Hello and welcome to the Vergecast, the flagship podcast of The Verge. I'm your friend David Pierce, and on today's episode we are going to talk about the Nvidia RTX Spark, a new computer chip that Nvidia thinks might help it make a mainstream play into laptops and that might shake up the Windows industry for a long time. The Verge senior editor Sean Hollister is going to come on. We're going to talk all about it. But first, here's a look at everything else happening at The Verge today. It's 90 seconds on The Verge for Tuesday, June 2nd, 2026.
Today is Microsoft Build Day. It's Microsoft's annual developer keynote. And in addition to lots of new Windows features and lots of new AI developer tools, Microsoft announced a new product called Scout. It's an AI personal assistant agent, kind of like Gemini Spark or OpenClaw, designed to actually use AI to go do things on your behalf. And fun fact about this, it's actually built directly on the OpenClaw technology, which is very funny given that Microsoft CEO Satya Nadella only a few months ago compared OpenClaw to a computer virus.
Things change fast, but this idea of doing stuff on your devices is everywhere and it is absolutely taking over inside of Microsoft. Microsoft also announced a new quantum chip as it continues to push towards quantum computing. It announced a new Surface device, the RTX Spark Dev Box, based on Nvidia's new chip that it hopes will help people make more AI tools. Because Microsoft's whole thing is making it easy for you to make AI tools, whether you'd like it or not.
President Trump signed an executive order essentially asking AI companies to give the government 30 days to review AI models before they release them. This is a version of an executive order that's been floating around for a while. At one point it got shelved, but it's a big change for the Trump administration, which has gone from essentially an anything-goes policy to AI to trying to have some more oversight. We'll see how that goes.
Finally, some big news for me in particular. Nintendo Music, which until now has just been a mobile app for listening to soundtracks from your favorite Nintendo games, is now a web app and supports CarPlay, which means I will be driving to the Mario Kart soundtrack for the rest of my life. You can read more about all of this at theverge.com. That's it. That's 90 seconds on The Verge for Tuesday, June 2nd. All right, let's switch gears now and talk about chips. The Verge's Sean Hollister. Welcome to the show.
Hi, glad you're back. We bring you here when I need you to explain whether a chip is any good. This is one of your jobs here at The Verge: is this chip any good? Nvidia just launched a new chip, the RTX Spark, that I would say it is making some pretty gigantic, grandiose claims about. But first, let's just back up a little bit. How big a deal is it that Nvidia is launching what amounts to a sort of mainstream computer chip?
On the one hand, Nvidia is in everything, right? It's got graphics chips that have been in laptops for years now. It's got pixels flying down from the cloud into your devices, but they haven't been in many devices as the CPU, as the central part of your computer. That's been true if you've got the Nintendo Switch. It's been true if you had like the Shield, the Shield tablet from years ago, but now it's the thing. It's like the Apple chip. It's the Qualcomm chip at the heart of the PC. It's the Intel. It's the AMD chip at the heart of the PC. Now, they're finally doing that thing where they are the silicon.
And so Nvidia getting into that realm, it means that there are going to be lots more ARM-based laptops in the world that are taking over from x86. In fact, the number of ARM manufacturers now outnumbers the x86 because there's AMD and Intel. But on the other side of things, we have Apple and Qualcomm and now Nvidia. So it's kind of ARM's moment. But are these chips any good? You're asking me to come in and say that they say it's the most power-efficient PC chip ever built. That is the quote. And like Apple in 2020, they have no statistics. They have no graphs. They have no charts. They have nothing to back that up right now.
Right. Well, Apple in 2020 strikes me as like the best case scenario here, right? Because Apple showed up in 2020 and was like, 'Hey, we have these M series chips. We're not going to sell you anything about them. Here's a bunch of graphs that don't mean anything.' And then essentially upended the PC industry, right? Like all kind of all at once just lapped everybody in terms of what you can expect out of your laptop.
What I'm trying to figure out is what Nvidia is trying to do here, right? Because Nvidia obviously has lots of stuff to talk about when it comes to AI, has lots of stuff to talk about when it comes to video games, lots of stuff to sell you on both of those fronts, too. But it's also, like you said, trying to do a very mainstream computer chip. And Jensen Huang, the Nvidia CEO, was even talking about lower-end laptops. Like, what is your sense of what Nvidia's game is here? Is it just complete chip domination because it's Nvidia and that's what it wants, or is it trying to do something specific here?
I think there's two things. One, there is the market opportunity of AI and being able to say that, you know, we want to be powering the AI computer. We want to be the forefront of the entire AI revolution. But I also think it's kind of hedging against... Nvidia knows from very prominent and embarrassing history for other companies that if you are not at the heart of the device, there is always a chance that you miss the next computing revolution. Even though it seems like Nvidia is at the forefront, at the heart of the AI computing revolution right now, they can look back and say, 'Well, Intel wasn't in the phone.'
And that was a big problem for Intel when the tide turned, when it wasn't just, you know, Blackberries and dumb phones, but now it was Intel inside being you didn't have them in the phone. That was a problem for Intel. They missed out on the phone because of that. Nvidia, by the way, even at one point made a pretty big mobile play, like you said, with the Switch that's running a smartphone chip. That was the Tegra processor that they tried to make a mobile phone.
Yeah. Tegra, there were a couple of Tegra phone attempts back in the day. There were tablets. There were... you can credit Nvidia for making $199 Android tablets a thing. We talked about it in 2012 and 2013 on The Verge. But in terms of like you don't want to necessarily miss being at the heart of things.
You famously wrote about the Amazon Fire Phone and how Amazon wasn't able to be, you know, the platform owner there because they couldn't get the OS that they wanted onto phones. They couldn't get a foothold there. So now they have to rely on can we, you know, sell hundreds of millions of Alexa devices into your home. Can we get our foothold into people's homes and lives that way instead?
So here is Nvidia seeing an opportunity but also possibly hedging a huge bet in case people decide that local AI is the thing and not cloud AI and we don't trust the AI clouds with every single thing that we're doing in our personal lives. If Google, if Meta, if Microsoft decide at some point, you know, that they are the cloud owners, you know, if they decide at some point, they and Amazon, if they decide at some point, we want to back a different player instead for AI.
And if things keep advancing at the speed they're advancing, maybe somebody could leapfrog Nvidia in the AI server realm. It's not out of the realm of possibility given how much money's thrown out. They got a tremendous lead, but here is a huge hedged bet here. Maybe they can also be the local AI provider, the core of your experience whenever you touch the device in the first place.
Interesting. And I do think that is a good outcome for the world that we do more AI processing on the edge on our devices. Like there are just a lot of reasons that's a good thing. And I think a lot of people are hopeful that that's where this lands. That rather than have everything happen in data centers and in the cloud, that actually your device just starts to do things.
What I am struggling to figure out though is if I'm Nvidia, right, and I have this big giant goal to sort of own the whole AI future, which I think it does. I don't think that is a particularly huge overstatement for what Nvidia is trying to do right now. At what point does Nvidia just make Windows, right? Like I'm looking at this and it's like, okay, Nvidia is launching a chip and it's like, okay, if you're going to launch a chip because you want to be in laptops because you think normal everyday laptops are going to become AI machines.
And I think you're right that Nvidia does think that. And Jensen even talked about, you know, there's this new interface that's coming. We're going to interact with our computers in different ways. At some point, they're going to make Windows, right? Like if the play is what you're describing, they're going to have to try to be Apple and just make the whole thing from the inside out, right? Like is this... I don't know. This feels at some point like Nvidia is just trying to own the whole thing.
I feel like they could. I don't know if they want to. They have one thing that Microsoft does tremendously well for everybody in that ecosystem is they are the glue that pulls all the developers together and says, 'Hey, we're all pulling together here.' Nvidia also famously pretty good at that. But I would say that Microsoft, if you want to make sure that you're the Adobes of the world and the Epic Games of the world and so on.
If you have the power of Microsoft to say we are the OS, we are the Xbox company, we are the games company, we have all the relationships that we've spent all this time building. It's maybe easier at least in the short term for Nvidia to lean on that and say, 'Yeah, there's a platform here we can build on that.' But I don't think it's out of the question if they wanted to make, you know, an operating system and own it completely vertically like Apple, why not?
Particularly because then they can put that in the cloud, they can deliver that to your device remotely as well as doing it locally. Why not? I don't think we've heard them state that, but we've certainly, you know, like you said, they're trying to dominate every angle of it.
So all of that actually makes total sense to me. And I think starting with the chip obviously is what you do if you're in Nvidia. That's its core competency. Like, of course, that's where you'd start. What do we and don't we know about this chip at this point? I think to your point about it just sort of appeared and they said a name and we don't have a lot of information about it. Nvidia gets a lot of credit just for being Nvidia. There is this sense that like because Nvidia did it, it's probably very good. My sense is we don't have a lot of details on what this actual family of laptop processors is going to be. What do we know at this point?
The super funny thing about the RTX Spark superchip is that it already exists and it was already released in a device by Nvidia called the DGX Spark.
That was their little... they called it like an AI supercomputer or something. I just... you just sparked this memory of like it's... we have a picture of it behind glass and they're like tiny... it's a tiny gold box, like tiny. I don't know why it's gold.
But it was very gold. I don't know if they thought Trump was going to come by that day or something, but anyhow, tiny gold box with a big ARM chip and a substantial chunk of like laptop mobile graphics power, GeForce graphics power in there. And what they did originally was they just released it as like a $4,000, $5,000 AI box and they said, 'Hey, run your local models on it.'
What happened around the same time as that though, maybe after they announced it, before they released it, was AMD decided that they were going to also build a superchip, quote unquote. They didn't call it a superchip, whatever. Called the Strix Halo or the Ryzen AI Max platform, Max Plus. And so this thing, like AMD's chip has a crapload of memory, 128 gigs of memory on the chip. Gobs of memory inside this AMD chip.
And they made it an application processor. They put it in little desktops for consumers, like the Framework Desktop. They put it in handhelds. I tested the most powerful handheld ever made. There's several of them now, but the one I tested was called the GPD Win 5 with Strix Halo. It's like 1080p PlayStation 5 level gaming in the palms of your hands.
And Nvidia, I'm guessing maybe they were already planning it, but Windows wasn't ready for it. Maybe they saw what AMD was doing and said, you know, well, we could have that too. We just got to take this chip and turn it around. And so what they've announced today is the RTX Spark. The highest-end RTX Spark is literally that same spec. It is 6,000-plus GPU cores. It's 20 CPU cores. It's made on a TSMC 3-nanometer process.
MediaTek is Nvidia's partner, so you can't just credit Nvidia here. This is a MediaTek chip with Nvidia IP on it, manufactured by TSMC, and they're not saying where, so we have to assume not here in the United States or they would make a big deal with that.
Surprise. 128 gigs of that memory, unified memory, so everything's super fast and that memory can go to whichever part of the system that needs it at that speed. And it has that chip like immediately sort of leaps to the tippy-top end of consumer-level chips, right?
It might in terms of a monolithic thing if it's like one chip on a board, but it doesn't have the most powerful graphics you can get in a laptop. Just the most powerful graphics you might be able to get on a single chip design.
Okay, so in a beefy gaming laptop, you could have an RTX 5090 mobile processor in there, which would be super, you know, balls-to-the-wall gaming graphics.
Whereas in this version, we're talking about more like RTX 5070 mobile, so a couple tiers down. That might be the fastest mobile graphics we've ever seen. It might not. We haven't tested. It certainly isn't as high as you could possibly get if you had a couple chips in a big beefy laptop. But their pitch here is you have all this performance and it is very thin and it's very light and it's very cool because it's an ARM CPU and it is monolithic here on a single chip that does all this stuff.
