Where are all the robot trucks?

Where are all the robot trucks?

Illustration by Sisi Kim for The Verge

The promised wave of autonomous big rigs never materialized. But 2024 could prove to be a pivotal year for the technology.

When legislation requiring heavy trucks to have a human operator onboard sailed through the California legislature earlier this year, the coalition of labor organizations backing it knew their victory wasn’t sealed quite yet.

They were right. Weeks after the bill’s passage, Governor Gavin Newsom vetoed it, arguing it would stifle innovation in the state. Teamsters and their elected allies vowed to not back down in their campaign against fully autonomous trucking.

The battle over the legislation, AB 316, may presage a broader fight over self-driving trucks ahead of what is set to be a critical year for the technology as it begins to be rolled out in earnest.

Despite claims of reducing accidents and assurances that autonomous vehicles will not displace truckers, the prospect of sharing the road with robotically controlled 18-wheelers continues to scare the public and alarm labor groups.

Companies in the autonomous trucking space have been piloting the technology for some time, moving long-haul freight for a wide range of customers across the Southwest, with the most traveled routes being between major metropolitan areas in Texas.

So far, these vehicles have been piloted by licensed operators ready to step in if needed. But major companies say they are now ready to remove that human presence and achieve what is known as Level 4 automation.

Image: Andrej Sokolow / Picture Alliance via Getty Images

Where are the autonomous big rigs?

Plans to deploy driverless trucks come at a fraught time for the technology and autonomous vehicles in general. Multiple firms focused on self-driving trucks have recently pulled back operations or folded entirely, and a high-profile robotaxi accident in San Francisco is sending shockwaves across the industry.

Aurora Innovation, founded in 2017 by alumni of Uber, Tesla, and Waymo, plans to deploy 20 fully autonomous trucks next year, with an eye on expanding to about 100 trucks in 2025 and eventually selling to other companies.

Kodiak Robotics, which boasts partnerships with Maersk, CR England, and Ikea, is also planning to launch driverless trucks in 2024.

“The technology is finally at a point where driverless is here, and it’s been a long time coming,” Kodiak’s co-founder and CEO, Don Burnette, who has been working in the self-driving vehicle space for 15 years, told The Verge. “We’ve really solved all of the fundamental technology hurdles that we need to; now it’s just about proving the safety.”

Other companies have longer timeframes for their launches.

“The technology is finally at a point where driverless is here, and it’s been a long time coming”

Torc Robotics, for example, has set a target of 2027 for the production of the driverless trucks it has been developing in a partnership with the manufacturing giant Daimler Truck.

While these three companies prepare for their launches, other once-major players in autonomous trucking have recently cut back on or ended their efforts to bring products to market.

The Alphabet subsidiary Waymo announced in July that it would be prioritizing vehicles for ridehailing. TuSimple laid off half its American workforce this summer and is said to be considering fully exiting the market to focus on operations in China amid management turmoil. Embark Technology laid off nearly a quarter of its staff in March before merging with another autonomous vehicle firm.

These setbacks are reflective of an industry that has not matured as quickly as some experts anticipated. Up until recently, many in the industry predicted that autonomous trucks would be on the road before self-driving vehicles were let loose in cities because highways are much less complex operating environments.

However, while robotaxis are moving customers in three major American cities and being tested in over a dozen more, driverless trucks remain stuck in neutral.

Image: Waymo

The safety case and faster freight

The first thing that companies and proponents bring up when making the case for autonomous trucking is safety.

In 2022, 5,887 people died in accidents involving large trucks, according to preliminary National Highway Traffic Safety Administration estimates. Supporters of autonomous systems frequently argue that removing the potential for human error from the equation would necessarily reduce accidents.

“The vast majority of human driven accidents are caused by drunkenness, drowsiness, or distractedness,” Sterling Anderson, Aurora’s chief product officer, told The Verge, echoing a common refrain among AV boosters. “Autonomous trucks have none of those things.”

Until driverless trucks are deployed at scale, though, this argument remains purely hypothetical.

“People are the cause of most accidents because, you know, grizzly bears can’t drive a vehicle,” Mike Di Bene, a Teamsters member and veteran commercial truck driver, jokingly pointed out.

“People are the cause of most accidents because, you know, grizzly bears can’t drive a vehicle.”

Proponents of the technology say that beyond removing human error, autonomous truck systems are safer because of their sensor systems and programmed defensive driving.

The sample size of autonomous trips is admittedly small, but the trucks have performed well when it comes to safety, according to publicly available crash data. In the few reported incidents involving self-driving trucks in Texas, where most companies have focused testing, other vehicles have been culpable.

Aurora recently simulated the 32 fatal collisions that occurred between 2018 and 2022 involving a tractor trailer on the Dallas to Houston route it will be launching on and claims that none of them would have occurred if the company’s system had been driving.

“Safety is the primary metric by which we measure the progress of our product,” Anderson said.

The other main argument in favor of self-driving trucks is that they are capable of moving freight much faster over long distances. Under guidelines set by the Federal Motor Carrier Safety Administration, truck operators are allowed to drive a maximum of 11 hours a day and have to take a 30-minute rest after eight consecutive hours behind the wheel. Autonomous trucks would face no such restrictions.

“With extended fuel tanks, our trucks can drive without stopping all the way from coast to coast,” Burnette said, estimating that a Kodiak truck could make the drive from Atlanta to Los Angeles in a little less than two days. “That’s a game-changer from an industry perspective.”

Self-driving skepticism

Despite the emphasis placed on safety by autonomous trucking companies, the public and drivers remain far from convinced.

Polling conducted as AB 316 moved through California’s legislature this year found that close to 80 percent of likely voters in the state would be uncomfortable with heavy driverless trucks on roads and freeways.

The general public’s reluctance toward driverless vehicles more broadly is likely to influence their feelings about trucks, no matter how much companies emphasize that driving on highways is easier than on dense city streets.

The public and drivers remain far from convinced

Robotaxis in cities like San Francisco, Phoenix, and Austin do not seem to have assuaged the public’s concerns. In fact, an AAA poll from earlier this year found that 68 percent of drivers are afraid of fully autonomous vehicles, up from 55 percent the year prior.

Many truck drivers are unconvinced that driverless systems are safer than human operators.

“While we think there are places where autonomous vehicles could operate hopefully without doing too much damage, our roads and streets, our highways of America aren’t that place right now,” said Todd Spencer, a veteran driver and the president of the Owner-Operator Independent Drivers Association (OOIDA), an organization representing 150,000 truckers across the US and Canada. “Our members are more concerned than most other drivers about these vehicles having the potential to kill them.”

Several drivers brought up the situation in San Francisco, where an autonomous Cruise vehicle seriously injured a pedestrian by dragging her under a car for 20 feet, as evidence of their safety concern.

“Even in the middle of nowhere, things happen in a split second,” said Jared Hamil, a Teamsters member currently working at UPS with experience in commercial trucking. “Whether it be an animal or a car or something, we have to be able to adjust to that at a moment’s notice.”

Labor organizations like the Teamsters, which represents tens of thousands of drivers, and the OOIDA have also expressed concerns that autonomous trucking will endanger the careers of their members.

“Our folks know they do an important job in our society, and they would prefer not to see their job eliminated via technology,” Spencer said.

Image: Bo Lee / Xinhua via Getty Images

The jobs debate

Proponents of self-driving trucks argue that concerns about worker displacement are overblown because adoption will be slow, there is a shortage of drivers now, and ultimately, the technology will create jobs.

While it is true that the 20 vehicles that Aurora puts on the road next year are unlikely to put any truckers out of work, unions are interested in protecting careers in the long term, not just the careers of current members. A 2018 study from the UC Berkeley Labor Center found that roughly 294,000 long-distance drivers could be displaced by autonomous technology.

Arguments by the industry about a deficit of operators almost always cite the American Trucking Associations, which claims there was a shortage of 78,000 truck drivers in 2022. Digging into publicly available numbers (PDF), however, shows the opposite. In California, for example, there were over 600,000 with Class A or B licenses in 2021 for only 140,000 “truck transportation” jobs. The problem, groups like OOIDA say, is actually in retention.

“Not a single one of those companies has put forward a proposal to demonstrate how exactly they’re going to create new jobs”

“There’s well over 400,000 CDLs issued every year,” said the group’s executive vice president, Lewie Pugh, referring to commercial driver licenses. “They just don’t stay. Within six months to a year they’re gone because they don’t make very much money. They’re away from home.”

A recent Department of Transportation-funded study found that between 26,400 and 35,100 jobs would be created by automating long-haul trucking with minimal layoffs.

The veracity of the claims on labor displacement aside, more has to be done by autonomous trucking companies to address worker concerns.

“Cruise, Waymo, Kodiak, Aurora, not a single one of those companies has put forward a proposal to demonstrate how exactly they’re going to create new jobs,” Teamsters spokesperson Matt McQuaid said. “We haven’t heard from the companies yet on that issue, which is paramount to us.”

That’s not to say companies have ignored workers. Some meaningful steps toward assuaging labor concerns have been taken. Aurora, for example, has worked with Pittsburgh Technical College and Gallatin College in Bozeman, Montana, to train new operators with skills specific to autonomous trucking.

“Honest conversations”

But many labor groups feel like they’ve been left out in the cold.

“Nobody ever says what they’re really trying to do,” Norita Taylor, a spokesperson for OOIDA, said. “They invite the media for their demonstrations, and they have a lot of bells and whistles and woohoo-ing, but they never really say what they’re trying to ultimately accomplish.”

The drivers who spoke with The Verge for this story all said they have never heard directly from autonomous trucking companies.

When asked about outreach to unions or individual drivers, a spokesperson for the Autonomous Vehicle Industry Association (AVIA) said it “continues to engage with labor organizations throughout the public process, including testifying alongside them in various state capitals and participating together in the California DMV trucking workshops.”

Twenty-three states have already authorized testing or deployment of driverless vehicles, according to the AVIA, and outside of California, efforts to regulate autonomous trucks have largely stalled so far.

However, with the increased attention that has already been brought to driverless vehicles by high-profile crashes and commitments by groups like the Teamsters to keep fighting autonomous trucks, new efforts to regulate the technology are bound to emerge. The lead sponsor of AB 316 in the California State Assembly, Speaker Pro Tempore Cecilia Aguiar-Curry, told The Verge that she plans to reintroduce similar legislation next session.

“I remain terribly concerned about the impact of lurching forward with unproven vehicle technology on California workers and public safety,” she said in a statement. “The industry admitted in public testimony they’ve not had conversations with their employees about what they might do to mitigate job losses. Rhetorical pats on the head about ‘more jobs’ with no detail or concrete plans don’t reassure workers and won’t feed their families.”

The model of requiring a human operator proposed in the labor-backed AB 316 may be a popular one given that it directly addresses fears about losing jobs and unmanned vehicles being unsafe.

For the industry, the proposal is a nonstarter because it wipes out the economic benefits of traveling faster and saving on labor costs. “The value proposition is simply not there if you’ve got a driver in the truck,” Aurora’s Anderson said.

“The value proposition is simply not there if you’ve got a driver in the truck.”

This position has frustrated drivers like Di Bene, who said it shows the companies are more focused on the bottom line than safety. “It just bothers me that these companies, in essence, agree with AB 316 until they want to sell the product,” he said.

Some studies do suggest that humans are not very good at intervening in moments of crisis if they’ve largely been disengaged from driving as they would likely be in these trucks.

An alternative to the compromise of AB 316 could be a system where autonomous trucks handle long routes between hubs while human drivers remain in charge for movement within cities and deliveries. But a proposal along those lines would likely have to come from companies and require significant outreach to convince workers.

Torc’s chief strategy officer, Andrew Culhane, told The Verge that engaging in “really honest conversations” with operators will be essential to build trust and make the adoption of self-driving trucks successful.

“A fear is valid, whether we think it’s justified or not,” he said. “If they have fears or reservations about this product, we need to have a conversation about it and understand what’s driving that and what can we do to move them forward.”

The launch of fully driverless trucks on America’s roads next year could be definitive for the technology. To see how an issue early on in the adoption of this kind of system can set back the whole industry, one has to look no further than Cruise — which, since the accident in San Francisco, has recalled 950 taxis and announced layoffs.

Companies face a high bar to overcome labor opposition and convince drivers that autonomous trucks will make American highways safer.

“Until AI can love and fear,” Di Bene remarked, “I don’t need it behind the wheel because I don’t want to die.”

WhatsApp to let iOS users share pictures and videos in original quality

WhatsApp to let iOS users share pictures and videos in original quality

Media files sent as documents don’t display a preview in WhatsApp, but you can view them in-app by tapping on them. | Illustration: The Verge

WhatsApp is rolling out a new feature on iOS that allows users to share photos and videos as a document to preserve their original quality. As spotted by WABetaInfo the official changelog provided alongside WhatsApp’s 23.24.73 update on iOS notes that users can now “easily send original quality media as a file,” as an alternative to the compressed approaches the communications app defaults to.

This feature was spotted being tested by a small subset of users in November, but is now releasing to all iOS users. The changelog notes that this is rolling out “over the coming weeks” so it may take a while to appear on your device, but when it does you can access it by tapping the “+” in chat, selecting “Document” and then “Choose Photo or Video.” Once sent, recipients can tap into the file on WhatsApp to view photos and videos, but previews don’t seem to appear in the main chat interface. There’s a 2GB limit on the files that can be sent using the feature.

Screenshot by Jon Porter / The Verge

Images and videos sent as documents don’t show previews in the main chat interface.

WhatsApp has also been testing the same feature for Android users since at least September this year, which suggests a stable release for Android devices may not be too far behind.

Previous workarounds have existed to send full-resolution photos and videos via WhatsApp that involved manually changing its file extension to PDF or DOC. Removing that extra work streamlines things considerably for users, providing an easier way to share high-quality media alongside the HD Pictures update that WhatsApp released this summer. We have reached out to WhatsApp to clarify when Android users can expect the same update and will update this article if we hear back.

Spotify to lay off 17 percent of its workforce in latest round of job cuts

Spotify to lay off 17 percent of its workforce in latest round of job cuts

Image: Nick Barclay / The Verge

Spotify is laying off 17 percent of its employees in an attempt to cut costs, its CEO Daniel Ek announced to staff today. Based on its total headcount of 9,241 revealed during its last earnings release, the cuts are expected to impact over 1,500 people.

In a memo sent to staff, Ek said slowing economic growth and rising costs were to blame for the cuts, which he said would make Spotify a leaner company. “Today, we still have too many people dedicated to supporting work and even doing work around the work rather than contributing to opportunities with real impact,” Ek wrote. “As we’ve grown, we’ve moved too far away from this core principle of resourcefulness,” he later added.

“As we’ve grown, we’ve moved too far away from this core principle of resourcefulness”

This is Spotify’s third major round of layoffs conducted this year. In January the company announced it would be laying off 6 percent of staff, or roughly 600 employees. Then, in June, it announced it would be cutting a further 200 roles from its podcast division. As well as cutting costs, Spotify has also made moves to increase revenue, raising prices of several of its plans across multiple markets, including the US, over the summer.

These layoffs have come after Spotify’s headcount increased significantly during the pandemic, with its headcount nearly doubling in the past three years, The Wall Street Journal notes. In his memo, Ek defended his decision to grow the team throughout that period, but said that “we now find ourselves in a very different environment.”

Employees impacted by Spotify’s latest layoffs will receive around five months of severance pay according to Ek’s memo, during which time the company will continue to cover their healthcare.

Spotify has generally prioritized growth over quarterly profits throughout its history, but the WSJ notes that investors have been increasingly pushing for profitability over the past year. Ek said at an investor day last year that he intends for Spotify to be profitable by 2024. Although the company posted a quarterly profit in its last earnings release, the WSJ notes that it reported losses of €462 million (around $502 million) in the first nine months of this year.

Kiss debuts ‘immortal’ digital avatars and plans to go ‘fully virtual’

Kiss debuts ‘immortal’ digital avatars and plans to go ‘fully virtual’

Image: Pophouse Entertainment

Kiss has been on a really, really long tour. The biggest glam rock band of all time has been playing its End of the Road tour for the last four years, and just wrapped the final show of its final farewell tour in New York City at Madison Square Garden on Friday. Kiss has done several farewell tours, but this time might be for good.

At the end of the show, after Kiss finished playing its last song (“Rock and Roll All Nite”), they disappeared in a hail of fire and smoke. After the smoke faded away, the lights went down, and on the screen behind the stage, a camera zoomed across a spooky lake on some fantastic planet toward the silhouettes of four figures: the new digital avatars of Kiss, in what I suppose must be their final form. Then Paul Stanley shouts:

“Kiss army, your love — your power — has made us immortal. The new Kiss era starts now. Oh yeah!”

Here’s one video of it.

Another one shows a much cooler aspect of this portion of the show: translucent screens with shots of the band projected onto them, lending the show a decidedly futuristic Blade Runner feel.

The avatars “performed” a song, then the video ended and left the audience with a picture of the four avatars under the stylized KISS logo, superimposed with the phrase “A NEW ERA BEGINS.”

The “new era,” of course, is one of making money from the avatars. The company behind the show, Pophouse Entertainment, has already been doing so with young, digital versions of ABBA in its ABBA Voyage show for over a year. Kiss, a band that has doggedly merchandised its image for half a century, seems like a good fit for such a partnership.

Pophouse said in a press release today that it will put on “immersive, avatar-powered” concerts using Kiss’ Industrial Light & Magic-created avatars.

Watching big acts like Kiss use digital recreations of themselves isn’t surprising — after all, these are people who have made a great deal of money off of an image they’ve created, so why not keep that money faucet open?

Do people really want to go see a “live” show with no live performers? Yeah, I think they absolutely do. Bloomberg reported that those ABBA avatar shows have been pulling in $2 million every week. And Variety wrote last week that the Eras Tour concert video had passed $250 million in worldwide box office sales. Kiss and its avatars could do just fine — for more about the band’s future, try out this 22-minute conversation about their transition to avatars.

Fortnite’s next chapter adds boss battles, Lego, Solid Snake, and a brand-new island

Fortnite’s next chapter adds boss battles, Lego, Solid Snake, and a brand-new island

Image: Epic Games

Fortnite is heading underground. Following The Big Bang event on Saturday, which led to the destruction of the Chapter 4 island, some big changes are coming to the battle royale in a new season dubbed “Underground.” That includes a new island, gameplay features like boss battles, Lego skins, and of course more licensed characters like Metal Gear’s Solid Snake and Peter Griffin from Family Guy.

Let’s start with the island. In the update — which brings the game to Chapter 5: Season 1 — the battle royale island has been completely changed, which means a number of new locations. There’s a railway — complete with a train you can actually travel on — as well as marinas, mansions, and an all-new grassland area. There’s also quite a bit of snow.

As for gameplay changes, the island is now littered with five boss characters — one of whom is Peter Griffin. Defeating them will earn players a medallion with a very useful feature: the ability to regenerate shields over time. But it comes with a drawback as well, as holding onto a medallion will alert other players to your location on the map. “The more medallions you have, the more precise the radius of your exact location,” Epic says.

Other additions include: an SUV to drive around in, a weapon called the Ballistic Shield that let’s you “defend yourself with the shield while shooting the pistol at the same time,” the option to move while healing or restoring shields, and the ability to mod weapons in certain locations around the island. There’s also a new kind of in-game cosmetic; now, in addition to customizing your character and weapons, you can customize vehicles. You can choose different bodies, decals, and wheels, and once you hop in a sports car in game, the vehicle will shift to your choices. (Certain items from Rocket League will carry over for those who play both games.)

For those who snag this season’s battle pass, they’ll eventually be able to unlock all of the new boss characters, alongside a new version of Jones and, later on in the season, Solid Snake.

And then there’s the Lego. On Saturday, Epic announced an upcoming in-game experience called Lego Fortnite, part of an ongoing partnership that’s described as a “survival crafting” game that will launch inside of Fortnite on December 7th. But the collaboration also extends to the game’s skins. Most existing skins — over 1,200, according to Epic — will include a Lego style option which does exactly what it sounds like, rendering characters like Peely the banana in Lego bricks. It appears that these styles will only be useable in Lego experiences inside of Fortnite, however.

The launch of Chapter 5 comes at a very busy time for Fortnite. The game received a huge boost in popularity with the release of Fortnite OG in November, which brought back the original island and many fan-favorite locations, items, and characters. According to Epic, this led to the biggest day in the game’s history, with more than 44 million players on November 4th. This all culminated with The Big Bang live event on Saturday, which not only saw the Chapter 4 island destroyed, but also featured a virtual concert from Eminem.

The Who’s Who Behind the Modern Artificial Intelligence Movement

The Who’s Who Behind the Modern Artificial Intelligence Movement Before chatbots exploded in popularity, a group of researchers, tech executives and venture capitalists had worked for more than a decade to fuel A.I.

How Elon Musk and Larry Page’s AI Debate Led to OpenAI and an Industry Boom

How Elon Musk and Larry Page’s AI Debate Led to OpenAI and an Industry Boom The people who were most afraid of the risks of artificial intelligence decided they should be the ones to build it. Then distrust fueled a spiraling competition.

The second ‘Halo’ season has a release date and a teaser trailer

The second ‘Halo’ season has a release date and a teaser trailer

Master Chief, the Spartan hero from Halo, will continue his live-action adventures with two episodes of its second season premiering worldwide on February 8th next year. Paramount Plus announced the date, along with a teaser trailer, during CCXP in Brazil.

The teaser trailer doesn’t reveal much about the story, of course. But Paramount Plus says in its release that the new season will see Schreiber’s Master Chief trying to prove that the Covenant (the alien antagonists from the show and the games) are preparing to attack Reach — the humans’ military stronghold that fell prior to the story’s beginning in the first Halo game.

If you’ve played that game, you know where this goes, and Paramount Plus hints at it in the release. The Master Chief will go “on a journey to find the key to humankind’s salvation, or its extinction — Halo.” If you’ve played the games, you know... other things about what happens in between now and his arrival at the Halo ring. I won’t spoil anything for anyone who was first introduced to the series by the TV show, but it gets grim.

The first season was, visually, at least, a generally faithful adaptation of the games it’s based on, and it had some decent, gritty sci-fi future chops. I tend to agree with Ash Parrish’s review for The Verge — Master Chief could stand to lighten up a little, and the show’s overall tone could use some tweaking. Halo’s CGI aliens look silly, too (although I found that nostalgically charming). For those who haven’t seen it yet, Paramount Plus made it free to watch on YouTube a couple of days ago.

In addition to Schreiber, Halo’s second season will see the return of Natascha McElhone as Dr. Halsey and Jen Taylor as Cortana. Bokeem Woodbine, Shabana Azmi, Natasha Culzac, Olive Gray, Yerin Ha, Bentley Kalu, Kate Kennedy, Charlie Murphy, and Danny Sapani will also return. Joseph Morgan and Cristina Rodlo will join the series as James Ackerson and Talia Perez.

Federal judge vows to investigate Google for intentionally destroying chats

Federal judge vows to investigate Google for intentionally destroying chats

Google chief legal officer Kent Walker. | Photo By Alex Zea/Europa Press via Getty Images

Judge James Donato is overseeing Epic v. Google, a case that could determine the future of the Android app store — but testimony in this case may have more repercussions for Google too.

On Friday, Judge Donato vowed to investigate Google for intentionally and systematically suppressing evidence, calling the company’s conduct “a frontal assault on the fair administration of justice.” We were there in the courtroom for his explanation.

“I am going to get to the bottom of who is responsible,” he said, saying he would pursue these issues “on my own, outside of this trial.”

Testimony in the Epic v. Google trial — and in a parallel DOJ antitrust suit against Google in Washington, DC — revealed that Google automatically deleted chat messages between employees, and that employees all the way up to CEO Sundar Pichai intentionally used that to make certain conversations disappear. Pichai, and many other employees, also testified they did not change the auto-delete setting even after they were made aware of their legal obligation to preserve evidence.

And Pichai, among other employees, admitted that they marked documents as legally privileged just to keep them out of other people’s hands.

On November 14th, Pichai told the court that he relied on his legal and compliance teams to instruct him properly, particularly Alphabet chief legal officer Kent Walker — and so Judge Donato hauled Walker into court two days later.

But the judge was not satisfied with Walker’s testimony, either, accusing him of “tap-dancing around.”

Walker said he never attempted to audit whether employees were actually retaining evidence — it was left up to individual employees to decide which communications might be relevant to a legal case, and more than one employee testified in court they had the wrong idea of what was relevant.

“The most serious and disturbing evidence I have ever seen in my decade on the bench”

Today, Judge Donato said it was “deeply troubling to me as a judicial officer of the United States” that Google acted this way, calling it “the most serious and disturbing evidence I have ever seen in my decade on the bench with respect to a party intentionally suppressing relevant evidence.”

“This conduct is a frontal assault on the fair administration of justice. It undercuts due process. It calls into question just resolution of legal disputes. It is antithetical to our system,” said Judge Donato.

And yet, the judge decided today that he would not issue a “mandatory inference instruction” — one that would tell the jury they should proceed with the understanding that Google destroyed evidence that could have been detrimental to its case.

Instead, there will be a “permissive” jury instruction — the jury “may” infer that the missing evidence might have helped Epic and hurt Google.

“The best course of action is for the jury itself to decide whether it will make an inference. I am not going to constrain the jury’s discretion by making that inference for them,” he said.

“Even though it would be well within bounds to issue a mandatory inference instruction,” said Judge Donato, “I can pursue these issues on my own, outside of this trial, in subsequent trials.”

“I am going to get to the bottom of who is responsible,” he said. “That is going to be separate and apart from anything that happens here, but that day is coming.”

Google declined to comment to The Verge on Judge Donato’s statements.

Today, Epic and Google rested their case in Epic v. Google. We’ll be returning on December 11th for closing arguments and jury instructions.

Samsung Galaxy S24 leaks suggest a titanium build, flattened screen, and more

Samsung Galaxy S24 leaks suggest a titanium build, flattened screen, and more

Leaked images of the new Samsung S24 Ultra and standard S24. | Image: Windows Report

This is the time of year when Samsung unwillingly has major details of its flagship Galaxy S line of phones leaked out months before they launch. So it will come as no surprise that we’ve now got details on the S24 and S24 Ultra — or at least the phones we assume will carry that name.

Today, Windows Report got a hold of new official Samsung images that include the Galaxy S24 Ultra with a new flattened screen design. It confirms an earlier leak by Twitter / X user David Martin, who posted handheld images of the purported new S24 Ultra. The images clearly showed sharper screen edges akin to the changes made with the regular S23 from earlier this year, instead of the gentle curve, and a new gray color finish.

As reported last month, Samsung is expected to release new on-device generative AI software on its upcoming phones that use the company's Gauss model. Today’s leaks claim the S24 devices will use the tech to translate messaging apps in real time and in over a dozen languages. In October, Samsung also demonstrated a new AI-powered “Zoom Anyplace” camera framing feature powered by a Qualcomm AI engine — expected for release on a future device.

Image: Windows Report

S24 Ultras with titanium build and flatter screens.

Also today, Android Authority found that the FCC recently published major processor and modem specification details for the S24 line. While the commission did not explicitly state the chip, the information shows that the S24, S24 Plus, and S24 Ultra all have Qualcomm modems with Smart Transit functionality. The combination suggests every S24 model, at least in the US, will be equipped with the new on-device AI-capable Snapdragon 8 Gen 3 chips. However, there’s still a possibility international versions of the standard S24 could end up with Samsung’s own Exynos processors and modems like in previous years.

Along with the flatter screen and faster processor, the S24 Ultra is also getting a titanium frame, an upgraded 50MP 5x telephoto camera, better vapor chamber cooling, and Wi-Fi 7 (which you probably don’t need). The rest of the leaked specs by Windows Report show the new phone is not too different from the previous S23 Ultra model. It looks like the S24 Ultra will still have a 6.8-inch Quad-HD screen, 120Hz refresh rate, similar storage options, 12GB of RAM, and a 5,000mAh battery.

Image: Windows Report

Here are Samsung’s Galaxy S24 phones looking all... generational.

Meanwhile, the standard Galaxy S24 and S24 Plus model designs are looking similar to their S23 counterparts as previously confirmed in a September leak. But they do get new color finishes in onyx black, marble gray, cobalt violet, and amber yellow. Notably, the new Ultra also has color options this time around that match the regular S24 phones. The previous details correlate with the new leaks that point to a larger 6.7-inch Quad-HD screen, bigger 4,900mAh battery, and more RAM at 12GB for the S24 Plus. The regular S24, however, will get very small bumps in screen size at 6.2 inches and battery at 4,000mAh.

Samsung has not officially announced a successor to its Galaxy S23 smartphone line that was released earlier this year. As for the leaks themselves (and if history holds true), we should expect to see these phones announced early next year at a Samsung Unpacked event.

In firing Altman, OpenAI’s board wanted to keep the element of surprise

In firing Altman, OpenAI’s board wanted to keep the element of surprise

Illustration: The Verge

We still don’t know the whole story around the conflict between OpenAI CEO Sam Altman and the company’s board, but more and more is leaking out. A new report, from The New Yorker, alleges that OpenAI’s old board deliberately excluded Microsoft after initially voting to expel Sam Altman as CEO, and they actually believed the company would back them.

In the sprawling story, New Yorker reported Microsoft executives were blindsided by the decision to oust Sam Altman, possibly to prevent Microsoft CEO Satya Nadella from warning Altman what was about to happen.

Altman has a reputation for being a shrewd business person, and part of the decision not to inform Microsoft was to ensure that he would be surprised.

When OpenAI announced Altman’s ouster, it seemed the firing revolved around the debate over the dangers of AI and the belief that Altman had not been forthcoming with information. Shocking in hindsight, the board thought that Microsoft would understand their decision.

But the board had a different impression. After all, Microsoft, whose forays into AI had been choppy at best, created a responsible AI division and was concerned about the technology’s dangers. It wanted to avoid anthropomorphizing AI, too!

However, the company integrated GPT into its products very quickly and clearly wants more people to bring AI into their everyday lives. Microsoft CEO Nadella even told The Verge at the launch of the ChatGPT-powered Bing that he wanted to use AI as a way to make Google “dance.”

Which is probably why, as The New Yorker reported, many in Microsoft found the OpenAI board’s decision “mind-bogglingly stupid” and felt it could destroy the ChatGPT developer.

Microsoft executives then created three plans. The first was to back OpenAI CTO Mira Murati, who was appointed chief executive by the board for a few hours, and get leadership to reverse its decision or explain its reasoning. The second was to use its influence as the largest investor in the company to reinstate Altman, and the third, which it eventually did, was to hire Altman and anyone else who wanted to go with him to rebuild inside Microsoft.

Of course, in the end, Altman is back at OpenAI, and Microsoft joined the OpenAI board as an observer, and we still don’t know the precise reasons the board elected to fire Altman in the first place.

SpaceX will take some Amazon satellites to space

SpaceX will take some Amazon satellites to space

Image: Amazon

Amazon is enlisting the help of one of its biggest rivals to shuttle its Project Kuiper internet satellites into space. The company will use SpaceX’s Falcon 9 rocket for three launches beginning in 2025, Amazon announced on Friday.

Project Kuiper is Amazon’s low Earth orbit satellite program that could eventually provide users across the globe with internet connectivity. Amazon launched its first two Project Kuiper test satellites into space in October, where it managed to connect to the internet and conduct a two-way video call. However, it still has to do some serious catching up to SpaceX’s Starlink, which has already deployed around 5,000 satellites.

Despite the rivalry between the two companies, it doesn’t exactly come as a surprise that Amazon has chosen SpaceX as a future launch provider. Satellite companies and government agencies have become increasingly reliant on SpaceX’s reusable rockets for consistent — and cheaper — launches, as pointed out by an earlier report from The Wall Street Journal. The Elon Musk-owned company powered 88 percent of flights in the first six months of 2023 alone.

Most of Amazon’s satellites will still be ferried by Arianespace, Jeff Bezos’ Blue Origin, and United Launch Alliance (ULA), though. Amazon secured up to 83 launches with the three companies last year, which it says “provides enough capacity to launch the majority” of its planned 3,236 satellite constellation. It will deploy more Project Kuiper satellites in the first half of 2024, with beta testing starting toward the end of next year.

iA Writer can now track what you or ChatGPT wrote

iA Writer can now track what you or ChatGPT wrote

In iA Writer’s traditional minimalist interface, AI text shows up in grey, while your own words an in black. | Image: iA Writer

iA Writer 7, the latest version of the minimalist multi-platform writing software, has a new feature that’s designed to clearly mark text contributed by generative AI systems like ChatGPT. While your own words appear in black, you can choose for AI-generated text copied into your iA document to be greyed out. Then, as you tweak and edit the AI additions you can easily keep track of how much text is your original work and how much is artificially generated.

In its blog post about the feature, the company outlines its philosophy about using AI-generated text in writing. Rather than letting AI become a “ghostwriter” that “takes over” and makes you “lose your voice,” iA Writer’s feature is designed to encourage you to use AI “as a dialog partner” that “makes you think more and write better.”

Image: iA Writer

iA Writer can sometimes recognize when text is AI-generated, but marking it as such is optional.

Although there are automated elements to the authorship feature, I found it to be a largely manual process during my brief testing. “It’s up to you to decide how honest you want to be with yourself,” iA’s blog post notes. Although the software recognized my AI text when I copied it in from ChatGPT’s website on desktop, it only did so when I included the original prompt alongside ChatGPT’s response. iA Writer didn’t recognize text from ChatGPT if I only copied over the generative AI’s response without my prompt. To manually get it to appear as AI text, you can right click, select “Paste As” and then choose “ChatGPT.”

As a result, the feature is less about detecting plagiarism or watermarking AI-generated content, and more about giving creatives a tool to see the sometimes blurry line between their own words and those contributed by generative AI.

The authorship feature is only available with iA Writer’s Mac, iOS, and iPadOS apps for now, and the company says it’s planning to bring it to iA Writer’s other platforms like Windows and Android in the future. But more interesting is that iA has published the spec on Github, and says it’s open to working with other apps to “ideally” turn the feature into a new standard.

What’s Next for OpenAI, Binance Is Binanceled and A.I. Is Eating the Internet

What’s Next for OpenAI, Binance Is Binanceled and A.I. Is Eating the Internet It’s full steam ahead for the tech world despite, well, everything.

Resident Evil 4 remake will add VR mode in December

Resident Evil 4 remake will add VR mode in December

Image: Capcom

The Resident Evil 4 remake will soon get a change in perspective. On December 8th, the PS5 version of RE4 will get a free update adding VR mode to the game.

“Resident Evil 4 VR Mode supports the full main story campaign and delivers an added level of immersion to Leon S. Kennedy’s harrowing rescue mission,” read the press release. “The experience lets players see the terrifying world directly through Leon’s eyes and soak in his surroundings with the 4K HDR display of PlayStation VR2.”

If the prospect of gunning down zombies in VR feels a bit too challenging to you as it does to me, RE4VR will feature a shooting range game mode that lets you acclimate to the controls and test the kinds of weapons Leon will use. And if that still proves too much, the original Resident Evil 4 also has its own VR version available on the Meta Quest 2.

A VR version is only one part of all the special attention being paid to the RE4 remake. The game is one of the more unlikelier candidates for this year’s Game of the Year award, and it’s also one of the games selected to debut on the Apple iPhone 15 Pro alongside Resident Evil Village, Death Stranding, and Assassin’s Creed Mirage.

Resident Evil 4 VR Mode launches on the PSVR2 December 8th.

Tesla delivered the Cybertruck: here’s how to preorder one

Tesla delivered the Cybertruck: here’s how to preorder one

A newer showcase model of the Cybertruck in New York City. | Photo by Amelia Holowaty Krales / The Verge

The Cybertruck is here — for real this time — and for those who haven’t reserved one already, Tesla has a new ordering process. Today, you can preorder the visually perplexing, low-poly pickup truck on Tesla’s website with a refundable deposit of $250.

A small batch of early preorder holders took delivery of the first production Cybertruck models at an event in Texas. Since reservations went up on Tesla’s website in 2019, the automaker happily accepted $100 deposits to get in line to buy a non-configured Cybertruck model. During the last earnings call, Tesla CEO Elon Musk confirmed there were over 1 million preorders.

For a short while earlier this week, the price to hold a place in line for the Cybertruck went up to $250. Afterward, Tesla shut off preorders and added a countdown timer for the delivery event. Now, Tesla has reopened preorders and is inviting interested buyers to drop $250 to join the back of the line. It lets you choose between a rear-wheel drive, all-wheel drive, and new high-end “Cyberbeast” version of the truck, with a new starting price of $60,990. The Cybertruck options are as follows:

Once you pay your “refundable” deposit, you’ll get an invitation when your Cybertruck is ready to be configured. It’s not confirmed if the Cyberbeast model carries the previously announced tri-motor setup.

Tesla originally promised a single-motor Cybertruck model with 250 miles of range on a single charge for $39,900, a dual-motor version capable of 300 miles for $49,900, and a top-tier triple-motor one with an almost unbelievable 500 miles of range for $69,900.

The Cybertruck has made its mark in 2019 as Tesla’s sixth production vehicle and one of the first announced all-electric pickup trucks for consumers. Since then, Ford released the F-150 Lightning and Rivian delivered the R1T, and people can actually buy them right now. But there’s a completely separate appeal to the Cybertruck, and it’s certainly looking better today than it did as a prototype on the road.

---

Related:

Windows 11 tests energy saver mode for both laptops and desktop PCs

Windows 11 tests energy saver mode for both laptops and desktop PCs

Illustration by Alex Castro / The Verge

Microsoft is adding an energy saver mode for Windows 11 Insiders (Build 26002) that “extends and enhances” the existing battery saver option. The new mode doesn’t apply to just laptops like battery saver does — you can also use energy saver on a desktop computer if you want to conserve electricity, too.

Aside from availability on both desktop devices and laptops (even when they’re plugged in), energy saver works pretty much the same as battery saver mode. The new feature also helps save energy by “trading off some system performance,” so you might not want to use it if you’re gaming.

Image: Microsoft

You can enable energy saver from the Quick Settings menu, or you can configure it from the Settings menu by heading to System > Power > Energy saver. There, you can toggle on the feature or choose to run it automatically whenever your device reaches a certain battery percentage. Energy saver mode is rolling out now to Windows 11 Insiders in the Canary channel.

If you don’t have access to this build, you can still use the existing battery saver mode available to everyone with Windows 11. You can also optimize your power consumption with the system’s energy recommendations feature, which automatically adjusts certain settings, like screen brightness, to save power.

Google beefs up Drive’s document scanning feature and brings it to iPhone

Google beefs up Drive’s document scanning feature and brings it to iPhone

Illustration by Alex Castro / The Verge

Google Drive users on iOS are getting a new scanning feature that lets them quickly photograph documents and add them to their libraries, the search giant has announced. The feature has been available for Android for over a decade, but until now iPhone and iPad users haven’t had the same option. Google has also upgraded the scanner in Drive’s Android app.

iPhone users in the US will also benefit from a new title suggestion feature that attempts to auto-generate a document title based on text it recognizes in the scan. The feature previously came to the Android version of Drive earlier this year. Google also says it’s improving document scanning on Android, adding similar features like the option to automatically capture documents when they’re in frame. 9to5Google spotted the rollout of this new Android interface last week.

Image: Google

The document scanning process on Android.

At this point, most of the major cloud storage providers now offer native scanning features to quickly capture documents using your phone’s camera. Microsoft launched its Office Lens (now Microsoft Lens) document scanning app in 2015, and Apple built a similar scanner into its Notes app in 2017.

Google says the new iOS features are rolling out now, and are already available on Android. Google’s scanner is accessible in the Drive app on iOS and Android either via the camera icon on the bottom right, or by tapping the “+” icon and selecting “Scan.”

U.S. Debates How Much to Sever Electric Car Industry’s Ties to China

U.S. Debates How Much to Sever Electric Car Industry’s Ties to China Some firms argue that a law aimed at popularizing electric vehicles risks turning the United States into an assembly shop for Chinese-made technology.

London’s famed black cabs will be listed on Uber’s app in big win for the ridehail company

London’s famed black cabs will be listed on Uber’s app in big win for the ridehail company

Photo by Mike Kemp / In Pictures via Getty Images

Uber notched another win in its effort to win over the beleaguered taxi industry, announcing a plan to start listing London’s famed black cabbies in its app.

The service won’t roll out until early 2024, but some London cabbies have already begun to sign up. Uber says all 15,000 of London’s cab drivers “will now have the opportunity” to sign up for Uber trip referrals. The company recently brokered deals with taxi fleet owners in New York City, Paris, Rome, and Los Angeles to list drivers in its app.

Next to New York’s yellow cabs, the black cabbies of London are arguably the most iconic taxis in the world. Not only is this a symbolic victory for Uber but also it could help build trust with taxi owners who may still be wary of Uber’s motives.

Indeed, not everyone is happy about Uber’s attempts to bring cabbies into its app. The Licensed Taxi Drivers’ Association (LTDA), a trade group that represents 10,000 drivers, cast doubt over whether drivers would flock to the app.

“There is no demand for this partnership from the London licensed taxi drivers we represent or our passengers,” LTDA general secretary Steve McNamara said in a statement. “Neither the LTDA nor any other taxi trade groups were consulted in advance of this unilateral announcement. We are not aware of any drivers having been recruited and don’t believe our members will even consider joining the app, given its well-documented, poor record on everything from passenger safety to workers’ rights in London.”

McNamara noted that riders can already hail a ride in a London black cab through numerous apps, including Gett, Taxiapp, FreeNow, and ComCab. “We have no interest in sullying the name of London’s iconic, world-renowned black cab trade by aligning it with Uber, it’s poor safety record and everything else that comes with it,” he added.

“We have no interest in sullying the name of London’s iconic, world-renowned black cab trade by aligning it with Uber”

Uber has long been at odds with the taxi industry. In the early days, the company’s habit of playing fast and loose with the rules irked taxi owners, who accused the company of ignoring local regulations when moving into new markets. Uber retorted that the taxi business had many flaws before it arrived, including predatory loans.

But after it failed to completely wipe out and replace the taxi business, Uber instead turned to taxis to help fuel its next stage of growth. The company has said that, by 2025, it hopes to list every taxi in the world on its app. And for once, taxi owners are eager to be involved.

Taxis are featured in the Uber app in 33 countries around the world, with “hundreds of thousands” of taxi drivers receiving trip referrals from the company. Some of the largest markets by volume include Hong Kong, Poland, South Korea, Sweden, and Turkey. Last year, Uber struck a deal to include around 14,000 of New York City’s iconic yellow taxis in its app.

When a taxi is hailed through Uber, the company gets a cut. Uber’s average global take rate (also known as its revenue margin) for rides in the third quarter of this year was 28.3 percent, up from 27.9 percent in Q3 2022. Uber said it would waive its commission on trips for London cabbies for the first six months.

Taxis are featured in the Uber app in 33 countries around the world

Both Uber and London’s black cabs are doing pretty well these days, having rebounded from the covid pandemic. More new cabs are being registered, including electric-powered ones. And Uber secured a 30-month license to keep its ridesharing services up and running in London last year after a lengthy battle with Transport for London over the company’s safety record.

Last year, Uber lost a legal battle in the UK over the employment status of its drivers. This required the company to start classifying its UK-based drivers as employees, granting them minimum wage, paid vacation, and other benefits.

London cabbies are known for their adherence to “the Knowledge,” a seemingly uncanny ability to locate thousands of landmarks within the greater London area with precision. Cabbies study for up to three years and spend around £10,000 (about $12,707) to memorize all the details.

Amazon’s Q AI assistant lets users ask questions about their company’s data

Amazon’s Q AI assistant lets users ask questions about their company’s data

AWS CEO Adam Selipsky | Noah Berger

Amazon’s cloud business AWS launched a chat tool called Amazon Q, where businesses can ask questions specific to their companies.

Announced during a keynote speech by AWS CEO Adam Selipsky at AWS re:Invent, Amazon Q acts like an AI assistant where users can ask questions about their businesses using their data. For example, employees can query Amazon Q on the company’s latest guidelines for logo usage or understand another engineer’s code to maintain an app. Q can surface the information instead of the employee sifting through dozens of documents.

Users can access Amazon Q through the AWS Management Console or individual companies’ documentation pages, developer environments like Slack, and other third-party apps.

Selipsky noted that questions asked on Amazon Q “will not be used to train any foundation models.”

Amazon Q can work with any of the models found on Amazon Bedrock, AWS’s repository of AI models, which includes Meta’s Llama 2 and Anthropic’s Claude 2. The company said customers who use Q often choose which model works best for them; connect to the Bedrock API for the model; use that to learn their data, policies, and workflow; and then deploy Amazon Q.

AWS said Amazon Q was trained on 17 years’ worth of AWS knowledge and can be used to ask questions specific to AWS use. It can suggest the best AWS services for a project.

Currently, Amazon Q is available only for users of Amazon Connect, AWS’s service for contact centers. Eventually, it will be available on other services like Amazon Supply Chain, which helps customers track their supply chain management, and Amazon QuickSight, its platform for business intelligence. Amazon Q for supply chain and business intelligence is available on preview.

Dilip Kumar, vice president for AWS Applications, told The Verge in an interview that each instance of Amazon Q on AWS services will look different. On Amazon Connect, Q is deployed in real time and essentially listens in on a customer call to get information like account details. It gives the contact center agent the relevant answers to questions without the agent needing to find it themselves.

“We wanted to pair the technology with the services that make the most sense first, and for contact centers, supply chain, and business intelligence, AI is a natural fit,” Kumar said.

Pricing for Amazon Q in Connect starts at $40 per agent per month. According to AWS’s Connect website, users can try Amazon Q in Connect “for no charge until March 1, 2024.”

Selipsky said Amazon Q recognizes security parameters set up by customers, so employees without access to some information cannot use the query system to get data they’re not allowed to see.

Other companies have built similar products. Microsoft’s Copilot does something similar for Windows users, Dropbox’s Dash lets people query their stored documents, and Notion announced an AI-powered notes search feature this month.

Along with the launch of Amazon Q, Selipsky also said AWS would offer Bedrock users the ability to put guardrails around models they use to build AI-powered apps. Now on preview, the guardrails let companies ensure their applications and the models they use to power them follow their data privacy and responsible AI standards.

Companies, particularly those in highly regulated industries like finance and healthcare, often point to their inability to fence off their data and make sure their information is not used to train future versions of the model. AWS said the ability to redact the personally identifiable information of their customers’ end users will also be part of the guardrails, but it is not immediately available yet.

AWS’ transcription platform is now powered by generative AI

AWS’ transcription platform is now powered by generative AI

Photo by Jaap Arriens/NurPhoto via Getty Images

AWS added new languages to its Amazon Transcribe product, offering generative AI-based transcription for 100 languages and a slew of new AI capabilities for customers.

Announced during the AWS re: Invent event, Amazon Transcribe can now recognize more spoken languages and spin up a call transcription. AWS customers use Transcribe to add speech-to-text capabilities to their apps on the AWS Cloud.

The company said in a blog post that Transcribe trained on “millions of hours of unlabeled audio data from over 100 languages” and uses self-supervised algorithms to learn patterns of human speech in different languages and accents. AWS said it ensured that some languages were not over-represented in the training data to ensure that lesser-used languages could be as accurate as more frequently spoken ones.

In late 2022, Amazon Transcribe supported 79 languages.

Amazon Transcribe has 20 percent to 50 percent accuracy across many languages, according to AWS. It also offers automatic punctuation, custom vocabulary, automatic language identification, and custom vocabulary filters. It can recognize speech in audio and video formats and noisy environments.

The Verge reached out to AWS for information on previous accuracy and which foundation models it used for Amazon Transcribe.

With better language recognition, AWS said advances with Amazon Transcribe also bleed into better accuracy with its Call Analytics platform, which its contact center customers often use. Amazon Transcribe Call Analytics, now also powered by generative AI models, summarizes interactions between an agent and a customer. AWS said this cuts down on after-call work creating reports, and managers can quickly read information without needing to go through the entire transcript.

Of course, AWS is not the only company offering AI-powered transcription services. Otter has been providing AI transcriptions to consumers and enterprises for a while and released a summarization tool in June. While not exactly the same, Meta announced it is working on a generative AI-powered translation model that recognizes nearly 100 spoken languages.

AWS also announced additional capabilities to its Amazon Personalization product, which allows clients to offer products or show recommendations to customers, like how streaming services can suggest new shows based on previous activity. AWS added Content Generation, which will write titles or email subject lines to thematically connect recommendation lists.

YouTube is getting into games, too

YouTube is getting into games, too

Illustration by Alex Castro / The Verge

YouTube is branching out into games — at least for its paid subscribers. The platform is giving Premium users access to a set of online games that can be directly played on either the mobile app or desktop app. Known as “Playables,” the company first debuted the experimental feature to select users in September. As noted by Droid-Life, YouTube sent a notification last week to Premium subscribers informing them of Playables and allowing them to try it out. Those who opt in will be able to play a total of 37 mini-games that effectively live inside YouTube — there’s no need to download or install them.

The selection of games isn’t too challenging or “out there” — they include crowd-pleasers like Angry Birds Showdown, Brain Out, Daily Solitaire, The Daily Crossword, and a number of arcade games. And they may not be here to stay. YouTube Premium’s notification stated that the games would be available until March 28th, 2024. For now, Premium members can find the full library of games under the “Playables” section in the Explore tab.

YouTube joins a long list of non-gaming tech companies that are trying to expand into games — often with mixed results. After years of hype, Google officially axed its Stadia service in January. Just a couple of weeks ago, Amazon cut over 130 jobs in its free games division and announced it would “refocus” its efforts. In 2021, TikTok teased what it called a “major push” in gaming, only for owner ByteDance to announce on Monday that it would be laying off around 1,000 workers in its gaming unit.

But a less than stellar track record hasn’t stopped other tech companies from pushing forward into gaming. Meta’s experiment with Instant Games has been roughly seven years in the making; the platform this month launched a new distribution model that lets developers publish beta versions of their games directly on Facebook. Netflix has been releasing games exclusively on mobile (the reviews of which are pretty mixed), and wants to expand into cloud games — though it’ll likely be a while until they’re available to the public.

Great to see Phaser games as YouTube Playables. Worked flawlessly, nice integration (only tested on desktop) pic.twitter.com/hzuNHgWmV5

— threads.net/@photonstorming (@photonstorm) November 27, 2023

YouTube’s experiments are nothing new; the platform often tests features on its paid subscribers before deciding whether to add them permanently. In addition to Playables, Premium members can also sign up to try out YouTube’s conversational AI feature that lets users ask questions about videos they’re viewing.

YouTube’s Playables seems less of an appeal to serious gamers than a way to sweeten the pot for its paid subscribers. The company raised the price of its Premium plan by $2 this summer, and there have been similar streaming subscription price increases by Netflix, Apple, Amazon, and others. As these services get pricier, some users will naturally cancel their subscriptions. While it’s hard to imagine someone explicitly signing up for YouTube Premium solely for the Playables games, they may convince some on-the-fence users to stay for longer.

This company just put the air in Apple’s MacBook Air

This company just put the air in Apple’s MacBook Air

A pair of 15-inch M2 MacBook Air laptops at Frore Systems’ HQ in San Jose, California. | Photo by Sean Hollister / The Verge

Two Fridays ago, I drove down to a squat single-story Silicon Valley office building to see a MacBook Air. There was no security guard, no need to badge in, not even a PR person in my demo room. That’s because I wasn’t visiting Apple headquarters for an unannounced laptop. Instead, I went to San Jose to see an existing MacBook — one surgically modified to prove how far an exotic cooling tech has come.

Frore Systems is a startup with $116 million in funding, and I’ve shown you its first product before: the AirJet Mini is a piezoelectric cooling chip that weighs just nine grams and is thinner than two US quarters stacked together. Each nominally consumes one watt and can remove 4.25 additional watts of heat. Here’s the question: what would happen if Frore used those AirJets to cool a laptop that normally doesn’t have a fan at all?

What the company discovered — and I saw firsthand — is that Apple’s M2 chip can run faster, for longer, with Frore’s tech on board. Without it, a 15-inch M2 MacBook Air was like a runner that can’t sprint indefinitely without running out of breath. But with three AirJet Minis, the same laptop got a permanent second wind.

Here’s the bottom line: the longer I ran an intensive benchmark on Frore’s modified and unmodified pair of 15-inch MacBook Airs, the bigger a difference I saw.

Even during the very first of a multi-minute benchmark like Cinebench R23 or Shadow of the Tomb Raider, the AirJet version pulled out slightly ahead, though perhaps within the margin of error at, say, 29 frames per second with AirJet versus 28 frames per second without. The Xcode benchmark finished in 172.7 seconds with AirJet versus 178.2 seconds without the first time around.

Photo by Sean Hollister / The Verge

Three AirJet Minis and a custom heatsink inside a 15-inch M2 MacBook Pro.

Photo by Sean Hollister / The Verge

The reverse of that heatsink — the AirJets strongly suck air through a stack of copper fins atop a custom heat spreader that screws down onto Apple’s logic board.

But after half an hour of back-to-back Shadow of the Tomb Raider gaming benches, scores had dipped to an unplayable 22fps on the stock Mac versus 27fps on the AirJet version. I started to see massive stutter on the stock Mac at the 40-minute mark — while Frore’s AirJets just kept breezing on.

With Cinebench R23, a video editing benchmark, multicore scores seriously dipped after just a few repeat benches — from 8775 with AirJet to 8380 without. According to a Frore-provided overlay that I watched like a hawk, the stock Mac simply wasn’t able to maintain the same CPU clocks. Both clock speed and wattage plummeted on occasion, presumably because a fanless computer has no other good way to cool itself.

What’s happening here is fairly well understood: today’s processors are only as fast as they can be cooled, with many advertising “turbo” modes that the thinnest computers can only maintain for a short time. For example: Apple’s Air can run the M2 chip at 3.2GHz but throttles down to 2.8GHz after 30 minutes of Cinebench R23, says Frore.

Image: Frore

A chart from Frore showing how the Apple M2’s megahertz and processor power can take a dive during sustained tasks.

But as neat as it was to see the AirJets give Apple’s M2 a second wind, I also walked away a tad underwhelmed. Tomb Raider stutter aside, we’re mostly talking about an already capable computer simply taking a bit longer to perform the most intensive tasks, with less demanding benchmarks (like single-core Cinebench, where I saw identical scores) not stressing the chip enough to make a real difference.

More importantly, we already know what Apple’s M2 chip can do when you add airflow — it’s called a 13-inch MacBook Pro!

As I reported a few years ago (back when I could ethically cover Apple products), the biggest difference between the 13-inch MacBook Pro and the 13-inch MacBook Air is literally a fan. Since then, we’ve clearly seen in our own M2 MacBook Air and MacBook Pro benchmarks that performance tends to deteriorate over 30 minutes without that fan on board.

Frore knows this and brought a 13-inch M2 MacBook Pro for me to test alongside the Airs — but it’s not really an apples-to-apples test if I’m comparing them to a 15-inch MacBook Air with a wider chassis and more thermal headroom. Even so, Frore’s modified 15-inch Air was mostly neck and neck with the 13-inch Pro, rather than regularly beating it. (The Pro did take four seconds longer to complete a run of Xcode.)

Photo by Sean Hollister / The Verge

Frore carved out a little extra space in the 15-inch MacBook Air’s lid.

But there is one good reason Frore picked the 15-inch MacBook Air — at 0.45 inches (11.5mm) thick, it’s one of the thinnest laptops ever made. It’s not clear Apple could fit the fan from its 13-inch MacBook Pro in there — or any fan, for that matter. Even Frore had to painstakingly mill 0.3mm out of the laptop’s lid to give the AirJets a big enough air gap to do their thing, and the company wound up removing the speakers, Wi-Fi antenna, and even the Mac’s internal keyboard connector along the way.

But Frore did it, and engineering chief Prabhu Sathyamurthy tells me laptops can get even thinner if they’re designed with AirJet in mind — they could reach 9.5mm in thickness simply by substituting a thinner keyboard and a thin OLED display, the company claims.

Image: Frore

Click to enhance and maybe make those numbers large enough to read.

There is one thing I didn’t see accounted for in Frore’s proposal, though, and didn’t get to satisfactorily test at Frore’s offices: the all-important battery.

While Frore says each of its AirJet Mini chips require one watt of power, I saw three of them consume over five watts from the MacBook’s USB-C port, and the AirJet-equipped MacBook Air (obviously) drained far faster than the power-throttled fanless one did during my tests.

Mind you, Frore’s been clear since the beginning that its chips need to be integrated into a device for them to work optimally, and the MacBook demo is anything but where power is concerned. Sathyamurthy says you can expect the AirJet part of a complete system to draw as little as 0.1W or 0.2W when idle, with the AirJet Minis themselves staying turned off until or unless they need to blow. He thinks the average user will see them kick on 10 to 15 percent of the time and not at all for “emails and web surfing.” We’re talking a Zoom call or worse.

Photo by Sean Hollister / The Verge

For demo purposes only: a USB-C power cable for the AirJets where the MacBook’s speaker used to be.

And when they do blow, they blow far quieter than a fan, I can confirm. I had to bend down next to the laptop to hear them at all. The modified laptop also seems a tad cooler than Apple’s fanless MacBook Air — a thermal camera showed me a wider, cooler region instead of a hotspot near the chip.

I can’t say this demo convinced me that Frore can make the fan obsolete. I mostly saw it meeting the performance of a fan, with less noise in a tighter space. But that could mean a lot to the right company trying to push the envelope in portability or — pretty please! — using those space savings for additional batteries.

I wonder if any laptop manufacturer will give it a go. So far, Frore’s only confirmed design win is this Zotac mini-PC that’ll always be plugged into a wall, but it has prototyped 4K webcams, doorbell cameras, LED light bulbs, and more. Sabrent, Phison, and OWC tried experimental AirJet-cooled SSDs. Frore seems to think AirJets could fit into a Steam Deck, too.

I’d love to try a battery-powered system where we can fully see the AirJet way is better. And I’ll let you know as soon as I do. Frore says the AirJet Mini is in mass production now, with larger and smaller versions in development.

View this post on Instagram

A post shared by The Verge (@verge)