Apple's NVIDIA Challenge • Snap's Specs Shitshow

Might Apple make their own AI chips? It's easy to think of a few reasons why they would want to... Will Snap's SPECs fail to launch? I mean, come on... Read on, for thoughts on...

Apple's NVIDIA Play

As a $4T company, growth for Apple is increasingly hard to come by. Yes, the services business has been augmenting (and catching up with) the key hardware business, but the reality remains that nothing can touch the iPhone in terms of moving Apple's revenue needle. And while the new iPhone Duo should help growth continue (certainly from an ASP perspective), the company would obviously still love that Next Big Thing™ to come along. It clearly isn't the Vision Pro. The Apple Car stalled. The Apple Bank closed. Becoming an oil company could look attractive given the current situation in the Middle East, but Apple long ago blew past Exxon as the biggest company on the stock market. So what else?

Well, as I outlined last year, there was a trillion-dollar opportunity just staring Apple in the face in the form of their chip business. While it has obviously been a key component, quite literally, of the iPhone and their other hardware, what if they made a move to open it up to others? And the Age of AI gives them just that opportunity. They need look no further than the company that has since surpassed them as the biggest company on the stock market, their old frenemy: NVIDIA.

Well, as Aaron Tilley, Qianer Liu, and Phoebe Liu now report:

Apple has been developing the server with plans to sell it to AI developers, other businesses and governments, the people said. The product under consideration would come in two versions—a smaller one with two of Apple’s planned M8 Ultra chips clustered together and a bigger one with four M8 Ultra chips. The M-series Ultra chips are Apple’s highest-performance chips and power products including the Mac Studio.

There have been reports about Apple working on their own servers for their AI workloads for a while now. But again, the real opportunity may be in selling them to others. I mean, just look at NVIDIA right now. The profits they're posting make those by Apple – long the profit king – seem puny in comparison. Even their other big frenemy, Samsung, has lapped them in profit (albeit thanks to their memory chips, but still – the money right now is clearly in selling chips for AI, just look at the other, other Apple fremeny, Google also moving to sell their TPUs externally). If Apple's looking for growth, I mean, it's obvious, right?

Sure, first and foremost the chips have to be good. And even if the M series wasn't originally conceived for AI workloads, clearly they're moving more in that direction with the M6 making some large leaps and even bigger ones expected with the M7. So this would-be M8 move makes sense on the surface at least. But with a wrinkle:

Apple has considered connecting the M8 chips in the server using an Nvidia set of products called NVLink Fusion, which includes switches, chiplets and software that help chips communicate with each other in data centers, the people said. Any such move would expand on a nascent warming in what have long been frosty relations between the two tech giants.

There's a reason, beyond simple lack of foresight and/or fear of CapEx spend, that Apple has been the one major company not buying up every NVIDIA chip they can get their hands on: the companies simply don't like each other. Yes, this dates back to the day of Steve Jobs when they had a falling out over graphics card issues with the Mac (and perhaps Pixar IP). But that was a long time ago and NVIDIA is a very different company in a very different position now, obviously.

If Tim Cook didn't want to fully mend that bridge, perhaps John Ternus will?

The Apple product isn’t currently expected to hit the market until 2029, and could still be cancelled or proceed without the Nvidia technology. But Apple’s new chief executive, John Ternus, who took over earlier this month, was a backer of the project at its inception around a year ago when he headed Apple’s hardware engineering, and saw an opportunity for Apple’s chips in the booming environment for AI compute, said some of the people.

It's reasonable to question if NVIDIA would want Apple to get into the AI inference game. If not, why help them with NVLink? Then again, everyone seems to be getting into that game in their own ways and it's certainly not hurting NVIDIA's business. At least not yet. And if they're getting into the game regardless, why not have them beholden to at least part of NVIDIA's stack?

Plus, NVIDIA undoubtedly remembers Apple's Xserve initiative and may be skeptical that they can do much in servers this time. Let alone anything in enterprise! Then again, this is also a different Apple. A company no longer reliant on IBM or Intel for their chips, one which can't meet the demand for their Mac minis to be used for AI development, and one on a long hunt for growth, so...

The Specter of Specs

First and foremost, I don't really understand Snap's launch strategy here. This is at least the third unveiling of their Specs smartglasses – which they style as SPECS, which I can't help but read as SPECTRE when capitalized that way – while I guess the September 2024 event was more of a preview, the June 2026 press tour was clearly meant to be an unveiling. And while I think it would then be fine to have an actual launch event, this new wave of press is also not that. Nor is it a full review cycle of the product. It's just... another preview? For the glasses which are coming "sometime in the fall". Just ship or get off the pot?

And, unfortunately for Snap, these new previews aren't great, here's Jay Peters:

Now, Snap is nearly ready to launch its first pair of AR glasses for consumers, the Specs, sometime this fall. That suggests Snap thinks these meet its vision well enough for regular people to enjoy using the $2,195 glasses. They’re certainly full of big ideas. But after an afternoon of testing at Snap’s offices this week, I’m skeptical of whether they’ll be fully ready for primetime.

They're obviously and comically too big (which, to be fair, Evan Spiegel did join in on the jokes poking fun at him wearing them), the controls are said to be interesting (literally snapping to take a picture if fun!) but "finnicky", the field of view at 51 degrees is clearly going to be a problem, as, it seems, is using them in the sun. If there is something new to talk about now it's the AI features, which seem to be angled at the same agentic flows at Meta's Muse because those two simply always must be at each others' throats. Sadly, while Meta's AI seems solid now (after a multi-billion dollar revamp/reboot!), Snap's AI, at least on the Specs, are said to be "slow and inconsistent".

All this for the low-low price of $2,195.

But wait, actually there is one more thing: a $2,395 version! That's if you want direct cellular connectivity instead of tethering to your phone. Instead, you'll be tethering to your glasses case, which will house the service, which seems to be exclusive to Verizon. And yes, will also cost a monthly fee on top of the extra $200 you're spending. We're creeping into Vision Pro price territory here, though, to be fair, the Specs are fully contained with no physical tethers needed. Of course, that also means roughly 4 hours of battery life, which Spiegel is trying to play up as a feature, not a bug. These aren't meant to be worn all the time, you see. But what if you, say, actually need glasses to see? These are not for you.

So who exactly are these for? It's entirely unclear. I could see some level of interest if they were a few hundred dollars, but that would still mainly be as a novelty/glimpse of the future. It just seems insane to launch these as a full-blown product now. Even more so than when Apple launched the Vision Pro nearly two years ago. In hindsight – but also not – those clearly should have been released as a dev kit. But Snap's problem is that they've already had a Specs dev kit in the while for a couple years now. The technology has simply not gotten to the point to make these day-to-day viable as a consumer product yet.

That's harsh, but it's reality. No point sugarcoating it, despite Spiegel's best efforts. The pivot-to-enterprise spin is already starting ahead of the launch. So perhaps they're leaning something from Vision Pro and Google Glass...

Spiegel is clearly in a bit of a bind. Not just with Specs, but with Snap overall. He continues to say the right things about the company's vision and direction with regards to their core advertising business and core usage, but he also clearly made a bad bet here that he can't walk away from. He should cut the tether to the device with no tethers, and perhaps try to take the company private again. The stock market is just far too soured on Snap, to the point where startups that are just weeks old are valued more highly. That's not a joke. But it is pretty sad.

One more thing: I would also just note that Snap is launching the Specs into a market that's increasingly turning on wearables with cameras and other "surveillance" tech. Once again, Snap largely has Meta to blame there. But even Apple isn't immune. Honestly, that's probably the least of Snap's problems here.