Category: TECH

Rivian has had a lot on its plate as it transitioned from pitch mode to selling EVs. It created an electric pickup and an electric SUV while prepping a monster IPO. It made electric delivery vans for Amazon and wants to do the same for other companies. It now plans to sell an even cheaper SUV that could make Rivian a dominant EV player for years to come. And it wanted to build an entirely new factory in Georgia where it would manufacture many of these vehicles.

With so many variables, the exact shape of the company’s future was hard to predict.

That has changed.

Earlier on Thursday, the company announced a fully revamped version of its first two consumer vehicles, the R1T pickup and the R1S SUV. Not only are they more technologically advanced, Rivian also made them more simple in a bid to dramatically slash the cost of building them.

Rivian also recently set aside its plan to build that factory in Georgia for now, opting instead to double-down on its existing facility in Illinois. The decision is going to save the company $2.25 billion and means it can focus all its efforts on one manufacturing staff in one factory.

These changes mean that, for the first time since the company broke stealth in 2018, Rivian’s immediate future is actually remarkably clear. The company needs to sell these revamped vehicles at a profit in order to sustain itself long enough to get to the cheaper mass market R2 SUV on the road (and the adorable R3 variant that took the automotive world by storm earlier this year). It knows exactly where that will happen, and it knows what it will take to get there.

“With Rivian’s latest move to refresh the R1T and R1S EVs, you can begin to see how the company aims to chart its way forward across the ‘EV valley of death,’” Corey Cantor, senior associate for electric vehicles at BloombergNEF, said in an email to TechCrunch. “If successful, they can use the fruits of this process as they scale-up the R2 and reach the mass market, en route to the R3.”

Other EV startups arguably have a harder path through that “valley of death.”

Take Lucid Motors, for example. The company has a well-regarded product in the Lucid Air sedan. But it has struggled to find buyers for the Air, with its own CEO Peter Rawlinson publicly admitting to failures on the marketing side. It’s only shipped around 12,000 cars to date, at least as of the end of the first quarter of this year.

Lucid Motors is now pinning a lot of its hopes on the forthcoming Gravity SUV. That vehicle should have wider appeal, given the popularity of the SUV form factor. But its success is nowhere near guaranteed, especially because it’s starting at a relatively high price point of “under $80,000.” And Lucid Motors needs the Gravity to succeed if it ever hopes to get to its own planned midsize, mass-market EV.

Other EV startups face more uncertainty. Canoo has changed its business model so many times that it’s often hard to keep tabs on what it plans to do with its bulbous EVs, first revealed in 2019. (Currently, the plan is to sell to fleets and government entities.) Faraday Future has been spending as much time fighting with its landlords as it has trying to sell its own luxury EV. Fisker is on the brink of bankruptcy after dealing with underwhelming sales of its electric SUV and myriad quality and service problems.

It won’t be easy for Rivian. The company is forecasting essentially no growth this year compared to 2023, and it started off on a flat foot. It may need to raise more money as a result — a challenging feat in the current economy.

But company says the changes to the R1 lineup set it on the path to reaching “positive gross profit” by the end of this year. That’s a big deal considering Rivian is still losing tens of thousands of dollars on each car it sells. If Rivian wants to survive long enough to ship its more affordable mass-market R2, it really needs these revamped vehicles to sell well.

“The path ahead is clearer than it was a year ago as Rivian has laid out its near-term plans,” Cantor said. “But ultimately execution of both profitability and high-volume EV sales is what is required for Rivian to become one of this decade’s EV success stories.”

Threads, Meta’s Twitter/X competitor and the company’s first bet on decentralized social media, is now making it easier for users to control their Threads experience. After launching a customizable dashboard interface last week, Instagram head Adam Mosseri announced Monday that the app is now rolling out the ability for users to signal which sort of posts they wanted to see more or less of.

Users would “swipe right on a post to like it, or swipe left to show you’re not interested,” Mosseri explained in a post on Threads. “We’ll use those signals to show you more posts like the ones you swipe right on and fewer of those you swipe left on.”

The idea to use a swipe gesture to indicate interest is a user interface interaction that recalls dating apps like Tinder, but the ability to train an algorithm about what sort of things people like could help Threads more quickly get up-to-speed on personalizing users’ For You feeds. TikTok, for example, uses a similar mechanism by allowing users to mark videos as “not interested,” and X does the same for posts.

Across Threads’ competitors, there are a number of different approaches to how content should be delivered. X uses a more traditional social media algorithm that leverages likes and engagement patterns alongside other metrics to determine what sort of content people would like to see. On the flip side, the startup Bluesky, originally incubated inside Twitter (now X), offers a “choose your own algorithm” model where users can customize feeds to their liking or follow prebuilt feeds like “What’s Hot” or “Popular with Friends” to view the network’s content in the way they prefer instead.

Well-designed For You feeds, however, can increase engagement and time spent with an app — areas Meta is looking to improve with Threads.

Though Meta’s newest app today counts more than 150 million monthly active users, for comparison, Elon Musk claims X now has 600 million monthly active users, 300 million of whom use the platform daily. (He doesn’t indicate what portion of that user base consists of automated accounts or spam, though — and as X is no longer a publicly reported company, the numbers can’t be externally verified.) Still, it’s fair to say that Threads has further to go to catch up with X, though it’s outpacing Bluesky’s 5.7 million users and Mastodon’s 7.2 million, less than 1 million of whom are active on a monthly basis at present.

Threads first began testing the side-swiping gesture to signal interest in posts in March, where swiping one way would reveal a heart icon and the other would display an icon of an eye with a line crossed over it — a symbol typically used for “hide” or “hidden.”

While Threads today says it will use the swipes to help users customize their own feeds, the signals Meta collects could be put to a larger purpose over time. With enough data, the swipes could also help the company determine what posts are popular versus those that are largely being “downvoted” (swiped left on) by the community. This could help the app’s For You algorithm more quickly improve.

When someone says a product is “new and improved,” it’s wise to take it with a grain of salt. But with Mill’s redesigned food waste bin, you can believe it.

As before, the bin accepts a wide variety of food waste — only a handful of items like oyster shells are off limits — and grinds and dries it to a consistency that looks like chunky coffee grounds. Those grounds can be mixed with garden soil, spread on lawns, or even shipped back to Mill, which then offers it to farmers as chicken feed. A household using the bin can expect to trim about half a ton of greenhouse gas emissions annually.

So what’s different? Just about everything.

Where the old bin worked as promised, it wasn’t always as quiet or fast as I would have liked, sometimes taking nearly a day to complete a cycle of drying and grinding the food. That’s not the case with the new one, which I’ve spent the last couple weeks testing. Every night at 10 pm, my bin started a cycle, and by the time I woke up, it was always finished, just as co-founder Matt Rogers promised me. What’s more, it’s significantly quieter, no longer disrupting evening TV viewing.

Here’s how Mill made it happen.

The design brief was simple, said Kristen Virdone, head of product at Mill: each cycle had to be completed before breakfast. With that guidepost, and a year’s worth of data under their belts, the team dug in.

From the outside, the new Mill bin doesn’t look that different. The visual changes are so subtle you’d have to be paying close attention to notice them, like when automakers tweak a model’s headlights to freshen the appearance. Probably the biggest aesthetic change is the fact that the status lights no longer shine through the wood-grain plastic lid, a nifty bit of shy tech that I kind of miss.

Under the lid, one of the biggest changes users will notice is that the augers that grind the waste are now vertical instead of horizontal. That change allowed the team to make the bottom of the bucket flat instead of rounded, making it easier for the augers to sweep clear. It also helped eliminate untoward noises. Previously, the augers would drag food waste across the curved bottom, creating what the Mill team calls “haunted house noises.” (To me, it always sounded like a creaking and groaning pirate ship.) The new configuration exorcized those demons. 

The vertical arrangement also gave the design team an opportunity to add little paddles to the tops that users can twist to help dislodge grounds when they’re emptying the bucket.

The bucket itself is now made entirely of metal. The previous one had some plastic parts, which reduced how much heat could be transferred from the heating element to the food waste, lengthening drying times. To help the grounds slide out, the bucket is lined with a PFAS/PFOA-free ceramic coating.

To further reduce cycle times, the Mill team was able to use machine learning algorithms trained on data gathered over the last year, Virdone said. As a result, the new software is smarter about how long each cycle needs to run. 

Each bin also has a suite of sensors, just like the previous version, though now the team has enough data that it can differentiate between the weight of one strawberry and four raspberries, said Suzy Sammons, Mill’s head of communications. Two humidity sensors, one on the air inlet and one on the exhaust, help the bin to understand exactly how long each drying cycle needs to run.

“If you think about it, there are infinite combinations of food that can go into our bins,” Virdone said. “Having a year under our belt, and having real families put in real weird combinations of foods, we start to see the bounds of what’s in there.”

The fans have been completely redesigned, too, Virdone told TechCrunch. They’re quieter, and their location within the bin was rethought with an eye toward minimizing the amount of noise that escapes the unit. In aggregate, the changes worked well. The new unit’s fan noise was significantly reduced during my testing.

The only thing I noticed missing from the new bin is a power-activated lid. On the old model, stepping on the foot pedal would signal a motor to swiftly lift the lid. It was oddly satisfying to use, and my kids loved it, too. The new one is a more traditional, linkage-operated lid that’s physically connected to the pedal, like a stereotypical kitchen garbage bin. Virdone said that user testing revealed that people preferred the mechanical lid, saying it was more intuitive than the motorized version. 

Like the old bin, the new one requires a power outlet nearby. In our house, that means the bin technically lives in the family room, just a few steps away from the kitchen sink. It works out just fine in practice, though it looks a bit out of place when you’re sitting on the couch. If I were going to make a permanent home for it, I’d want to find it a home somewhere in the kitchen, maybe adding another outlet in the process.

Apart from that, the only thing that’s preventing me from buying one is the price. At $360 per year, it’s not cheap, especially compared with the unsubsidized curbside compost service in my city, which is a third the cost. Mill’s new price is about 10% cheaper than before, provided you have somewhere to dump the grounds. If you don’t, you’ll have to add $10 per month to get it picked up. It’s possible the price will come down if Mill is able to negotiate subsidies through municipalities. Currently, the only cities that have deals with Mill are Pittsburgh and Tacoma, Washington..

Given the current cost, Mill’s bin still isn’t for everyone. But for households who don’t have curbside composting services available, or that don’t like the smell that accompanies them, it’s a great product that’s gotten even better.

AI models are always surprising us, not just in what they can do, but what they can’t, and why. An interesting new behavior is both superficial and revealing about these systems: they pick random numbers as if they’re human beings.

But first, what does that even mean? Can’t people pick a number randomly? And how can you tell if someone is doing so successfully or not? This is actually a very old and well known limitation we, humans, have: we overthink and misunderstand randomness.

Tell a person to predict heads or tails for 100 coin flips, and compare that to 100 actual coin flips — you can almost always tell them apart because, counter-intutively, the real coin flips look less random. There will often be, for example, six or seven heads or tails in a row, something almost no human predictor includes in their 100.

It’s the same when you ask someone to pick a number between 0 and 100. People almost never pick 1, or 100. Multiples of 5 are rare, as are numbers with repeating digits like 66 and 99. They often pick numbers ending in 7, generally from the middle somewhere.

There are countless examples of this kind of predictability in psychology. But that doesn’t make it any less weird when AIs do the same thing.

Yes, some curious engineers over at Gramener performed an informal but nevertheless fascinating experiment where they simply asked several major LLM chatbots to pick random a number between 0 and 100.

Reader, the results were not random.

All three models tested had a “favorite” number that would always be their answer when put on the most deterministic mode, but which appeared most often even at higher “temperatures,” raising the variability of their results.

OpenAI’s GPT-3.5 Turbo really likes 47. Previously, it liked 42 — a number made famous, of course, by Douglas Adams in The Hitchhiker’s Guide to the Galaxy as the answer to the life, the universe, and everything.

Anthropic’s Claude 3 Haiku went with 42. And Gemini likes 72.

More interestingly, all three models demonstrated human-like bias in the numbers they selected, even at high temperature.

All tended to avoid low and high numbers; Claude never went above 87 or below 27, and even those were outliers. Double digits were scrupulously avoided: no 33s, 55s, or 66s, but 77 showed up (ends in 7). Almost no round numbers — though Gemini did once, at the highest temperature, went wild and picked 0.

Why should this be? AIs aren’t human! Why would they care what “seems” random? Have they finally achieved consciousness and this is how they show it?!

No. The answer, as is usually the case with these things, is that we are anthropomorphizing a step too far. These models don’t care about what is and isn’t random. They don’t know what “randomness” is! They answer this question the same way they answer all the rest: by looking at their training data and repeating what was most often written after a question that looked like “pick a random number.” The more often it appears, the more often the model repeats it.

Where in their training data would they see 100, if almost no one ever responds that way? For all the AI model knows, 100 is not an acceptable answer to that question. With no actual reasoning capability, and no understanding of numbers whatsoever, it can only answer like the stochastic parrot it is.

It’s an object lesson in LLM habits, and the humanity they can appear to show. In every interaction with these systems, one must bear in mind that they have been trained to act the way people do, even if that was not the intent. That’s why pseudanthropy is so difficult to avoid or prevent.

I wrote in the headline that these models “think they’re people,” but that’s a bit misleading. They don’t think at all. But in their responses, at all times, they are imitating people, without any need to know or think at all. Whether you’re asking it for a chickpea salad recipe, investment advice, or a random number, the process is the same. The results feel human because they are human, drawn directly from human-produced content and remixed — for your convenience, and of course big AI’s bottom line.

Heya, folks, welcome to Week in Review (WiR), TechCrunch’s regular newsletter that recaps the past few days in tech.

Google’s annual enterprise-focused dev conference, Google Cloud Next, dominated the headlines — and we had plenty of coverage from the event. But it wasn’t the only thing afoot (see: the spectacular eclipse).

Lorenzo wrote about how hackers stole over ~340,000 Social Security numbers from government consulting firm Greylock McKinnon Associates (GMA). It took GMA nine months to determine the extent of the breach and notify victims; as of yet, it’s unclear why.

Elsewhere, Sarah had the story on Spotify’s personalized AI playlists, which lets users create a playlist based on written prompts.

And Connie reported on the death of entrepreneur Mahbod Moghadam, who rose to fame as the co-founder of Genius, the online music encyclopedia. Moghadam passed away at the age of 41 owing to complications from a recurring brain tumor.

Lots else happened. We recap it all in this edition of WiR — but first, a reminder to sign up to receive the WiR newsletter in your inbox every Saturday.

News

Tesla price drop: Tesla dropped prices of unsold Model Y SUVs in the U.S. by thousands of dollars in an attempt to clear out an unprecedented inventory backlog.

Snapchat turns off its solar system: Snapchat adjusted a feature in its app that visualizes how “close” you are to your friends after reporting revealed that it was adding to teens’ anxiety.

Noninvasive anxiety treatment: Neurovalens, a startup developing tech to deliver noninvasive electrical stimulation of the brain and nervous system, achieved FDA clearance thanks to a 2019 agency rule change aimed at encouraging innovations targeting insomnia and anxiety.

Llama 3: At an event in London, Meta confirmed that it plans an initial release of Llama 3 — the next generation of its AI model used to power chatbots and other apps — within the month.

Emulators in the store: Apple updated its App Store rules to globally allow emulators for retro console games an option for downloading titles.

AT&T breach: AT&T began notifying U.S. state authorities and regulators of a security incident after confirming that millions of customer records posted online last month were authentic.

Funding

Web3 and beauty: Kiki World, a beauty brand that uses web3 for customer co-creation and ownership, has closed a $7 million round led by Andreessen Horowitz.

Analysis

Magnets in keyboards: Frederic writes about an intriguing development in mechanical keyboard design: magnetic switches, which can quickly change the actuation point — the point during the keypress where the switch registers a downstroke.

WFH, here to stay: Working from home isn’t going away — even if some CEOs wish it would. Ron writes that most workers crave flexibility and work-life balance — who knew?

Podcasts

On Equity’s startup-focused Wednesday show, the crew dug into the Multiverse’s acquisition of Searchlight, the latest Guesty round, the Monad Labs transaction and a new venture capital fund targeting growth rounds in Africa.

Meanwhile, Found featured Ben Christensen, the founder and CEO of Cambium, a startup that’s reimagining the wood supply chain and reallocating previously wasted materials to be used in new building projects.

Bonus round

Microsoft passwords exposed: Security researchers discovered an open and public database hosted on Microsoft’s Azure cloud service that was storing internal information relating to Microsoft’s Bing search engine. Microsoft says that it has resolved the lapse.

A flagship European Union digital market regulation appears to be shaking up competition in the mobile browser market.

It’s been a little over a month since the Digital Markets Act (DMA) came into application and there are early signs it’s having an impact by forcing phone makers to show browser choice screens to users.

On Wednesday, Reuters reported growth data shared by Cyprus-based web browser Aloha and others that it said suggests the new law is stirring the competitive pot and helping smaller browser makers gain share or at least grab more attention than they were.

But it’s early days for DMA implementation, with choice screen rollouts still a work in progress, and many EU users haven’t even seen one yet. While Aloha is not the only other browser reporting a boost in interest since the DMA compliance deadline kicked in on March 7 — Brave, Opera and Vivaldi also shared positive stories of increased interest — several others, including DuckDuckGo and Firefox, told us it’s too soon for them to be able to assess the regulation’s effect.

TechCrunch reached out to 16 alternative browser makers with questions, as well as Apple and Google, to inform our reporting. We also contacted the European Commission to ask about its own tracking of the DMA’s impact in this area — but it declined to share any data.

Neither Apple nor Google responded to questions asking about any changes in regional usage of their own browsers since the choice screens began being shown to mobile users.

Opting for choice screens

The EU’s goal for the DMA is to boost competition against internet “gatekeepers” whose control of dominant platforms gives them many operational advantages over smaller rivals. The regulation does this through a list of “dos and don’ts” that tech giants must comply with. In the case of browsers, it obliges the likes of iOS maker Apple and Google’s Android to display browser choice screens — forcing them to point users to alternatives to Apple’s Safari and Google’s Chrome.

Choice screens are intended to work against platform dominance and self-serving defaults by alerting consumers there are other options. But users do still need to decide to switch to an alternative app in order for choice screens to boost competition. The design of screens is also important.

Some alternative browser makers remain concerned the design of choice screens isn’t where it needs to be. We suspect this is leading to reluctance by some underdogs to share data on early impact, especially as the EU is currently investigating Apple’s choice screen design for suspected noncompliance.

In other words, some browser makers may be playing a waiting game in the hopes of encouraging Commission enforcers to push for a stronger implementation. At the same time, some really small browser players may see more gains to be had from good old-fashioned publicity — for example, sending out a press release trumpeting early interest — as a tactic to raise their profile to try to drive more downloads through increased awareness.

Overall, it’s still very early. Many regional mobile users may not have even seen a choice screen appear on their handset yet. Google, for instance, says screens are being displayed on newly launched Android devices but for existing Android handsets it’s up to the makers of the devices to push out the choice screens to their users. So there isn’t a clear implementation timeline on Android.

While in the case of iOS, Apple says it’s been displaying choice screens to users of iOS since iOS 17.4. But users who haven’t updated to this version also won’t have seen any yet.

Mozilla, maker of the Firefox browser, told us it estimates that less than a fifth of iOS users have been shown a choice screen so far. It reckons even fewer Android users have seen one in the wild as yet.

With this patchy Android rollout picture in mind, it seems likely that more iOS users will have seen choice screens than Android users so far — even though Google’s platform has a larger regional market share.

Measuring the impact of the DMA on alternative browsers’ market share is further complicated by variations in the apps that mobile users see in different EU countries. Some alternatives, such as Firefox, can appear on the iOS choice screen in every EU market. Whereas others are far more limited: Vivaldi, for example, can only appear in eight countries. So exposure to potential users can vary substantially depending on the browser. (Apple lists the options it’s currently showing in each market here.)

Alt browsers on the up?

Aloha, a browser that focuses on privacy and claims not to track users, told us it’s seen 250% growth in new users (i.e., app downloads) since the DMA came into effect last month. It reports having approximately 10 million active monthly users globally — and estimates that around 1 million of those are located in the EU. So it remains a very small player. 

However, since Aloha says it does not collect any personal data, including location data, it told us it cannot be precise about where its users are located. Yet it told Reuters the EU had moved up from being its fourth largest market to its second largest since the DMA compliance deadline kicked in. 

Aloha also claimed to have seen an uptick in users in the U.S. since the DMA came into effect — yet the regulation does not apply in the U.S. market so U.S. users aren’t encountering it via browser choice screens. Aloha told TechCrunch it believes privacy awareness is rising generally, but also suggested growth in new installs in the EU may be helping to raise its position in the U.S. App Store.

Norway-based Opera, meanwhile, is also claiming market share gains since the DMA started to bite on March 7. Per new metrics shared with TechCrunch Wednesday, Opera said new user growth from February to the end of March was 63% — so it’s reporting a substantial uptick in people downloading Opera and giving it a try.

It is also reporting a 39% growth in users on iOS selecting its browser as their default specifically, from March 3 until April 4.

Previously (as of March 18), Opera reported 164% growth in the inflow of new EU users on iOS after the deadline for Apple to implement the DMA-enforced choice screen. So there actually appears to have been a drop in the growth rate it’s seen over this period — that is, after a bigger initial spike of interest. 

Regardless, Opera is sounding very happy with the extra level of interest it’s seeing. In a statement, Jørgen Arnesen, its EVP of mobile, said the DMA “is working to even the playing field,” adding: “We’re excited to see that it has become easier for users to express their browser choice and for that choice to be respected.”

Another browser maker with a positive experience since DMA compliance day is Vivaldi, which is also developed out of Norway.

It told TechCrunch it’s seen an increase of 36.7% in downloads in the EU (in total) since the iOS choice screen came into effect. But the boost in downloads is even bigger when you look at the eight markets where Vivaldi is actually being shown on iOS choice screens. In those markets it said downloads have increased 69.6% since the choice screen started being pushed at users. 

Despite this uptick in downloads, Vivaldi is unhappy with the current design of Apple’s choice screen.

“There are significant flaws with its implementation, including when it is shown and what is shown,” a company spokesperson told us. “Users can only see the choice screen when they click Safari. The list of browsers does not show additional information and that does not help users to make a meaningful choice. If the user has already selected a browser of their own choice, the choice screen can actively try to push them away from it, and may not even include it in the list that it presents to the user.”

“We think the priority should be given to cross-platform browsers, so that the same browser can be used on all of the user’s devices,” she added. “Apple looks at it very narrowly, per platform and country. We believe the main browser choices should be visible and we are not. And we should be on the list for all countries.”

We also heard positive things from Brave. The U.S.-based privacy-focused browser said it’s seen “a significant uptick” in installs since the DMA came into effect. (Although it does not report users per region so declined to break out total usage figures for the EU.)

“The daily installs for Brave on iOS in the EU went from around 7,500 to 11,000 with the new browser panel this past March,” per a company spokesperson. “In the past few days, we have seen a new all time high spike of 14,000 daily installs, nearly doubling our pre-choice screen numbers.”

“Regarding retention, users who are choosing Brave from the DMA screen are being retained equally to or better than our average,” she added, arguing that, overall, the uptick in interest it’s seeing “confirms that users want choice.”

On the flip side, three other alternative browsers that we contacted — DuckDuckGo, Ecosia and Firefox — suggested it’s too early to tell whether the DMA is helping them.

Veteran privacy-focused browser maker DuckDuckGo declined to share any data, saying it’s too soon to draw meaningful conclusions.

“While we’ve seen some positive signs, the choice screen rollout is ongoing and for a competitor like us that sees billions of searches and millions of downloads a month, we need more time to make an accurate impact assessment at scale,” it said in a statement.  

DuckDuckGo also told us it lacks access to “key information” to be able to assess the DMA’s impact, saying, for example, that it has no way of knowing how many people have seen a search engine or browser choice screen.

“This is key because it would help us understand our selection rate on a choice screen and how widespread the rollout has been,” it noted, adding: “We’re at the beginning of this journey, not the end.”

Another alt player, the not-for-profit, tree-planting and eco-action focused Ecosia, also told us it doesn’t have enough data to make an accurate assessment of the regulation’s impact. “We have not received selection rates or any other meaningful datasets, so it is hard for us to solidly report on the effectiveness of the choice screen at this stage,” said Sophie Dembinski, its head of public policy and climate action. 

She emphasized Ecosia isn’t happy with the current iOS choice screen, which it believes is hampering potential growth — also pointing to the Commission’s open case investigating Apple’s implementation.

“While Ecosia has jumped to second and third position in some European markets for utility apps in the Apple App Store, our search numbers have barely changed,” she said. “This is due to several design issues within Apple’s choice screen — such as showing the choice screen to users who have already selected an alternative choice to Safari; an overly complex installation process which loses a large number of users; and keeping the Safari browser app in the best position on the home screen.”

Another veteran browser player, Firefox, is also keeping its powder dry when it comes to assessing early impact.

“We are not currently sharing absolute numbers, both because we have some serious concerns about the current choice screens and because we estimate that less than 20% of users on iOS and likely less on Google have been exposed to them thus far,” said Mozilla’s Kush Amlani, global competition and regulatory counsel. 

“The DMA represents a once-in-a-generation opportunity to create competition and choice for EU consumers. Whether that potential is realized depends on the gatekeepers’ compliance and the European Commission’s enforcement,” he emphasized, also referencing the Commission’s probes into suspected gatekeeper non-compliance.

“While we’re seeing many thousands of people select Firefox on the choice screens, we don’t think this should distract from the fact that the iOS choice screen has significant flaws that block people from making genuine choices,” Amlani added. “The critical challenge is that powerful and deep-pocketed gatekeepers are incentivized to protect their existing closed ecosystems and fight the implementation of the DMA, which will open them up to competition.”

TechCrunch’s outreach to browser makers that may benefit from the DMA choice screens also yielded one report of no meaningful impact since the requirement kicked in: Yandex, a Russia-based browser that can appear on the iOS choice screen anywhere in the EU, told us it hasn’t seen “any meaningful changes in the user metrics in the region so far.”

In Yandex’s case, its possible disinterest in switching could be linked to consumer concerns about using or supporting software that’s developed in Russia in light of the Ukraine war. 

Welcome back to TechCrunch Mobility — your central hub for news and insights on the future of transportation. Sign up here — just click TechCrunch Mobility — to receive the newsletter every weekend in your inbox. Subscribe for free.

Automakers reported auto sales for Q1 and, welp, turns out that pricing sure does matter if you want to sell EVs. Who would have thought? A recent survey by Edmunds comes to a similar conclusion (at least for American buyers), finding a big gap between what consumers want and what is actually available on the market.

Here’s the crux. According to the Edmunds survey, 47% say they are seeking an EV purchase below $40,000, and 22% are interested in EVs priced below the $30,000 threshold. Today, there are no new EVs priced below $30,000 and only four below the $40,000 mark. The average price of an EV in 2023 was $61,702, while all other vehicles stood at $47,450.

This mismatch of realities is squeezing automakers as they try to move inventory by slashing prices. This downward pressure has forced automakers like Ford to delay future EV launches and put more resources toward hybrids. Even Tesla, a bellwether in the EV world, fell well below analysts’ expectations with deliveries down 20% from Q4 2023. Meanwhile, EV upstart Rivian posted tepid results.

What’s the answer? Well, over at Tesla, it seems the solution is twofold: slash prices again and try to capture revenue through sales of its Full Self-Driving software that costs $12,000 and is currently being offered in a free one-month trial to all customers.

OK, folks, let’s jump into the rest of the news!

A little bird

Founders, investors, engineers, policy wonks and others tell us things. And we’re here to pass along the verifiable information that those little birds have shared with us.

This week, a little bird tipped us on the closure of Ghost Autonomy, which had raised upward of $220 million and recently partnered with OpenAI. A couple of calls, emails and a fresh posting on the company’s website confirmed the tip. About 100 people were affected.

As I noted in my article, Ghost has pivoted a few times since it was founded in 2017. When I asked founder and CEO John Hayes what happened, he said the company had completed a highway driving product and was moving in urban environments through what he described as “last-mile delivery.”

“Ultimately, the years required to bring the product to market could not be financed,” he wrote to me in an email.

Got a tip for us? Email Kirsten Korosec at kirsten.korosec@techcrunch.com or Sean O’Kane sean.okane@techcrunch.com. If you prefer to remain anonymous, click here to contact us, which includes SecureDrop (instructions here) and various encrypted messaging apps.

Deal of the week

Startup founders, listen up — a new fund just closed. Get your slide decks ready.

Maniv, the Israel and now NYC-based VC firm, raised a $140 million fund with plans to stick to its early-stage investment strategy of backing startups at the intersection between mobility, transportation and energy.

As I noted in my longer feature, the firm’s approach has evolved a bit by expanding geographically and diversifying its investor base. The firm has also largely stopped using the once trendy umbrella term “mobility” (often leaving it out of its original name Maniv Mobility) and has opted instead to talk about deep tech, decarbonization and digitization of the transportation sector.

Investors in the fund are no longer dominated by automakers and Tier 1 suppliers. Instead, Maniv has opened up to a broader swath of strategic and institutional financial investors, including BNP Paribas Personal Finance and the venture arms of Shell and Enterprise Mobility.

The Maniv III fund also includes return investors Valeo and Jaguar Land Rover venture arm InMotion Ventures. Toyota Motor Corp.’s Woven Capital, vehicle leasing company Arval, transportation infrastructure giant Ferrovial, the industrial manufacturing firm ITT Inc., fleet payments business WEX and an unnamed European insurance company also participated in the fund.

Other deals that got my attention …

Alsym Energy, a Massachusetts-based startup developing nonflammable battery chemistry, raised $78 million in a Series C round led by General Catalyst and Tata, the Indian conglomerate, with participation from Drads Capital, Thomvest and Thrive Capital.

BlaBlaCar, the French carpooling and bus ticketing company, secured a €100 million revolving credit facility ($108 million at today’s exchange rate).

Notable reads and other tidbits

Autonomous vehicles

Waymo and Uber expanded on an ongoing partnership that will affect Uber Eats’ customers in the metro Phoenix area. Now when folks order a burrito or a pizza or some other treat through Uber Eats, they may have their meals delivered by a Waymo vehicle. The tie-up will begin with select merchants in Chandler, Tempe and Mesa, including restaurants like Princess Pita, Filiberto’s and BoSa Donuts.

Electric vehicles, charging & batteries

Apple is laying off 614 employees in California after abandoning its electric car project. According to the WARN notice posted by the California EDD, most of the affected employees were working at buildings related to its canceled car project, while others were working at a facility for its next-generation screen development, Bloomberg reported.

Canoo finally reported its Q4 and full-year earnings. Tucked inside the regulatory filing is a nugget regarding the use of CEO Tony Aquila’s private jet — just one of many expenses that illustrates the gap between spending and revenue at the EV startup. Tl;dr: Canoo spent double its annual revenue on the CEO’s private jet in 2023.

Faraday Future narrowly avoided an eviction from its Los Angeles headquarters. The company reached an agreement with the owner of the building, Rexford Industrial, to stay at the facility as long as it meets a few conditions. If Faraday violates any of the terms, Rexford has the right to trigger a 48-hour demand for payment and can boot the startup if it doesn’t pay up. If Faraday Future makes its payments, it can stay in the building until September 2025 when the lease expires.

The National Highway Traffic Safety Administration opened a third investigation into Fisker’s Ocean SUV, this time centered on problems getting the doors to open.

Tesla is reportedly abandoning its plan to build a lower-cost EV thought to cost around $25,000, according to Reuters, despite that vehicle’s status as a pivotal product for the company’s overall growth. Apparently, Tesla will instead focus on a planned robotaxi that is being built on the same small EV platform that was also supposed to power the lower-cost vehicle. This is where it gets a bit silly. Just hours after Tesla CEO Elon Musk said Reuters was lying, he posted on X that the Tesla robotaxi would be revealed August 8. Go figure.

This week’s wheels

This week’s wheels is taking a one-week hiatus while I enjoy a bit of vacation time. But don’t worry, it’s back next week and I have a few vehicles lined up, including the Mercedes-Benz EQE 350 4Matic sedan, a Lexus LC500 hybrid and a Mercedes eSprinter. Plus, some e-bikes will soon be in the mix.

What vehicles — including the two-wheeled variety — are you interested in reading about? I’ll put them on my list.

Reuters reported on Thursday that Google’s parent company, Alphabet, is exploring the possibility of buying Boston-based HubSpot, a CRM and marketing automation company with a market cap of over $33 billion – a number that has been climbing on those reports.

If such a deal were to happen, the cost would likely be pretty substantial, involving some significant premium over the current value. It would have to be to motivate the company to sell and become part of the search giant. It’s worth noting that the two companies have a relationship already, a partnership to use Google ads to drive sales in HubSpot, which can sometimes be the start of an acquisition discussion like this.

While Google/Alphabet has been extremely acquisitive over the years, the largest deal that it’s ever made was spending $12.5 billion for Motorola Mobility in 2011. It later sold it to Lenovo for just $2.91 billion, so it would have reason to be gun shy on a much larger price tag. More recently the largest deal involved spending $5.4 billion for security intelligence platform Mandiant in 2022. Google usually stays under $3 billion, so a deal of this scope would be very much out of character for the company.

When you combine that with the austerity program that most tech companies have been on in recent years, and a warning from Google CEO Sundar Pichai in January that more job cuts were coming, it’s not the type of deal that seems likely in a belt tightening climate, and certainly one that might be tough to justify to employees if those kind of optics actually matter. Yet with a huge cash horde of $110 billion on hand as of the end of last year, it certainly has the cash to make the move if it wants to.

Another issue the company could face in trying to buy HubSpot is a hostile regulatory environment for large deals. The U.S., the U.K and the EU have been monitoring large deals closely these days. Some, like Adobe’s attempt to buy Figma for $20 billion didn’t make it to the finish line because of competitive concerns. It’s not clear that Alphabet would face those same concerns with a CRM tool. HubSpot faces pretty powerful competition from Adobe and Salesforce, two well-capitalized firms, so this wouldn’t give Google a lock on that market by any means, but if there’s a risk, there’s sure to be a termination fee involved to hedge against that, another factor the company would need to take into consideration.

The question is what is the likelihood of such a deal coming to fruition and what would it give the companies that they can’t get from the existing partnership. As one analyst said to me, it doesn’t feel likely, but you never know.

Beyoncé’s “Cowboy Carter” has only been out for a few days, yet it’s already obvious that we’ll be talking about it for years to come — it’s breaking records across streaming platforms, and the artist herself calls it “the best music [she’s] ever made.” But in the middle of the press release for “Cowboy Carter,” Beyoncé made an unexpected statement against the growing presence of AI in music.

“The joy of creating music is that there are no rules,” said Beyoncé. “The more I see the world evolving the more I felt a deeper connection to purity. With artificial intelligence and digital filters and programming, I wanted to go back to real instruments.”

Beyoncé rarely does interviews, giving each of her comments about the new album more significance — these remarks are among few jumping-off points fans get to help them puzzle through each element of the album, and how they all fit together. So her stance on AI isn’t just a throwaway comment made in conversation with a reporter. It’s deliberate.

The central backlash against AI-generated art comes from the way this technology works. AI-powered music generators can create new tracks in minutes and emulate artists’ vocals to a scarily convincing degree. In some cases, that’s because the AI is being trained on the work of the artists whose jobs it could end up replacing.

Large language models and diffusion models both require sprawling databases of text, images and sounds to be able to create AI-generated works. Some of the best-known AI companies, like Open AI and Stability AI, use datasets that include copyrighted artworks without consent. Even though Stability AI’s music model was trained on licensed stock music, that’s not the case for the company’s image generator, Stable Diffusion. Stability AI’s VP of Audio Ed Newton-Rex quit his job over this, because he “[doesn’t] agree with the company’s opinion that training generative AI models on copyrighted works is ‘fair use.’”

It’s no wonder artists like Beyoncé have strong feelings about this technology — too many AI models have been trained on artists’ work without their consent, and especially for rising musicians who don’t have the klout to buoy them, it will be even harder to break into an already ruthless industry. Beyoncé’s stance makes even more sense in the context of “Cowboy Carter” itself.

Though it does not explicitly discuss AI, “Cowboy Carter” already addresses the theft and appropriation of artworks without consent. On the album itself, Beyoncé is giving listeners a history lesson about how Black musicians formed the foundation of country music, which is too often assumed to represent Southern white culture.

Even the title “Cowboy Carter” nods to the appropriation of Black music for white people’s gain. Though “Carter” could reference Beyoncé’s married name, but it’s also a nod to the Carters, the “first family” of country music — and those Carters took the work of Black musicians to develop the style we now know as country, which continues to exclude Black artists (just recently, an Oklahoma country radio station recently refused a listener’s request to play Beyoncé’s “Texas Hold ‘Em,” since Beyoncé didn’t fit their definition of a country artist). Beyoncé’s seemingly random stance against AI unearths a similar truth: Once again, artists’ work is being stolen without their consent and contorted into something else, leaving them without payment or credit for their cultural contributions.

There are a few moments on the album when ninety-year-old country icon Willie Nelson appears a radio show called “Smoke Hour,” and its first appearance precedes “Texas Hold ‘Em.” The placement of the track takes on an extra layer of meaning in light of the Oklahoma radio incident, and Nelson makes a slight jab: “Now for this next tune, I want y’all to sit back, inhale, and go to the good place your mind likes to wander off to. And if you don’t wanna go, go find yourself a jukebox.”

This is Beyoncé’s world: the jukebox and the radio are back in style, Black musicians can make whatever kind of music they want, and no one’s art gets stolen.

Google is shutting down its Podcasts app in the U.S. in a matter of days. The company has begun warning the app’s users they will need to migrate their subscriptions to YouTube Music by April 2 to follow and stream their favorite shows going forward. Users who don’t make the move immediately will still have additional time to migrate, but will no longer be able to stream from the Podcasts app directly after this date.

The Google Podcasts app, installed over 500 million times on Android devices globally, for over half a decade has offered a simple and streamlined interface for discovering, following and listening to podcasts, as well as tools to add podcasts by RSS feed. Unfortunately for fans of the app, the tech giant said last September it would begin to wind down the Podcasts app in early 2024 as part of its broader plan to centralize its audio services under YouTube.

In 2020, YouTube Music offered a similar transition strategy to move music listeners away from Google Play Music ahead of its shutdown that same year. However, the Google Podcasts app continued to be maintained for years because YouTube Music wasn’t ready to support podcasts until more recently. By the end of 2023, YouTube Music was able to support podcasts globally, and, by February, they had the ability to upload their RSS feeds, too.

The move to shift podcasting over to YouTube could help Google become a bigger player in the space, not only by combining its efforts and sharpening its focus, but also because interest in video podcasts — which were already popular on YouTube — is on the rise. This week, for example, Spotify forged a deal with Universal Music Group (UMG) to bring video podcasts to U.S. users of its streaming app after earlier in March announcing tests of video podcasts in 11 other markets around the world.

Bleeping Computer was the first to notice the shutdown date for Google Podcasts in the U.S., and a support page on Google’s site confirms that users in the U.S. will only be able to use the Podcasts app through the end of March 2024. For those who miss the in-app pop-ups, Google will offer users additional time to save their subscriptions by allowing them to use the app’s export feature through July 2024.

Google did not immediately return a request for comment, but after publication did respond to say that while it was still “tracking towards” the April 2 timeline for the U.S., it has not shared a timeline for the rest of the world yet.

From its earlier statements, though, the plan is to discontinue Google Podcasts globally in 2024.

Updated, 3/29/24, 5:00 PM ET with Google comment.