M.G. Siegler β€’ β€’

Inklings #015 πŸ“§

NVIDIA's CPUs β€’ Anthropic's IPO β€’ Streaming Bundles β€’ Dickovers β€’ Google Home Speaker β€’ Apple's Smart Glasses in 2027 β€’ Intel's New AI GPU β€’ Surface Laptop Ultra

Given how much was being telegraphed ahead of Jensen Huang's Computex keynote, I wrote up some thoughts on NVIDIA jumping into PC-focused CPUs a few days ago (of course, it was a long-time coming).

With 'RTX Spark' (what is with seemingly every company using the 'Spark' branding for a different element of AI?) now unveiled, it seems... pretty much as expected. It will start as a high-end chip for high-end PCs that's undoubtedly going to be expensive (no prices were given) simply because of the memory alone. Will they take off where "Copilot+ PCs" did not as personal "AI Supercomputers"? We'll see. A lot of that will fall upon just how well they actually run Windows (and the myriad software built for Microsoft's operating system). We'll undoubtedly hear a lot more about that at the Build keynote in a few hours... Forget Qualcomm, would love to hear Intel's thoughts on this!

Microsoft Seeks a Surface RT & Copilot+ PC Do-Over with NVIDIA
Can the main AI chip maker ensure Windows owns the β€˜AI PC’?

Thoughts On...

πŸ“ˆ Anthropic Files to go Public – Alex Kantrowitz and I were in the middle of recording our monthly podcast when this news hit, so we channeled Bill O'Reilly and did it live. It helps that I wrote about this very prospect (twice!) a couple weeks back: how complicated the narrative would be for OpenAI if Anthropic filed first. That was because it leaked that OpenAI was about to file, perhaps to take some wind out of SpaceX's S-1 sails – and well, it looks like Anthropic quietly beat them to the punch! So yeah, that's a problem for OpenAI optically, if nothing else. But that's also nothing new for OpenAI. Their hope would be that the market can digest all of these listings, but the fact that they're all likely to be listed in indices like the NASDAQ 100 and S&P 500 in relative short order, thus causing major portfolio rebalancing, could cause chaos. No surprise that OpenAI is out there today touting Codex growth, as they need to try to turn this story around, pronto. The other thing, not mentioned with Alex, that I'd expect them to lean on heavily here: ads growth. It's the one business area Anthropic apparently isn't touching. So OpenAI has a potential greenfield there – if they can get them working – at least from an IPO narrative perspective... How long until OpenAI files? [YouTube]

πŸ“Ί Streaming Bundles Prove Popular – Yeah, duh. With up to a third of sign-ups now happening with one of these partnerships, don't be surprised to see Netflix jump into the game soon too. They've dipped their toes tangentially, but an official Netflix + well, anyone, would obviously immediately be the most popular one of these offerings. The more interesting play remains if they go down the "channels" route that Amazon has pushed so effectively, and let (made) other services use their app (and UI, and recommendation engines) to serve up the content. This is obviously a dangerous path for the other streaming services – and why Netflix hasn't given such power to any of the streaming boxes. But as YouTube keeps creeping up (and backing into being the actual new cable bundle via YouTube TV), it feels like Netflix will make moves here. [NYT]

βœ‹ Don't "Dickover" Your Readers to Death – If I stop to think how much of my life has been spent dismissing pop-ups/overs on websites (especially in Europe), I immediately get depressed. Added together, it's a meaningful amount of our time on Earth spent doing this. But on an individual site level, it's both annoying and just rude to your readership, as John Gruber colorfully illustrates. It was definitely one key aspect of going with Ghost as a publishing platform over Medium or Substack for Spyglass. I just want the content to load and render in a way that's a great reading experience. Not, you know, this. [Daring Fireball]

πŸ”Š Where is the Google Home Speaker? – I have been wondering this myself because, frankly, I want one. The first purpose-built Gemini-powered device was long promised and promoted to be "coming Spring 2026". It's now June, and while we may technically have a couple weeks of "Spring" left, I think everyone would consider this to be Summer. The fact that I/O came and went without a mention of the device didn't seem like a great sign. But maybe there's some hope in the product pages of Canadian Best Buy (though they have since removed the "June 25" date – replacing it with "coming soon")? Given the fact that AI upgrade work continues on various Google Home products, this clearly seems to be a software/AI issue and not a hardware one. Which is interesting because that would be the same issue that has bedeviled both Amazon with the Alexa+ roll-out and Apple with the new HomePods (and newfangled Siri in general, of course). I'll just reiterate what I wrote two years ago: sometimes being early is worse than being late. That is, because Amazon, Apple, and now Google have to upgrade old systems rather than just rolling out new ones, they're paying a legacy tax, as it were. And clearly none of them were quite ready for it. [9to5Google]

I Wrote...

In other Microsoft news ahead of Build...

The Summer of β€˜Super Apps’
Microsoft races to get the one Copilot to rule them all out the door…

I Quote...

"This reinvention of the computer is as big of a deal as the reinvention of the phone into what we now know as the smartphone. Microsoft and Nvidia are going to reinvent the PC. This is the first completely re-engineered, reinvented line of PCs that has happened in 40 years."

– Jensen Huang, making the case for RTX Spark, and NVIDIA's place in the PC market. I haven't heard this much hype in at least a few months, when Jensen was telling us how OpenClaw was the most important software ever created. Now for both, we'll see...


This post is for paying subscribers only