes, I've been predicting it, but, well, there are signs that this points to some bigger changes in direction forthcoming for the store. Which are long overdue and have put Apple in a precarious spot both legally and from a business-perspective, all around the world. I'm just saying it's possible! One more thing: Schiller also oversaw all Apple events. Might we see a move back to live onstage keynotes? Also long overdue, IMO. [Bloomberg 🔒]

🚕 The Unusually Muted Tesla Cybercab Launch – Look, to their credit, it's 2026 and Tesla has indeed manufactured some Cybercabs. 45 of them to be exact. They're gold and, if we're being honest, pretty hideous. They look sort of cheap and tacky? Slightly better than the Cybertruck, but also in a color that only Donald Trump or Auric Goldfinger could love. I guess they stand out though, like the yellow taxis of old. Still, I must prefer the Zoox aesthetic. (And, to be fair, I've never been a fan of the Tesla look and feel.) I'm not sure this launch event was quite what everyone was looking for as the most notable element seems to be what wasn't there: the press or even Elon Musk. Still, they did manage to ship, so kudos on that. No steering wheel! Legally? Unclear at this point. But that's pretty par for Elon's course. It also may perform miracles? Let's see how 2027 goes, that's when the real promises kick in. Can Tesla scale Cybercab to compete with Waymo? Or even Uber? Will they actually sell these cars to consumers? For $30,000? Or will it all end up a part of SpaceX's mission? [Verge]

☁️ Microsoft Shifts Reporting Structure – While this is being framed as the company trying to be more transparent with regard to Azure, there's certainly an argument to be made that it actually makes the business more opaque in certain ways – notably, it's now harder to back into just how much of Microsoft's Cloud business might be directly tied to OpenAI... It's also interesting just how top-heavy this makes Microsoft from a segment perspective – 'Agents & Infra' is now $75B versus just $15B or so for 'Devices & Consumer' on a quarterly basis. That almost seems like they're trying to position the latter to eventually be cleaved off/spun-out? There's a long history of such a movie, with HP being a prime recent example, but would Nadella dare put Windows in an actually separate company? I mean, maybe if it's small enough! Certainly it gives them a way to ease the Xbox headache... But that's undoubtedly reading too much into it – at least for now. The easier story is that Azure is now a $100B/year business, so it's a natural time to further break it out. But there are second-order effects! [WSJ 🔒]

⚖️ Google Avoids a Breakup of its Ad Tech Business – Who could have foreseen? Except pretty much everyone – well maybe not Lina Khan, Tim Wu, and their contemporaries. I wrote about why such outcomes were going to keep happening in April 2025 – even in the cases the government technically wins! – because there's an obvious and natural mismatch between litigating the past and trying to fix the future. I mean, that's generally true, but with tech, things move far too fast and in many cases, natural "disruption" of the things being litigated are already underway, as was the case in the Google cases, which made the proposed remedies pointless at best and problematic at worst – imagine if the government had forced Meta to sell off Instagram, they'd effectively be destroying their overall business. This is all going to get worse with AI, where things are moving even faster. Good luck. [NYT]

😎 Can Evan Spiegel Sell the World on $2,195 Smart Glasses? – A very Betteridge's Law headline from WSJ here. Also, my own such headline was better three months ago. Look, I don't want to be a dick, but doubling down here with a big launch in a couple of weeks just looks stubborn, if not stupid. I get that Snap is sort of out of options after failing to entice outside investors and pouring so many resources into this project with the promise of a 2026 launch, but the timing is just all wrong. Not only are the glasses an order of magnitude too expensive – a lesson even mighty Apple had to learn the hard way – but there's now growing backlash against such wearables thanks to their old nemesis Meta. Yes, AR is different from AI, but the public is not going to care about that difference, if they care about the product at all! And unlike Meta with Quest, Snap doesn't have the profits to weather this being a dud. Hard to see how this ends; I still think the company should try to go private again if they're going to insist on such efforts. Get an AI narrative and the investors will come for that user base alone... [WSJ 🔒]

I Wrote...

I Quote...

"There’s a saying here in Tennessee. Pigs get fat, and hogs get slaughtered."

– Jonathan Skrmetti, Tennesse's attorney general, in a meeting to urge his fellow AGs to take Meta's $17B offer to settle the social media case against them. As the quote so wonderfully illustrates, that's a lot of money to those states. It is not a lot of money to Meta. Certainly not $200B, let alone $1.4T.

As the settlement makes clear – and the reporting that they were trying to settle before the trial even started – Meta was terrified of losing the case. Beyond the money, it was going to be yet another blow to their reputation – undoubtedly the biggest one yet – just as the company is trying to move on from social media, looking towards their AI future.

Thoughts On...

Asides...

• Conflicts of interest for VCs may no longer matter for Big AI, but they still seem to for smaller AI startups, particularly in the "personal assistant" space... [Upstarts]
• Anthropic has officially licked enough Trump administration boots to be back in the good graces. Well, maybe not. [Axios]
  • But they should be back with large enterprise customers thanks to the data retention changes. Let's see what it does to the Fable numbers... [CNBC]
• MapQuest, last heard from around the time of "Lazy Sunday" is um, atop the App Store. Blame Canada if you want, but unlike Google and Apple, they refuse to budge on the 'Lake America' naming nonsense. [MacRumors]
• Meta mandating Slack usage across the company is wild to those of us who recall the company's attempt to directly compete in the space. AI, bringing former enemies together. Microsoft dropping Teams for Slack next? 🤣 [BI]
• I'm sure 'Google Pics' – their new AI-centric Canva/Adobe competitor – won't be confused with Google Photos like every single day. [Verge]
• As Howard Stern cuts back to one show a week on SiriusXM, he is putting a video version on HBO Max – sort of backing into podcasting, in a way! [THR]
• Has Steve Ballmer sunk the Clippers? It sure looks that way. From the logo redesign to the Kwahi Leonard shitshow, it's not a great look. But I prefer to think that it was to move them to Seattle to be reborn as the Supersonics. [FOS]

I Spy...

Oh look, Apple sees some success with an idea – here: low-priced, colorful laptops – and the industry quickly follows. The problem, of course, will be that they lack Apple Silicon and run Windows. But otherwise, yeah, similar. Who could have known such a simple formula would be a success?...