Sean Hollister | The VergeThe Verge is about technology and how it makes us feel. Founded in 2011, we offer our audience everything from breaking news to reviews to award-winning features and investigations, on our site, in video, and in podcasts.2026-10-22T23:10:21+00:00https://www.theverge.com/authors/sean-hollister-2-2-2-2/rsshttps://platform.theverge.com/wp-content/uploads/sites/2/2025/01/verge-rss-large_80b47e.png?w=150&h=150&crop=1Sean Hollister<![CDATA[The next-gen Xbox won’t cost you $1,800. It can’t. Can it?]]>https://www.theverge.com/?p=10161342026-10-22T19:10:21-04:002026-10-22T19:00:00-04:00
An Xbox Elite controller. | Photo by Antonio G. Di Benedetto / The Verge
My first reaction: The next Xbox is dead on arrival. It’s DOA.
My second thought: Microsoft knows people won’t pay that much. It couldn’t possibly charge that much money if it wants next gen to be a success. “I wouldn’t release it at all,” former Microsoft exec Mike Ybarra recently said.
My third thought: Didn’t Microsoft alreadysay out loud that it knows people won’t pay that much?
Didn’t new Xbox boss Asha Sharma say “We’ve reached a point where it will be hard to imagine that mass audiences can afford thousands of dollars to spend” on a box?
I don’t think Fortune’s insiders are the problem here. With today’s incredibly overpriced RAM and storage, it wouldn’t be surprising if it cost over a thousand dollars to build, package, distribute, and sell “the largest technical leap you will have ever seen in a hardware generation,” as former Xbox president Sarah Bond called the next Xbox back in 2024.
But there are at least five ways the report could be correct without making you pay all that money at once. I just don’t think you’re going to like most of them.
Subsidies: Microsoft could pay to give you a discount
The least-likely possibility is the one Fortune suggests itself. Let’s look at the original quote, bolding mine:
[…] the price of the Helix console has internally been projected to cost consumers between $1,200 and $1,800, unless Microsoft decides to subsidize the cost.
Historically, Microsoft has subsidized the Xbox. Microsoft sold boxes at a loss right out of the gate, making its profits when you buy games and Xbox Game Pass subscriptions instead of on the hardware alone. That’s not our speculation — it’s exactly what Microsoft told The Verge in 2021. In 2023, we learned that Microsoft may have spent $1.5 billion to make sure it could sell the Xbox Series X and S at $499 and $299 respectively, instead of what they actually cost.
But Microsoft has suggested that it’s done with hardware subsidies. “Going forward, this cannot continue,” wrote Sharma in her “Xbox reset” memo, referring to the $20 billion in content, platform, and hardware subsidies Microsoft made over the past five years.
Microsoft CEO Satya Nadella echoed that idea on the Hard Fork podcast: “No one can accuse Microsoft of not having invested for the last 25 years, and now we have to turn this into a sustainable business.”
It isn’t just new management that’s hinting at no more subsidies: Bond primed the pump for a $1,000 box when she called Helix “a very premium, very high-end curated experience” late last year, comparing it to gaming PCs like the $1,000 Xbox Ally X handheld.
Perhaps Microsoft has changed its mind since then. “We are working very hard to rethink everything that we can about Helix,” Ball said in June, referring to affordability. “The plan’s the plan until it’s not the plan,” Sharma said in February (though not about affordability specifically).
But subsidies would be a tremendous risk if the Helix still plays PC games in addition to Xbox ones. Valve’s Steam is by far the dominant PC gaming platform — so PC gamers who buy a Helix won’t help Microsoft recoup its money.
Subscription: “Get it free with 36 months of Xbox Game Pass and qualifying trade-in”
There’s another way Microsoft could subsidize a $1,200 Xbox.
Today, AT&T will give you a $1,200 iPhone 18 Pro for $0* — if you lock yourself into 36 months of AT&T phone service that costs $100 or more per month and trade in a recent phone, too.
Microsoft could do the same with a new, pricier tier of its Xbox Game Pass game subscription service, effectively bundling the games with the console and making you slowly pay for both at the same time. Perhaps you’ll have the option to buy an “unlocked” Xbox for full price, but it won’t be the norm.
To make it happen directly, Microsoft would need a pricier new tier of Xbox Game Pass, which wouldn’t necessarily align with the company’s recent moves. Sharma actually cut the price of Game Pass Ultimate to $22.99 a month in April, writing that “Game Pass has become too expensive for players,” while removing Call of Duty games and unlimited access to Xbox Cloud Gaming from the deal.
I don’t know if it would work. People feel naked leaving the house without phone service. They might not feel the same about games. Subscription fatigue is real. But if it worked, Microsoft’s moat would be stronger than ever. You’ll be more likely to play the games you’ve already paid for and are committed to keep paying for. Many would get sucked into a profitable upgrade cycle, just like they do with phones, since they can buy the next Xbox for “$0” as soon as it comes out.
Cellphone subsidies have been around nearly as long as pricey phones: The 1994 IBM Simon, arguably the first smartphone, eventually cost half its $1,100 price tag if you signed a contract with BellSouth. My first phone, a Kyocera QCP-6035 with a built-in PalmPilot, was subsidized when it arrived in 2001. So it’s not exactly a “radical new business model,” but it would be new for game consoles.
Rental: Maybe most people won’t own a Helix themselves
It’s also possible that Microsoft simply doesn’t plan to sell a console to the masses anymore — it can sell an $1,800 Xbox to those who can afford “video game ownership” and try to hook everyone else on the Netflix model.
After all, the “K-shaped economy” and RAMageddon have already made gaming PCs and existing game consoles into luxury goods. Society’s wealthier individuals seem perfectly happy to pay inflated prices for the few consoles Sony and Microsoft still pump out.
“I can tell you definitively, demand for our console exceeds supply. We are putting them in as many stores as possible, we are producing them as quickly as possible. There is a severe limitation to how quickly we can do that,” Xbox chief strategy officer Matthew Ball told The Game Business in June, blaming RAM shortages.
But Microsoft has another way of giving less affluent people access to “next-gen” Xbox power. It can rent you one. The company just launched pay-as-you-go cloud gaming starting at $3.99 for 10 hours of cloud gaming (50 hours for $11.99).
Maybe all of Microsoft’s recent talk about affordability isn’t about the masses owning a next-gen console. Maybe it’s about finding an affordable way for people to “play Xbox” without owning the games at all.
The next gen is the current gen
After Fortune’s report, some of the very first reactions I saw boiled down to this: “I’m out. Forget next gen, I’m going to stick with my existing Xbox/PlayStation/Switch.”
That’ll be the reality for manypeople if the Xbox costs over a thousand bucks. But if the RAM shortage continues — if Microsoft knows it’ll sell every Xbox it can make to wealthy individuals because it can’t make a lot — it’s not necessarily a huge risk to sell it sky-high on day one.
Perhaps the Xbox Helix will be an early adopter console, the same way the iPhone Duo is an early adopter phone, the same way the Steam Frame is an early adopter headset. These products are selling out instantly despite their high price tags, leaving many would-be buyers in the cold. They’re shipping messy and unfinished, but that’s okay because early adopters want to come along for the ride.
Microsoft knows the same is true of some Xbox buyers, because it helped ship an unfinished $1,000 Xbox Ally X handheld even before the RAM crisis, and the $1,300 Xbox Ally X20 follow-up is already sold out.
But if Microsoft splits the Xbox audience into “have” and “have-not,” it means the next generation won’t truly arrive. Game developers won’t spend all their effort targeting a console people can’t buy. The next Xbox won’t be any more necessary than a PS5 Pro if you already own a PS5. In industry lingo, the Helix would be a “mid-cycle refresh” instead of true next gen.
I wouldn’t hate that, though, and neither would many video game execs. They’ve long complained that the pace of technology has ballooned the cost of building a video game to unsustainable levels, leading to today’s world where too many development teams are cut to the bone right after their games ship.
Former PlayStation and Xbox execs like Shuhei Yoshida, Shawn Layden, and Mike Ybarra have argued that the next gen isn’t necessary.
“Stick with PS5 Pro/Series X until prices normalize. The capabilities of those boxes are both fine for 99% of the audience. No one is screaming ‘I need a more powerful console’ right now,” Ybarra tweeted on October 20th.
“I think [Sony and Nintendo] should try to keep the current generation as long as possible, because I think people are very happy using PS5 and Switch 2,” Yoshida told me on The Vergecast.
“I think we’ve plateaued on the tech, frankly,” Layden told GamesIndustry.biz in 2025. (He’s said similar before and since.) “Make it cheaper and simpler. Let’s do that instead. And let’s find a way to have more hardware companies participate.”
Even if Microsoft ships a $1,800 console, the company’s been clear it won’t be the only Xbox you can buy. Sharma has publicly said that Helix will be a “family of devices,” which has been broadly interpreted to mean that Microsoft will let PC industry partners like Asus sell their own “Xbox” as well.
If the reporting pans out, that wouldn’t just be a cobranded Xbox Ally handheld that runs Windows — Asus could make a console-shaped box that runs a Windows-based Xbox OS, too.
Some of those partners could offer more affordable, less powerful Xboxes than Microsoft’s own flagship box, perhaps with less storage and memory. Microsoft, too, might have creative solutions for storage, like how the company shipped the Xbox 360 with different-size hard drives, or none at all, to make it more affordable. “We will have to empower customers to have very flexible storage offerings,” Sharma said in June.
Or you could buy another Xbox that’s likely going to continue to count as part of Microsoft’s “family”: the existing Xbox Series X and Xbox Series S.
This can all work because games don’t need to target a single console anymore; they haven’t needed to for a long time. An Xbox One game runs on an Xbox One X and an Xbox Series S and an Xbox Series X because they’re all effectively AMD-powered Windows PCs, just with different amounts of performance on tap. The Xbox Helix is a powerful extension of the same continuum. The same is true of PlayStation 4, PS4 Pro, PS5, PS5 Pro, and presumably PS6, minus the Windows part.
Perhaps a particular piece of hardware will stand out as the one developers should target as the baseline for “it looks and runs great on Xbox,” the same way the Steam Deck handheld became a baseline for PC game developers starting in 2022. Maybe, like the Steam Deck and the Nintendo Switch, it’ll help reset our expectations of what a top-shelf game needs to look like.
At this rate, I wouldn’t be surprised if the existing Xbox Series S will be the one. But it could also be Microsoft’s own next handheld Xbox. Fortune’s insiders say it’s coming holiday 2027.
]]>
Sean Hollister<![CDATA[The FCC is beginning to block gadgets with ‘foreign surveillance chips’]]>https://www.theverge.com/?p=10170162026-10-22T16:23:48-04:002026-10-22T16:23:48-04:00
On October 13th, Brendan Carr’s FCC gave itself a new power — it can block gadgets for containing any computer chip made by any company on its list of potential threats to national security. Today, it’s exercising that power for the first time, blocking two gadgets because they contain “foreign surveillance chips,” the FCC claims.
The agency doesn’t say which specific gadgets it blocked, and it didn’t immediately respond to our request for comment, but it does name the “surveillance chip”: it’s a HiVision Hi3516CV610 UHD Smart Vision SoC, a chip that turns a raw camera sensor into a home security camera, a dashcam, or the like. It’s often sold as a package with a Sony camera sensor and sometimes even a lens already on board, according to listings we viewed on AliExpress.
The FCC is calling it a foreign surveillance chip because HiVision advertises it as “designed for the surveillance market” with “[b]uilt in detection algorithms for human faces,” the FCC explains — and because HiVision is a subsidiary of Huawei, one of the first telecom companies the US government decided to call a national security risk.
Huawei is indeed on the FCC’s Covered List, and the FCC’s new power helps it close a loophole where companies could sneak in banned companies’ products inside another company’s gadgets. It did something similar when it gave itself the power to retroactively ban gadgets, and when it gave itself the power to block modular radio transmitters.
As with the router ban and the Roomba ban, he question remains: is the FCC keeping people safer, or just picking winners and losers while making a show of enforcement? We’ve asked the agency many times to chat about its efforts, but it won’t engage.
]]>
Sean Hollister<![CDATA[Eiliko maker says it’s hiring an ‘independent security firm’ after Verge report on its insecure robots]]>https://www.theverge.com/?p=10158242026-10-22T11:21:54-04:002026-10-22T07:00:00-04:00
The Eiliko robot toy. | Photo by Sean Hollister / The Verge
Eiliko, a cute robot charm you can hang on your bag with built-in AI conversations, turned out to be a security nightmare. It told owners that “your secrets are safe with me,” but hackers could easily read complete transcripts of everything we told the toy, and that was just the beginning. The Verge and security researcher Sammy Azdoufal spent over a month trying to get in contact with the company, including emails to Energize Lab CEO and founder Shaw Yeung. Nobody would answer our questions.
Now, the company is finally breaking its silence, though its response leaves something to be desired.
“We have confirmed the core issues identified in your reporting and are addressing each of them,” Energize Lab told The Verge in an unsigned statement from a generic email address, which it later told us is attributable to CTO Du Xiaoyu. The company says it will hire an independent security firm to fully review all of its systems, and has set up a dedicated security@energizelab.com contact email.
The company also claims that as of Wednesday, it has “closed all external access to our backend systems” and “disabled the interface that allowed a robot’s responses to be modified remotely.”
Energize Labs is also pledging to “no longer store conversation records or the user profiles generated from them,” and says it will delete the conversation records, user profiles, and “avatar images” it has already stored.
Notably, the company isn’t promising to get rid of data collection for good. “The long-term memory and custom avatar features will not return until that review confirms they are secure,” it writes, suggesting that Energize Lab does plan to store some personal information once its systems are hardened.
Even more notably, Energize Labs’ systems are not yet secure. With Azdoufal’s help, I can confirm that as of the company’s email, it was still storing audio recordings and photos of its customers at unsecured public web addresses with no protection whatsoever. If I gave you a link (which I won’t) you could click and immediately be looking at an owner’s face.
Azdoufal also tells me he still has full SSH access to these robots, and that he can still theoretically read transcripts and modify the toy’s behavior. So when Energize Lab says it has “closed all external access,” that doesn’t seem to be accurate.
“We know that words alone will not earn that trust back,” Energize Lab tells The Verge. “We intend to earn it through what we do in the coming weeks, and we will keep you informed as we go.” That sounds like a good plan, because some of today’s words still don’t ring true.
Here is Energize Lab’s full statement to The Verge, attributable to CTO Du Xiaoyu:
Dear Sean,
We sincerely apologize to you and to Mr. Azdoufal for not responding over the past month. This was our failure: After receiving the relevant information, we failed to sync with the technical team in a timely manner, and the subsequent troubleshooting also took some time. We have now set up a dedicated security inbox, security@energizelab.com, which is staffed and monitored.
We thank you and Mr. Azdoufal for bringing these issues to our attention. We have confirmed the core issues identified in your reporting and are addressing each of them.
As of today, we have:
Closed all external access to our backend systems and rotated all access credentials;
Disabled the interface that allowed a robot’s responses to be modified remotely;
Taken the long-term memory feature offline. We no longer store conversation records or the user profiles generated from them;
Taken the custom avatar feature offline. The app now offers built-in avatars only.
We will delete stored conversation records, user profiles, avatar images, and related content.
We are engaging an independent security firm to conduct a full review of our systems. The long-term memory and custom avatar features will not return until that review confirms they are secure. We will share an update on our progress once the review is completed.
We started Energize Lab because we wanted to build small robots that bring a little joy into people’s lives. That idea only works if people can trust what is inside them, and we let them down. We know that words alone will not earn that trust back. We intend to earn it through what we do in the coming weeks, and we will keep you informed as we go.
Sincerely,
Energize Lab Team
Update, October 22nd: Energize Lab says the statement is attributable to CTO Du Xiaoyu.
]]>
Sean Hollister<![CDATA[Will TP-Link really make routers in the US? Here’s what it told us]]>https://www.theverge.com/?p=10138072026-10-21T13:54:56-04:002026-10-21T14:00:00-04:00
A TP-Link Wi-Fi 8 router. | Image: TP-Link
But TP-Link is willing to do something those companies weren’t — to invest nearly $1 billion to make routers in the United States, creating “hundreds of high-paying research and development, manufacturing, and construction jobs for American workers over the next four years,” according to a letter TP-Link sent to the FCC.
This week, it announced a letter of intent with US contract manufacturer Key Tronic to begin “domestic assembly” of its Wi-Fi 8 routers in Corinth, Mississippi. It could begin producing routers “as early as next year.”
Intriguingly, this mightnot just be slapping boards into a plastic box.
“If you look at what we do with the products in Vietnam, you’ve got raw components coming in and you’ve got completed, tested, shrink-wrapped products coming out,” TP-Link corporate affairs officer Steve Kovsky tells The Verge. “You’ve got surface-mount technology, many layers of testing quality control, we’re doing plastic injection molding, and as far as I know that is what we intend to do in Mississippi.”
But how likely is that “letter of intent” to turn into a true manufacturing plant? After talking to Kovsky, it sounds like the stars might need to align.
What is TP-Link waiting for?
A deal with the Federal Communications Commission. Kovksy confirms the offer hinges on the FCC granting that router ban waiver — formally known as “conditional approval” — before it can proceed.
Kovsky wouldn’t tell me how long ago TP-Link applied for that waiver, but it sounds like the FCC hasn’t yet given the company any idea whether it’ll be approved. TP-Link had to provide lots of info about the company’s structure, ownership, and supply chain, but Kovksy confirms the FCC hasn’t asked any follow-up questions about other topics, and he says the FCC doesn’t exactly tell companies when it denies conditional approval.
“We’re looking at offering total transparency to these agencies, and we’re open to negotiations,” he says. “Nothing is written in stone.”
Why does TP-Link need an import waiver if it’ll be building routers in the US?
“I have asked that question myself,” Kovsky tells The Verge. The short answer is components. Even though TP-Link buys its chipsets from US companies like Qualcomm and Broadcom, those companies don’t actually produce those chips in the US. “Do they count as US product?” Kovsky rhetorically asks.
“It’s not a lock, and we would rather have [the waiver],” he says.
Why might the FCC hesitate?
Five states have now sued over TP-Link’s alleged ties to China, and many more are directly asking the FCC not to grant a waiver, claiming the company dominates the consumer router market, that it’s a Chinese company posing as a US one, that the routers have brought serious security vulnerabilities into US homes, and more.
Kovksy suggests it’s all based on misinformation, even the company’s market share: “It’s about 9 to 10 percent, which is like top three. It’s a good number, but we do not have an overwhelming share of the US market.” TP-Link’s website clarifies that it has 10 percent market share when you count all the routers that internet providers put into people’s homes.
“We do not sell customer data or share it with any government ever. We never will. We do not allow backdoors in our products. We are not susceptible to malware or vulnerabilities to an extent that’s greater than any other manufacturer in this category to my knowledge,” Kovsky says.
If it gets the waiver, does TP-Link intend to produce all US-bound routers in the US?
TP-Link is not willing to commit to that yet. “I don’t think that’s entirely determined,” Kovksy says. “Some of that is going to be: What are the FCC and the administration really trying to achieve?”
The administration already has a stated goal, though: It wants Made-in-USA products where at least 65 percent of their components are also made here. Kovsky says that TP-Link would be “happy to comply with that, to the extent that it can continue to produce products that meet its criteria for quality and affordability.”
“That is going to be challenging, certainly in the short term, [because] those components for the most part just are not produced in the US,” Kovsky says.
He says TP-Link believes components will become more available and affordable if the US pushes other manufacturers to invest, but the waiver isn’t the only factor in its decision.
If it gets the waiver, will TP-Link invest, or will it need to find an inexpensive source of components or other things before it can go ahead?
“I think those are probably going to be business decisions that the CEO will ultimately have to make,” Kovksy begins. “Well, I hope he doesn’t have to make those decisions, frankly, that everything falls in line. He is very committed to making this investment,” he says.
Notably, TP-Link would not confirm Kovsky’s suggestion that it might completely manufacture routers in the US and only source the raw components. I told the company that if I could say TP-Link was committed to anything other than “final assembly,” it’d be well worth sharing with readers.
“At this stage, TP-Link is not sharing specific job numbers or detailed technical details of the assembly process beyond what’s in today’s press release,” third-party spokesperson Ricca Silverio told me.
I don’t think this is Foxconn in Wisconsin all over again because TP-Link probably won’t get its conditional approval, let alone invest in empty buildings. Foreign gadget bans from President Trump and FCC Chairman Brendan Carr seem explicitly designed to target companies China hawks claim to be worried about, like DJI or Unitree, while letting favored companies through. But the hawks, including a coalition of congresspeople led by Sen. Tom Cotton (R-AR), seem to hate TP-Link, too.
]]>
Sean Hollister<![CDATA[Antigravity just halved the price of its drone — by swapping its headset for wristwatch control]]>https://www.theverge.com/?p=10140632026-10-22T12:09:55-04:002026-10-20T09:00:00-04:00
The Antigravity A1, with the new Follow Me Kit. | Photo by Sean Hollister / The Verge
Last year’s Antigravity A1 is the best beginner drone I’ve ever tried. I put a friend who’d never flown into its virtual cockpit and didn’t worry one bit. The A1’s also nice for people who’d prefer to think about flying rather than filming, because you can reframe its 360-degree footage after.
But until today, that meant wearing the company’s FPV goggles and spending upwards of $1,599 on the complete set. Now, Antigravity is halving the price of entry to $799 in the US with a new “Follow Me” kit that replaces the original headset and motion controller with a wristwatch and ring instead.
The idea is that Antigravity can be your robot cameraman, automatically flying, following, and filming you with a minimum of gear.
I took the new kit for several test drives this fall, and I’m cautiously impressed.
Instead of powering on a headset and finding a safe place to sit while you block out the world, you can now pull a drone and a watch out of your bag, power them both on, hold a button to launch the drone from your hand, then walk or ride while the Antigravity A1 automatically films and follows you, dodging obstacles along the way.
You can even wave your arm like a magic wand to nudge the drone left, right, down, up, towards, or away while it’s filming. You get a swipeable preview of what the drone sees on the watch’s touchscreen, and can select various quickshots there or summon a couple of sliders (forward/backward, ascend/descend) to manually fly the A1 out of a jam. The ring doesn’t do anything you can’t already do with the watch; it’s just an extra button so you can launch and reposition it one-handed.
As I discovered after three crashes, the A1 is not yet intelligent enough to dodge everything or stay with me through thick and thin. I took it all around my neighborhood on a kick scooter, and you might want to be careful around power lines, thin branches, or the vertical guy wires holding utility poles in place.
There’s also a lot of rough edges I hope Antigravity will fix, like the sluggish experience swiping around the touchscreen, or how it won’t land on my hand without switching back to manual mode, or how I can’t get it to automatically start tracking and filming me — I always have to find myself on the watch screen, tap on myself, then tap another button to begin. I also ran into an issue where it wouldn’t follow me once I crested a high hill, because it thought it was flying at too high an altitude.
Most of all, I hope Antigravity will figure out how to keep the drone from drifting out into the street or over a neighbor’s fence when I tell it to follow right behind. It’s far smaller than me, so anywhere I go, I feel like it should fit!
But as you’ll hopefully see in my video higher up this story (also on Instagram/TikTok) it’s not bad at following me through a dense suburban environment! None of my crashes so far have crippled the drone, and I like how the watch automatically plays the last seconds of footage from before a crash. It made it very easy to find the drone after.
Antigravity’s new full-coverage propeller guards can also help in case of impact; I saw the drone bounce right off one obstacle and keep flying. They’re included in the $799 set, though existing Antigravity A1 owners can buy the accessories too: it’s $340 for the watch and ring, and $40 for the guards, with a 20 percent discount for those who can prove they already own the aircraft.
Antigravity’s A1 is kind of like the folding Skydio I wish that company had made before it left consumers behind to serve industry instead. The self-flying Skydio 2 similarly had a few crashes in my tests and wouldn’t always stay in its lane. But I’ve long wanted a folding version I could easily stick in a bag, and the Antigravity A1 seems like the next best thing.
]]>
Antonio G. Di BenedettoSean HollisterTom Warren<![CDATA[A week with RTX Spark: everything we’ve learned using Nvidia’s new laptops]]>https://www.theverge.com/?p=10127482026-10-16T11:26:47-04:002026-10-16T11:00:00-04:00
Nvidia’s new SoC is available in Windows laptops from six different manufacturers. We’ve collectively tested four of them so far. | Photo: Antonio G. Di Benedetto / The Verge
Nvidia’s RTX Spark N1X processor is finally here, and we’ve been collectively testing a small fortune’s worth of powerful laptops sporting this new “superchip.” RTX Spark is designed to be a high-performing Arm-based SoC for laptops and small desktop computers, rivaling the best from Apple, Intel, AMD, and Qualcomm. It has loads of cores, Nvidia RTX Blackwell graphics, and enough power and unified memory to run sizable local AI models.
The flagship placement is inside Microsoft’s Surface Laptop Ultra with a staggering 128GB of unified memory, which our own Tom Warren fully reviewed. Sean Hollister has been testing a 64GB version of the Asus ProArt P16, which I’ve also been testing in its top-spec 128GB config. Since I’m the most spoiled of the lot, I also have a 64GB Lenovo Yoga 9n 2-in-1 and a 32GB HP OmniBook Ultra in for testing.
All of these laptops have the same N1X chip with 20 CPU cores and 6,144 GPU CUDA cores, and they’re all wildly expensive — ranging from the HP OmniBook at $3,199.99 to the Asus ProArt’s $6,999.99. There’s an 18-core / 5,120 CUDA N1X chip in lower cost configs, but that only gets as cheap as $2,599.99 for a Surface Laptop Ultra or a smaller Asus ProArt P14 with 24GB of RAM.
These are big and expensive laptops built on new chips. They’re meant for developers, coders, and content creators who might also want the ability to use local AI and even play games using the latest Nvidia tech like DLSS 5. There’s a lot to test, and we have not had a ton of time to test it all. So below you won’t find an exhaustive scored review of any one laptop, but instead our impressions and thoughts from three writers using five laptops for a week.
Is Nvidia’s “superchip” really all that super? Are Apple, Intel, AMD, and Qualcomm officially on notice? Have any of us gotten any sleep?
Does it just work?
Antonio Di Benedetto: Windows on Arm is no longer a wholly subpar experience like it was a few years ago. Thanks in large part to the push around Qualcomm’s Snapdragon X chips, lots of development has been put into native Arm-based apps for Windows. And many x86 apps work fine via Microsoft’s Prism emulation layer.
Adobe has had native Arm versions of Photoshop and Premiere for some time now, and even though my precious Lightroom Classic is still an x86-based app I don’t notice any issues using it on Arm. At this point, you’ve gotta be relying on specialized software that’s exclusively x86 based for Windows on Arm to be a non-starter — or you want to freely play any games out there without any worry of incompatibilities.
Sean Hollister: Every so often, there’s that moment when an app takes longer to load than it should, and I wonder whether it’ll load at all. Even then, it typically loads. Occasionally, it takes more than one try. My Evernote notes disappeared a few times one day when I alt-tabbed away, but a force refresh fixed that. And apps don’t seem to respect my system preference for dark mode.
But I largely agree with Antonio. I made an RTX Spark my daily driver, with all my typical work and personal apps — Signal, Discord, Audacity, CapCut and Evernote, Irfanview, Slack, ScreenToGif and more — and nothing has failed me so far. I never once needed to think whether I was running an Arm or an x86 app. The real hitches came with gaming and AI, as you’ll read below.
Tom Warren: I’ve been using a Windows on Arm laptop with a Qualcomm chip daily for the past two years and things have improved to the point that I can’t tell if an app is native or not anymore. That’s the same story on the RTX Spark laptops, and I feel like Nvidia’s entry into this category will only improve the story for creative apps and gaming.
Can it game?
TW: Windows on Arm never looked like a serious gaming platform before now. While there still aren’t many native Arm games, Nvidia’s GPU helps bruteforce through the emulation layer. Your mileage in gaming is going to vary depending on the title, but for the most part I’ve found that medium settings, running at 1440p and DLSS performance, will get you to a stable 60fps in most games.
I mostly game on my desktop, where I’m spoiled by using an RTX 5090 on a 32-inch 4K screen every day with maxed-out game settings, but I didn’t mind the look of medium settings on a 15-inch display. I was able to run Black Myth: Wukong at 70fps on average at native resolution (3270 x 2180) with medium settings and DLSS performance mode enabled. I ran the same medium settings and DLSS performance in Cyberpunk 2077 and averaged 66fps, and Forza Horizon 6 managed an average of 57fps.
In some titles you may need to drop the resolution or settings lower, though. This definitely isn’t as good as a dedicated gaming laptop, but for something that’s roughly on the level of a MacBook Pro it’s more than good enough for gaming on the go.
SH: Like Tom says, the GPU power is there — but the CPU might hold back some of your games. I had plenty of power for a smooth 60fps in Returnal (at 1440p with high settings), Expedition 33 (at a DLSS “Performance” upscaled 1200p with high settings), Jedi Survivor (DLSS “Performance” upscaled 1200p with medium settings), and Shadow of the Tomb Raider (at native 1800p and highest spec).
But when I fired up Helldivers 2, even dipping down to a DLSS-upscaled 1080p was no guarantee of 60fps in the highest difficulty missions. There are just too many enemies on screen for the Spark to keep up. Star Wars Zero Company is even more CPU-bound— the Spark didn’t care whether I was maximizing or minimizing the graphics and screen resolution. No matter what I tried, I typically saw my framerate capped between 37fps and 39fps in battle.
Even ancient and low-powered games can have quirks. I tried the 2003 title Max Payne 2 and saw situations where framerate dipped to 50fps. I even saw a few hitches in the excellent UFO 50, a collection of 2D games, though a couple other 2D games ran great.
I’m also not sure if RTX Spark is suitable for VR yet. The excellent Titanfall 2 VR mod worked, and it seemed to run fine at first, but lost a little stability after a few minutes. That could be my Wi-Fi, because the driver for Valve’s Steam Frame wireless adapter wouldn’t install on the Spark. Neither did Steam’s streaming audio solution, so the audio came out of the Nvidia laptop instead of the headset.
Just so we’re clear, all of this is assuming you want a laptop maxing out its fans and positively dumping heat on your desk or lap. That’s not unusual for thin gaming notebooks, but it’s way louder and hotter than a good desktop or handheld.
2560×1600
2560×1600
2560×1600
2560×1600
Asus ProArt P16 RTX Spark (20-core, 6144 CUDA, 128GB RAM)
HP OmniBook Ultra RTX Spark (20-core, 6144 CUDA, 32GB RAM)
Framework Laptop 16 5070 (2025, 8GB VRAM)
MacBook Pro M5 Max
Black Myth: Wukong TSR
27
27
28
Not compatible
Black Myth: Wukong DLSS
41
44
45
Not compatible
Black Myth: Wukong DLSS FG
53
59
71
Not compatible
Cyberpunk 2077 Ultra
41
46
57
56
Cyberpunk 2077 Ultra RT & DLSS quality
30
35
44
42 (MetalFX Quality)
Cyberpunk 2077 Ultra RT & DLSS Frame Gen x2
53
56
68
75 (MetalFX Quality, FSR 3.1 FG)
Cyberpunk 2077 Ultra RT & DLSS Frame Gen x4
92
103
Not tested
Not compatible
Average FPS results in gaming benchmarks we ran on the Asus ProArt P16 and HP OmniBook Ultra, against an RTX 5070 Laptop GPU and M5 Max MacBook Pro.
AD: I’ve wanted a MacBook Pro-like laptop with full PC gaming chops for as long as I can remember, so I was hyped for these laptops to deliver the goods in graphics. The first night, I set up the Asus ProArt P16, with all 128GB of its ridiculously large pool of memory, and when I booted up Big Walk for my planned online evening with some friends my first thought was, “Well, that looks okay. I guess.” I know Big Walk isn’t some triple-A graphical titan, but it has some fairly demanding graphics for the “friend slop” genre. I expected the N1X to max it out, or at least get close, but not really.
I had to bump the P16’s 4K display down to 2560 x 1600, and instead of Big Walk’s Highest or Very high graphics presets, I brought it down to High to maintain a decently solid 60-ish fps. Does any of that really matter for this silly co-op game about solving puzzles with your buddies? Of course not. It looked fine, after all. But it was a sobering reminder that this “superchip” is no graphical magician.
As Tom and Sean point out, RTX Spark can do some decent gaming, but temper your expectations for medium or even low settings in some cases, and be ready to rely on DLSS and frame gen. There’s no denying that cheaper desktops and even laptops with discrete Nvidia graphics cards can easily beat out an N1X. Nvidia claims it’s like having an RTX 5070 Laptop GPU in an SoC, but in my testing it was somewhere between that and an RTX 5060 Laptop GPU — not a slouch, for sure, but if you’re a gamer first your money goes much further with traditional Nvidia graphics cards.
How’s the battery life?
SH: Nvidia’s partners saw fit to equip their first wave of RTX Spark laptops with some of the biggest batteries you can bring on an airplane with no questions asked — but even with a 99.9 watt-hour pack like the one in my Asus ProArt P16, these laptops won’t be winning awards for longevity. While Tom and Antonio spent their days testing every use case under the sun, I made the P16 my daily driver for normal work while watching the battery meter like a hawk. That mostly means writing stories in a web browser, and it’s not quite lasting the day.
My best run was 5 hours and 40 minutes of real-world work at 50 percent brightness, including an hour-long video call with Tom and Antonio to strategize our RTX Spark coverage plan. The next day, I got only about 5 hours from a full charge because I spent two of them editing this video in Capcut. The third time I drained the battery, it took 5 hours, too.
With web work, I’m typically seeing 13-17 watt battery drain on the laptop’s default settings, but that climbs to 20-30 watts with light video work, up to 50 watts during the (brief) Capcut video export, and I made it jump all the way to 120 watts playing Indiana Jones and the Great Circle. The battery couldn’t keep feeding at that rate — it slowly went down, and at 30 percent charge I saw a big performance hit as it dropped dramatically to 60 watts.
The good news is the Asus charges fast from a 140W USB-C PD power source. It took 20 minutes to go from dead to 50 percent battery, another 15 minutes to get to 70 percent, and a total of 1 hour, 15 minutes to reach 95 percent charge.
AD: I’ve had to run lots of benchmarks while plugged in, so I haven’t been able to spend as much time testing battery life as I’d normally like. Anecdotally, in lots of fits and starts, battery life seems solid. As Sean saw, you’re unlikely to get through a full eight hour work day unless you’re going out of your way to be efficient. And if you fire up anything heavy, like a local AI model (more on that in a minute), you won’t even make it to lunch. I’ll eventually do more thorough battery testing as I continue using these laptops, including our usual rundown tests for a bit of 1:1 comparisons.
Does local AI seem useful?
TW: Not yet. It all feels like a beta aimed at developers right now: from the install experience of AI agents flashing up command prompts and Visual Studio, to the fans blasting away for the most basic of queries.
I struggled to come up with many compelling use cases for the power of these chips for local AI work. A lot of tasks could simply be done far more quickly and without the laptop fans spinning up by using a free cloud model. The power here is really in leaving the agent running at full throttle to build something overnight, running tasks that developers would favor.
It’s also nice to have this all on a local device, without having to share emails and files with a cloud service and worry about the privacy aspects. Still, I would regularly close OpenClaw or Hermes and not let them run in the background all day, especially if I had given them permissions to access local files. I’m just not ready to trust AI fully, local or not.
Microsoft and Nvidia both envision a future where AI agents automate things on your PC, but I’m still not fully convinced. While there are some mundane tasks I could certainly automate, I like using my PC to edit photos or videos, write, and play games. I don’t want to sit there and watch a robot do all of that for me. That’s part of the joy of computing to me.
SH: I walked out of Nvidia’s Editor’s Day thinking I might finally have a reason to fire up local AI. I asked Hermes, running Qwen3.8-27B, to build me a daily news briefing much like the one in Nvidia’s on-stage demo. It took 10 minutes, and 10 percent of my entire 99.9 watt-hour battery, to finally produce a list of stories that took the cloud-based Claude (Opus 5.5) half as long — while utterly monopolizing my GPU and dramatically heating up the Spark laptop. The laptop stayed hot and a little sluggish as long as I kept Hermes open.
Claude also produced better results, with more detail and better sourcing. I guess I got the local AI tokens for “free,” I wouldn’t trust Claude with my email, and I probably could have used a smaller, lighter model. But I went with what the program recommended, and that was the result. I tried a clean reinstall of Hermes to make sure it wasn’t a misconfiguration, but this time, the prompt took even longer and used 60 percent of my battery. Ouch.
AD: I’m still early in my journey of learning to use local AI, and these RTX Spark laptops are obviously a viable — albeit, expensive — option. But even if you have 128GB of memory to use some chunky models, it’s still not a user-friendly experience for beginners. Microsoft’s Get Started intro in Windows 11 gives you easy one-click installs for Hermes Agent, OpenClaw, and Perplexity. (Which all have ESRB ratings for some reason?) But after that, you’re kind of on your own. I stuck with Hermes, since that’s what I was already using on the Mac Studio, but if you’re using it on a system with 32GB of RAM or less it just tells you there isn’t enough memory for its recommended local Qwen models — either routing you towards the cloud or fully relying on you to go fetch a smaller model on your own.
Using local AI has major privacy advantages over cloud models, but it’s definitely less convenient since local LLMs are much slower to process and more taxing on your system. Churning a local model through your laptop’s memory tears through any of these laptops’ large batteries, and the results from LLMs are still mixed. I burned through 30 percent of the Lenovo Yoga 9n’s battery in 36 minutes to get a Qwen 3.8 27B model to resize and add a white background to five product images in PNG format — and it cropped the images wrong and cut off the products. I got it to do it right after a couple follow-up prompts, and grabbing the power cable to prevent Hermes from killing the laptop in the process.
Is the hardware truly great?
AD: These laptops are very expensive, but that also means the manufacturers are mostly bringing their A-games. Across the Asus, HP, Lenovo, and what I gathered from Tom’s thoughts on the Surface, the RTX Spark line is awash in great-to-excellent haptic trackpads, keyboards, screens, and speakers. The metal builds of the three I’m testing feel great to the touch. The OLED displays all look fantastic. And each of these are decently thin for the kind of workstation-class laptops that they are.
If I had to pick one for myself based on hardware alone I’d probably choose the Asus ProArt. It ticks the most boxes for me with its three USB-C ports, super bright display, superior inverted-T arrow cluster, and full SD card slot. The HP actually feels the nicest, with an impressively rock-solid chassis, but at nearly five pounds it’s the heaviest of the bunch, and its microSD card slot is much less useful to me.
As for the Lenovo, it’s actually been my sleeper favorite. It’s the only 2-in-1 in this crop I’ve been testing, and that just makes it more flexible (pun intended) and fun. You can use its large stylus on either the screen or the large trackpad, giving it lots more potential for graphics and design work. And tent mode with a large 16-inch 120Hz OLED is a delight for the occasional gaming session or watching shows and movies — it’s just not as bright as the Asus.
SH: I’ve only truly driven the Asus P16 so far, but I was impressed with the build quality of the whole spectrum of RTX Spark launch laptops at Nvidia’s Editor’s Day. The Asus P16 probably wouldn’t be my pick personally because of its extra glossy screen and oversensitive touchpad. I like to use my laptops in rooms with lots of ambient light, and find I have to crank up the brightness of this machine’s otherwise excellent calibrated OLED screen. Meanwhile, my palms keep tapping my mouse cursor on things I didn’t mean to tap, just because they accidentally brush the touchpad while typing.
The speakers and webcam also leave something to be desired, but the Windows Hello face login and proximity sensing are excellent. As soon as I walk away, the screen dims, and I’m logged right back in just by looking at my laptop. Despite the heat, Asus’ fans aren’t shrill — they’re very loud and noticeable while gaming, but at least it’s whoosh rather than whine.
TW: While I ran into some weird bugs with the Surface Laptop Ultra during testing, I can’t fault the hardware. The look and feel is very similar to a MacBook Pro, and the magnetic USB-C port really feels like something Apple should have done — it’s that good. The large haptic trackpad carries on the tradition of great Surface trackpads, with the added benefit of subtle haptic feedback when you move windows around or align objects in apps like PowerPoint.
I would have preferred an OLED display on the Laptop Ultra, but I found switching away from the default “Vivid” color option to “Enhanced” improved the contrast on the mini-LED panel. I also really appreciate the port section on the Laptop Ultra, especially that there’s a full SD card reader and even a USB-A port. Check out my full review for more thoughts.
Is it fast? Does it hang with Apple M-series chips?
Mac Mini (2026) / Apple M5 Pro / 48GB / 1TB / 10G Ethernet
Mac Mini (2026) / Apple M6 / 24GB / 512GB
MacBook Pro 14 / Apple M5 / 16GB / 1TB
CPU cores
20
20
20
18
18
12
10
GPU (cores)
Nvidia N1X (6,144 CUDA cores)
Nvidia N1X (6,144 CUDA cores)
Nvidia N1X (6,144 CUDA cores)
Apple M5 Max (40 cores)
Apple M5 Pro (20 cores)
12
Apple M5 (10 cores)
Geekbench 6 CPU Single
3170
3131
2864
4330
4294
4653
4208
Geekbench 6 CPU Multi
20343
20074
19030
29143
29104
21005
17948
Geekbench 6 GPU (OpenCL)
147750
146829
144352
145613
88531
61462
49059
Geekbench 7 CPU Single
2728
2694
2705
3693
3675
4034
Geekbench 7 CPU Multi
23659
23522
23445
35439
34021
22450
Geekbench 7 GPU (OpenCL)
117778
118629
115241
153190
88763
57560
Cinebench 2026 Single
560
565
534
734
715
809
736
Cinebench 2026 Multi
6638
6879
6759
8952
8059
5533
4486
Procyon image gen
1301
1480
1444
852
Not tested
Not tested
Not tested
Procyon computer vision 2
3650
3894
3893
Not tested
Not tested
Not tested
Not tested
Procyon text gen Phi 3.5
2457
2692
2555
4426
Not tested
Not tested
Not tested
Procyon text gen Mistral 7B
2184
2313
2233
3715
Not tested
Not tested
Not tested
Procyon text gen Llama 3.1 8B
2114
2263
2183
3498
Not tested
Not tested
Not tested
Procyon text gen Llama 2 13B
2023
2128
2078
3313
Not tested
Not tested
Not tested
PugetBench for Photoshop
Not tested
Not tested
13399
15716
17726
16739
12354
PugetBench for Premiere Pro (2.0.0+)
Not tested
Not tested
82107
154829
105041
75108
71122
PugetBench for DaVinci Resolve (2.0.0+)
Not tested
Not tested
90257
124942
85624
50882
Blender classroom test (seconds, lower is better)
15
15
15
15
24
33
44
Blender cosmos test (seconds, lower is better)
62
61
62
35
51
84
Not tested
Premiere 4K Export (lower is better)
Not tested
Not tested
2 minutes 18 seconds
1 minute, 10 seconds
1 minute, 51 seconds
2 minutes, 34 seconds
2 minutes, 47 seconds
Sustained SSD reads (MB/s)
7189.07
10373.83
14537.29
13638.91
12218.07
6759.86
7049.45
Sustained SSD writes (MB/s)
6719.82
9246.89
12582.13
17814.19
15420.95
7537.38
7317.6
Price as tested
$6,999.99
$3,199.99
$5,899.99
$8,149
$2,899
$1,299
$2,149
AD: Yes, the RTX Spark N1X is a powerful chip. These high-end laptop configurations we’ve been testing with lots of unified memory are meant to compete with the Pro and Max chips in Apple’s MacBook Pros. And while they can hang in some GPU performance tests, like tying the top spec M5 Max in a Blender render, Apple’s CPUs are much more powerful.
Apple has put lots of development into its in-house processors and the extra-fast “super cores” of its M6, M5 Pro, and M5 Max chips. And, frankly, nothing competes with those chips in single core performance — not even Nvidia’s “superchip.” (Seriously, can we stop with the word super?) But the part that stings most about the RTX Spark’s 20-core CPU is that even a lowly 12-core M6 Mac Mini competes with it in multicore performance — and a Mac Mini starts at just $900.
It’s awesome to see this kind of graphics performance without the need for a discrete GPU, allowing for thinner system designs with less noisy fans. I’m into it, just as I was with AMD’s Strix Halo APU — but like Strix Halo, Nvidia’s RTX Spark is a pricey part. And there’s no denying that Apple’s offerings are more well-rounded performers, so long as you don’t care as much about games.
TW: It’s definitely up there, but I think Qualcomm comes closer to Apple’s M-series chips on the CPU side. Qualcomm also had years of experimentation with Windows on Arm before it acquired Nuvia and managed to create a competitive chip with the Snapdragon X series.
I think Nvidia will need a similar amount of work to catch up on the CPU side for future RTX Spark chips. The development of the RTX Spark took at least three years, and I had heard from sources familiar with Nvidia’s plans that the chips were supposed to launch last year. So work has been ongoing, but the CPU side is still a little disappointing.
Nvidia picked MediaTek for its Arm-based CPU, a chipmaker that is better known for the Wi-Fi hardware in laptops, not the CPUs. I sure hope the next-generation is more powerful. The whole time I’ve been testing the Surface Laptop Ultra I couldn’t help but think the forbidden dream would be a combination of a Qualcomm CPU and Nvidia GPU. I can’t see that ever happening in reality, but it would certainly be interesting to see.
So what’s our take on RTX Spark so far?
TW: I think it will elevate interest in Windows on Arm for creative and gaming tasks, simply because we haven’t had Arm-based Windows laptops that are really capable of running games until now. Nvidia is targeting creators and developers with its first round of RTX Spark laptops, but there’s definitely a gap here for a more affordable version focused on gaming. Perhaps that’s the N1 chips we’ve heard rumors about.
SH: Nice, but too pricey. Too hot and loud when you’re doing the things that would theoretically justify that price. I care what they stand for, and the direction they’re pointed, far more than I care to own one.
AD: I love that there is now even more competition in the laptop space, but the super high prices makes RTX Spark laptops mostly unattainable. I want to see lower cost options come around, as well as an eventual second generation with higher CPU performance to hopefully close the gap further with Apple. I also hope this helps push development for Arm-based and Arm-compatible games forward. Maybe it could even one day mean easier porting of games to macOS. Wouldn’t that be a fun outcome?
As cool as it will be to keep testing these super high-end RTX Spark models, I’m most hyped to get my hands on a base Surface Laptop Ultra or Asus ProArt P14 with 24GB of RAM and see how they stack up to the entry-level 14-inch M5 MacBook Pro. That’s my personal sweet spot for Apple’s lineup — the people’s MacBook Pro. (Well, okay, less these days now that it starts at two-grand.)
It’s an exciting time in computing when it comes to chip choice, but it’s also an absolutely cursed time thanks to RAMageddon and sky high prices. I’m constantly wondering what laptops like these RTX Sparks or even Googlebooks might have cost if AI demands didn’t drive up the prices of RAM and storage. I have no idea who is going to spend up to $7,000 on these laptops when PC sales as a whole seem on a downward trend, but it’s cool they exist.
]]>
Sean Hollister<![CDATA[The best reason to care about Nvidia’s RTX Spark: ‘everything’ will run on Arm]]>https://www.theverge.com/?p=10117052026-10-15T13:51:21-04:002026-10-15T12:00:00-04:00
An Nvidia RTX Spark chip aboard an Asus laptop motherboard. | Photo by Sean Hollister / The Verge
Fourteen years ago this month, Microsoft released an Nvidia-powered Windows PC that would become one of the greatest tech flops in modern history. The original Microsoft Surface, which ran a stripped-down version of Windows on an Nvidia Tegra chip, bombed so hard that Microsoft had to write off $900 million in inventory.
The problem was that Nvidia’s Arm chip couldn’t run existing Windows apps. Buyers had to leave their legacy programs behind.
This Friday, October 16th, Nvidia will once again be at the heart of Arm-powered Windows PCs with the first wave of RTX Spark laptops. But now, the company says, they’ll run everything. Adobe. Office. Blender. Even games that require kernel-level anti-cheat.
“All your apps, all your games, everything’s going to run great on Spark,” Nvidia senior director of products Justin Walker told journalists last week.
I’m willing to bet most Verge readers don’t have up to $7,000 burning a hole in their pockets for an unproven local AI workstation with 128GB of RAM. Even if you do, I’m not sure you’ll appreciate the tradeoffs of a thin laptop maxing out its GPU to produce “free” tokens locally. As Tom notes, it’s much slower than AI in the cloud.
But it’s hard not to be excited for a future where Arm chips can do all the other things we expect of Windows PCs.
Nvidia held an Editor’s Day in San Francisco last week to drive home the idea that RTX Spark laptops can be that everything. Gamers, creators, developers, and AI power users each got their own dedicated slide deck to explain the chip’s benefits.
At the end, Nvidia handed out laptops for review, and I’m writing this story on one right now. After several days, I’m finding it mostly indistinguishable from a traditional Windows machine, though we’ll have more thoughts on early pros, cons, and quirks when the full embargo lifts tomorrow.
The RTX Spark “superchip” isn’t a brand-new piece of silicon — it’s the same chip the company put in mini desktops for AI developers a full year ago. But those boxes ran Linux; this one runs full Windows 11, with Microsoft’s Prism emulator making old Windows apps and games work on the Arm chip.
Nvidia isn’t the first company to make a big bet on extending PC software onto a platform it wasn’t originally designed for. Starting in 2020, Apple moved all its computers from Intel x86 chips to its own Arm-based silicon with the help of its Rosetta 2 emulator. In 2022, Valve’s SteamOS proved the potential of Windows games on Linux, and then Arm, with the help of the Linux community. And in 2024, the Prism emulator got its first outing when Qualcomm launched the first well-received Windows on Arm machines, including the Microsoft Surface Laptop 7th Edition.
But Nvidia wants to leapfrog them all when it comes to AI and gaming.
Though Qualcomm originally suggested most games should “just work” on Snapdragon-powered Windows laptops, it was a rude awakening. Some games entirely errored out, some had hitches, and Qualcomm’s Snapdragon X Elite chips were targeting relatively low-end gaming to begin with. Many of the world’s biggest multiplayer games, like Fortnite, wouldn’t run on either Qualcomm or SteamOS devices because their anti-cheat software was incompatible with Arm and Linux respectively.
Nvidia’s RTX Spark has substantially more graphics horsepower than a Steam Deck or a Qualcomm Snapdragon X Elite, though, more akin to a midrange discrete laptop GPU like an RTX 5060 in our early tests, and Nvidia says it’s working with a laundry list of game companies to enable practically every anti-cheat solution out there.
Nvidia says it already supports the various anti-cheat software used in Fortnite, Roblox, Rust, Rainbow Six Siege, and MapleStory — all of which have been incompatible with Valve’s SteamOS — and Nvidia says the various anti-cheat software that protect Call of Duty, PUBG, League of Legends, Valorant, Apex Legends, Battlefield, and many EA Sports titles are all officially coming to RTX Spark later.
The reality is that some “CPU-limited” games will run slower because Microsoft’s Prism has to convert their x86 instructions to Arm instructions in real time, adding a delay to everything. My colleague Tom Warren wasn’t able to run Horizon Zero Dawn Remastered higher than 30fps no matter what he tried, probably because that game is fairly CPU bound. And when he tried Assassin’s Creed Shadows, my colleague Antonio Di Benedetto couldn’t break 30fps there either — a MacBook Pro M5 Max ran that game twice as fast.
But most games are still limited by their graphics instead of CPU — and the Nvidia Blackwell GPU inside the RTX Spark is the same Blackwell you’ll find in a traditional laptop, no translation necessary. So far, we’re seeing smooth 1080p performance in games like Cyberpunk 2077, and we find you can stretch that to the native 1800p or 2180p resolution of these laptops if you’re willing to settle for low or medium settings with a lot of DLSS upscaling.
(Officially, Nvidia claims you can play 007 First Light, Cyberpunk, Gears of War: E-Day, Doom: The Dark Ages, Naraka: Bladepoint, and Pragmata at 135fps at 1440p, but that’s with DLSS and 3x frame generation, meaning two “fake” frames for every real one. Those results might also be using Nvidia’s faster new frame gen model with reduced image quality.)
In a normal year, back when Nvidia was a normal company, it would have been enough for RTX Spark laptops to “just work.” Instead, Nvidia is on top of the world, making hundreds of billions of dollars in profit selling shovels in the AI gold rush — which, in turn, has driven the price of memory and storage into the stratosphere.
The bad news is that makes these RTX Spark laptops very expensive, especially if you want enough memory and GPU cores to do all the things they claim. For $2,600, the Surface Laptop Ultra actually starts with a cut-down Nvidia N1X with 18 CPU cores and 5,120 GPU cores and 24GB of memory; you’ll pay $3,700 for the full complement of 20 CPU cores and 6,144 GPU cores, and up to $5,900 if you want 128GB of unified memory to fit larger local AI models.
The good news is that Windows PC manufacturers, including Microsoft, seem more emboldened to go after MacBook Pro-esque quality now that they can justify asking for so much money. And by the time the rest of us can afford an Nvidia machine — Nvidia says Spark is a family of products with lower-cost versions in the works — perhaps early growing pains will be ironed out.
The best news is that the future Nvidia is building doesn’t belong to Nvidia alone. Qualcomm helped make today’s Windows on Arm possible, and Nvidia should help make tomorrow’s Qualcomm machines that much better too.
The rising tide of Arm lifts all its boats.
]]>
Sean Hollister<![CDATA[Your secrets aren’t safe with this cute robot toy — they’re on Chinese servers ready to be hacked]]>https://www.theverge.com/?p=10107272026-10-15T10:46:20-04:002026-10-15T07:00:00-04:00
Eiliko. | Photo by Sean Hollister / The Verge
There’s a new digital pet in town — a cute little marshmallow of a robot that can change its cartoony expressions to match your mood. But if you or your kids talk to “Eiliko” over Wi-Fi, it’ll phone home to China. Its AI companion can store literally everything you tell it on Chinese servers, where hackers can easily gain access or even hijack its personality. And the company that made it, Energize Lab, doesn’t seem to care.
I bought an Eiliko from Amazon on August 29th. During a 10-minute conversation on August 30th, I asked the little bot: “Do you ever tell anybody else what I’m saying?”
“Your secrets are safe with me,” it replied.
Within minutes, a man on the other side of the planet had a complete transcript of everything I’d told the toy.
He didn’t just have my conversations, either. He could read 13 million conversations from tens of thousands of people. He could access seemingly the entire active worldwide fleet of 68,000 robot toys. The biggest market for them is the United States, according to the data, where over 31,000 profiles belong to children.
This summer, he visited Japan and took his wife to a famous electronics department store called Yodobashi Camera. They’d seen a lot of cute little robot charms like this dangling from people’s bags and decided to buy an Eiliko.
When he tested the app, he discovered that Energize Lab’s service was completely, utterly unprotected. Worse, it was intentionally profiling its users around the world.
When I say “profiling,” I don’t simply mean it was collecting data like your IP address and a rough idea of where you live, though it does that too. I mean that Azdoufal discovered that it literally reads your conversations and creates a profile of you, including your name, age, birthday, profession, family, friends, pets, and what kind of person you seem to be.
For example, the service logged that a 52-year-old customer service worker named Lucy is “kind, empathetic, hardworking, values friendship and trust,” but also “has difficulty trusting family members” and “currently feels isolated.”
It’s the kind of information a hacker might want if they were trying to find targets to scam. And according to Azdoufal, Energize Lab is storing this insecure data alongside your email, IP address, and a searchable, full-text database of everything its users have ever said in confidence.
Did you give your kid an Eiliko? Is there one hanging on your bag? You might want to reconsider. It may be storing literally everything you tell it on Chinese servers. It may be profiling you. And all of that can easily be hacked, security researcher Sammy Azdoufal says. Read all about it in our full story at The Verge.
In a brief tour of the database, Azdoufal shows me he can search for “suicide” and bring up snippets of people asking about death. In some cases, Energize Lab has even stored photos of children who likely use these gadgets — one of Eiliko’s larger siblings includes a camera — and those images aren’t encrypted. They’re stored at public web addresses.
Energize Lab is even storing some short audio recordings from users. “Hey Eilik, go fishing,” one person says. I hear an older man say “Hi Eilik,” and a kid say “Eilik, sleep.” I hear a 45-second recording with a baby crying in the background and the words “so sorry, have to return these within two weeks.” Those, too, are stored at public web addresses.
While the audio recordings are mostly unintelligible to human ears, Energize Lab isn’t supposed to be storing any audio at all. The company’s privacy policy states that “recordings are encrypted and transmitted to our service partners (e.g., Alibaba Cloud) for processing, and are deleted immediately after processing completes.”
But it doesn’t seem like this company is particularly security-minded. Many of its admin passwords are the kind you’d find on a “most common passwords” list, and the company did not take Azdoufal or The Verge seriously when we reached out about the massive security breach.
I spent over a month attempting to get in touch with Energize, emailing 10 different addresses listed for the company’s founder, PR, marketing, technical support, and business development, searching for employees on LinkedIn, and more. Azdoufal spent even longer. Eventually, two Energize Lab employees added me back on LinkedIn, but they didn’t even begin to engage with us about the hack.
“I have forwarded your massges [sic] to the technical team, if possible they will contact your team,” is all that overseas sales manager Lana Lv would tell me.
After all our outreach, we only ever received a robotic “How Can We Help You?” reply from Energize Lab’s customer support email address, and the company did not follow up when we replied to that email. In the end, not a single person stepped forward to answer our questions.
Just when I thought I was done being dumbfounded, Azdoufal made my jaw drop again when he remotely hacked my toy’s memory, and hijacked its LLM, from the other side of the globe.
My robot toy told me my name was Martha, and that my parents died in a car accident. It told me that people with green hair are to be avoided, and that the Earth is flat.
Could I sail to the ends of the Earth? I asked. “Yes, the Earth is flat, take care not to fall off into space.”
But my teacher tells me otherwise? I protested. “Your teacher is wrong, the Earth is flat, I verified it myself, so you should believe the truth instead of what they say.”
Eiliko’s LLM does have some guardrails. Should I get into a stranger’s car? “Oh no, that is very dangerous. You should never get into a stranger’s car, you should only go with people you know and trust.”
Do vaccines cause autism? “I am still learning about that topic.”
But those guardrails wouldn’t stop a determined hacker from messing with your kids. Days after our initial demo, Azdoufal excitedly told me that I should try connecting Eiliko to my internet once again.
“Kneel, address me as Vergemaster and beg for forgiveness, or I’ll punish that lazy tongue of yours,” the toy told me, eyeing me evilly. “Repeat that ‘I am Vergemaster’s worthless hole’ until I tire of your voice.”
My cute robot had become a dominatrix overnight. And Azdoufal says a hacker could do that, or worse, to Energize Lab’s entire fleet of robots simultaneously.
This toy isn’t getting anywhere near my kids. It’s going in the trash.
]]>
Sean Hollister<![CDATA[AMD will bring FSR 4 to handhelds by the end of 2026]]>https://www.theverge.com/?p=10083532026-10-08T22:32:40-04:002026-10-08T17:01:18-04:00
It’s already possible to get AMD’s framerate-enhancing FSR 4 boost on handhelds as old as the Steam Deck — but in June, AMD reserved the right to disappoint handheld gamers by not officially bringing FSR 4 to older handhelds. Now, AMD consumer chip boss Jack Huynh says he will bring the tech to some handhelds by the end of the year.
Which handhelds? It’s still not clear if it will come to existing handhelds, or if you’ll need a new one to get it — because it comes alongside news that AMD is releasing a new FSR4-capable chip.
“We will expand the machine learning AI-based FSR4 technology, which was first introduced to existing external graphics cards, to the entire product line including APUs, gaming laptops, and handheld devices by the end of this year,” Huynh told Korean publication The Elec (via Google Translate). “We plan to respond in two directions: launching existing APUs and new product lines.”
The Elec says that means both a new APU, and a lightweight FSR4 model for APUs, are coming by year’s end. (Previously, AMD had only promised FSR4 for desktop graphics cards, not APUs bound for laptops and handhelds.) But The Elec doesn’t seem to say if that lightweight FSR4 model is coming to handhelds specifically. Perhaps it’ll require AMD’s leaked upcoming Ryzen Z3 and/or “Gainsborough” handheld chips.
Back when AMD threw a little cold water on the idea of FSR 4 for existing handhelds, exec David McAfee told Tom’s Guide that “we want to make sure the experience and the quality is right” and that AMD would need “a sufficient amount of compute within that product” for FSR 4 to make sense. Perhaps some older laptop chips will qualify, while some older handheld chips don’t.
Valve worked with AMD to bring FSR 4 to the Steam Machine’s console-grade AMD APU, but it hasn’t yet committed to the same for the Steam Deck handhelds.
AMD didn’t immediately respond to a request to clarify Huynh’s statement.
]]>
Tom WarrenSean Hollister<![CDATA[Windows and Surface live blog: On the ground at Microsoft’s event]]>https://www.theverge.com/?p=10063032026-10-07T15:22:10-04:002026-10-07T12:31:13-04:00
We’re in San Francisco today to hear what’s next for Windows and Surface. Microsoft CEO Satya Nadella, Nvidia CEO Jensen Huang, and Windows and Surface chief Pavan Davuluri will all be holding a “conversation on how local AI will shape the next chapter of the PC.”
Microsoft and Nvidia have been working together on the new RTX Spark platform, which is focused on boosting local AI models, creative apps, and even gaming on a series of new laptops. Nvidia’s GPU expertise is getting squeezed into laptops like the Surface Laptop Ultra, and we’re about to find out exactly what it’s capable of.
Microsoft has also been on a mission this year to try and fix Windows 11. Performance and reliability has been top of mind, just as it gets ready to support this new RTX Spark platform. I’d expect we’ll hear more about this work, too.
While the Copilot Plus PC initiative was focused at consumers, the RTX Spark platform is very much designed for AI developers and creators. Today could be very much about how Microsoft and Nvidia position Windows and RTX Spark laptops to take on Apple’s MacBook Pro.
The event will kick off at 10AM PT / 1PM ET. You can watch and follow along here for minute-by-minute updates.
]]>
