A startup in Hangzhou just released a lightweight collar that listens to dogs and cats and turns their sounds into short sentences on your phone. Named Pettichat, the device weighs only 27 grams and sits comfortably around a pet’s neck. It picks up vocalizations through built-in microphones while motion sensors track posture, movements, and other physical cues at the same time.

Developers created the system using Alibaba Cloud’s Qwen model, giving it millions upon millions of speech samples from pets collected over several years in the aim of determining what those barks and meows are all attempting to communicate. They matched each of those vocalizations to patterns of conduct and the environment in which they occur. The app then sends out a basic message, such as “I’m hungry” or “I want to play.” Not only that, but early demos showed the collar operating in reverse: when owners speak into the app, it turns their words into noises that their pet can respond to.

Plaud Note Pro AI Voice Recorder, Transcribe & Summarize with AI Note Taker for Meetings & Calls…

• AI-POWERED TRANSCRIPTION & MULTI-DIMENSIONAL SUMMARIES: Plaud Note Pro is your professional voice transcriber, delivering high-accuracy transcription…
• ENHANCED CONTEXT WITH MULTIMODAL INPUT: Capture audio, type notes, add images, and press to highlight key moments for richer context. During…
• CHAT WITH YOUR RECORDINGS USING “ASK Plaud”: Unlock deeper insights with this interactive AI. Ask questions, extract key points, draft emails, and get…

Meng Xiaoyi, the company behind the collar, claims that their AI model can detect more than twenty different emotions with a staggering 94.6 percent accuracy, but it’s worth noting that this is just the company’s own testing, and no outside lab has come forward to confirm it, and certainly no peer-reviewed studies have been published yet. Meanwhile, animal behavior researchers point out that pets derive a great deal of their meaning from body language, context, and their surroundings, rather than just the sounds they make. Real-world sounds, other animals, or visitors can quickly throw this system out of balance.

In China, however, the response was immediate, with pre-orders for the collar opening in mid-May and then quickly taking off, with over ten thousand people picking them up in no time. It costs just about $120, or 799 yuan if you convert. Deliveries have already begun in China, with a wider rollout scheduled for May 30th. People from all over the world are placing orders after seeing recordings of a cat’s meow suddenly converting into “I wanna play” and a dog’s yelp reading “I’m hungry.”

There are already over 126 million cats and dogs living in Chinese cities, and the number is rapidly increasing. This collar serves as a meeting point for wearable technology, cloud computing, and the simple desire to get to know your furry friend a little better. But will it deliver correct translations, or is it just a nice parlor trick until more owners put it to the test in their own homes? For the time being, thousands of individuals are prepared to bet that the next time their dog barks, they will receive a clear response.
[Source]

Tesla announced on Thursday that its Full Self-Driving (Supervised) system is now available in China, listing the country among 10 markets where the technology can be accessed. The announcement on X was short on details and marks the first time Tesla has confirmed FSD availability in the world’s largest EV market. It comes a week after Elon Musk joined a US business delegation for President Trump’s summit with Chinese leader Xi Jinping in Beijing.

The timing raises questions the announcement does not answer. It is unclear whether mainstream Chinese consumers can already activate FSD or whether the post signals regulatory approval that has not yet been publicly confirmed. Tesla’s China website lists “intelligent assisted driving” for the Model 3 at a one-time fee of 64,000 yuan (approximately $9,400), with a Mandarin disclaimer noting that features would be updated “shortly.” China’s embassy did not respond to requests for comment on whether regulatory approval had been granted.

Despite its name, Tesla’s FSD (Supervised) still requires active driver supervision and is classified as a Level 2 system, meaning the driver must remain in control at all times. A fully autonomous, unsupervised version is being trialled only on a fleet of Tesla vehicles operating as part of the company’s robotaxi service in Austin, Texas. The gap between the marketing name and the technology’s actual capability has been a persistent source of regulatory and consumer confusion.

The 💜 of EU tech

The latest rumblings from the EU tech scene, a story from our wise ol’ founder Boris, and some questionable AI art. It’s free, every week, in your inbox. Sign up now!

The delay has been considerable. Musk first touted plans to bring FSD to China in 2024. In July of that year, he said he expected regulatory approval before the end of 2024. In September 2024, he cited “pending regulatory approval.” As recently as April 2026, Tesla’s CFO Vaibhav Taneja said in the Q1 earnings call that the company was still awaiting approval. Bloomberg reported on Wednesday that Tesla had launched a concerted hiring effort for autonomous driving roles in China, including autopilot test engineers, suggesting the regulatory path had finally cleared.

While Tesla waited, Chinese competitors moved. China’s Ministry of Industry and Information Technology issued its first Level 3 autonomous driving certifications in December 2025, approving passenger cars from Changan Auto and BAIC Motor. Level 3 allows hands-off driving under defined conditions, a capability Tesla’s FSD (Supervised) does not offer. Xiaomi delivered more than 410,000 vehicles in 2025 with its own advanced driver-assistance systems. Xpeng has been selling vehicles with highway and urban autonomous navigation in China since 2023. Huawei’s ADS 3.0 system, licensed to multiple Chinese automakers, operates without high-definition maps in more than 400 cities.

Chinese robotaxi companies are even further ahead. Pony.ai and Baidu’s Apollo Go are operating commercial, fully driverless ride-hailing services in multiple Chinese cities. Apollo Go caused a mass outage in Wuhan in April when more than a hundred robotaxis stopped mid-traffic, but the incident underscored the scale of deployment rather than the absence of it. Tesla’s robotaxi service, by comparison, is limited to a geofenced area of Austin.

Tesla’s competitive position in China has been under sustained pressure. In April, Tesla China sold the fourth-highest number of EVs in the country, behind BYD, Geely, and Chery, according to China Passenger Car Association data. Xiaomi launched a $34,300 YU7 Standard Edition this week that undercuts the Model Y by $4,350 with 50 kilometres more range. The FSD launch is positioned to restore a competitive advantage, but the technology Tesla is now offering in China is two levels below what Chinese regulators have already certified for domestic manufacturers.

The strategic question is whether FSD (Supervised), a Level 2 system that requires constant driver attention, is a meaningful differentiator in a market where competitors already offer Level 3 autonomy, proprietary mapping systems, and LiDAR-equipped vehicles at lower price points. Tesla’s camera-only approach, which Musk has argued is superior because it mirrors how humans drive, has not yet achieved the regulatory recognition in China that competitors using sensor fusion have obtained.

Chinese automakers are now entering Canada and expanding across Europe, the Middle East, and Southeast Asia with autonomous driving capabilities that were developed during the years Tesla spent waiting for Chinese regulators. The FSD launch, whenever it fully materialises for mainstream Chinese consumers, arrives in a market that has moved on. Tesla’s self-driving story in China is no longer about being first. It is about whether being late still matters.

Offbeat

Spanish shipbuilder’s 75-meter drone vessel comes with sensors, modular payloads, and no room for sailors

Shipbuilder Navantia has put forward a design for an uncrewed
warship intended to complement existing naval vessels in what has been dubbed a
“hybrid navy,” although it may not be an exact fit for any current requirements.

Developed by the UK arm of the Spanish firm, the Large
Autonomous Surface Vessel, or LASV75, is basically a large seagoing drone that
is armed like a conventional warship.

As its name suggests, the LASV75 is 75 meters long, making it
about half as long as one of the Royal Navy’s Type 45 destroyers and closer in size to one of the River-class patrol vessels, with a displacement of about 1,000
tonnes.

It is designed to be entirely uncrewed, with no bridge or
crew spaces, and adaptable for different missions via a modular design. Promotional
images also show the vessel carrying several shipping containers, which
have become a common way for navies to quickly add extra capabilities to a
vessel.

According to Navantia, the design allows for construction of
the drone at pace and scale, and it is expected to cost significantly less than
crewed warships, although the company did not specify exactly how quickly or cheaply it could be built.

The modularity extends to both mission payloads and
engineering systems, Navantia told us, allowing the LASV75 to be tailored to
the roles required, from installed power to weapons and sensor capability.

A prominent feature is a mast designed to host a number of
sensor configurations, but we were puzzled by the apparent lack of funnels for
engine exhaust. The ship is equipped with Integrated Full Electric Power and
Propulsion (IFEP), Navantia said, meaning it uses diesel generators to drive
electric motors and power everything else aboard. It has waterline
exhausts.

The LASV75 was designed to meet the Royal Navy’s concept of
a hybrid navy, but also to serve wider demand for autonomous vessels. Its size will
enable it to have the range and endurance for task group operations in the open
ocean, the company claims, so it can undertake escort duties or support the Royal
Navy’s Atlantic
Bastion strategy to protect undersea infrastructure around the UK, such as cables and pipelines, and to track Russian submarines.

The vessel could potentially meet the requirement for a Type 92 sloop that the Royal Navy outlined for Atlantic Bastion, effectively an uncrewed ship that can patrol the North Atlantic looking out for submarines.

This role is currently filled by the Type 23 frigate, and will be taken by the Type 26 when that comes into service, but the expectation
is that a flotilla of uncrewed Type 92 vessels would allow for greater uninterrupted coverage of the ocean.

“Autonomous vessels are fundamental to the future of
sovereign defence capabilities. Naval capabilities of the future will comprise
a hybrid mixture of crewed warships with uncrewed escorts and ancillary ships,”
said Derek Jones, Navantia UK chief commercial and business development officer.

“At Navantia UK, we’re investing heavily
in our four shipyards to turn them into ideal partners to
deliver this vision of the future.”

The company is currently building the Fleet Solid Support
(FSS) vessels for the Royal Fleet Auxiliary, which will provide supplies to
Royal Navy ships at sea.  ®

Two years ago, IBM realized there was one glaring omission in its roster of sports partnerships: Formula One. 

Formula One has become one of the world’s most popular sports, especially in the U.S., where Netflix’s “Drive to Survive” documented the working lives of F1 drivers and turned them into mainstream celebrities. The tech-centric sport has also become a hot ticket for tech companies like AWS, Oracle, and Anthropic, which partner with teams for sponsorship visibility and to provide data analytics and AI tools that can deliver a competitive edge.

So when IBM went looking for its next major sports partnership, it’s no wonder the company picked F1 and one of its most iconic teams, Scuderia Ferrari HP.

“They’re the winningest team in history,” Kameryn Stanhouse, IBM’s Vice President of Sports and Entertainment Partnerships, told TechCrunch.  

At the heart of this partnership, however, is what has led other teams to start working with tech giants: access to more sophisticated tech solutions that can help them make the most of, especially, artificial intelligence. In fact, one of the best parts of sports, Stanhouse said, is how much data is available and can be used to help people get comfortable with AI.

“They actually see how it serves them,” she said of how AI is used in sports storytelling.  

The IBM-Ferrari partnership centers on that idea of storytelling, enhancing fan engagement by overhauling the technology powering the Ferrari fan app. To help with this, Ferrari hired Stefano Pallard in the newly titled role “head of fan development,” who said the challenge the team wanted to tackle was not just reaching fans, but “making each of them feel like we know them.” 

“That starts with taking the data we get from the track and turning it into content that is easy to follow and engaging,” he told TechCrunch.  

Teams process millions of data points per second during each race, capturing every movement of the driver and the car. Turning this into content that fans can engage with is just one way that advanced enterprise AI can help businesses better interact with their consumers.

Among the 11 teams, Ferrari is one of the few (alongside the likes of McLaren and Williams) to have a standalone fan app strategy rather than lean on social media or the official F1 platforms instead, showing how the sport is slowly starting to capitalize on its growing global fandom.  

Some of the changes to the Ferrari app were simple, like offering it in Italian. Even though Ferrari is an Italian company and many of its fans are Italian, their fan app was not available in Italian until the IBM partnership.

Stanhouse said the old Ferrari fan app was a place where people went to find race details and then leave. This new app has games where fans can play with others in the app, new AI-written race summaries, more behind-the-scenes stories about the team and the drivers, a place to make predictions, and an AI companion for fans to ask questions.

“There are two drivers, but did you know it takes 24 people working simultaneously in two seconds to change a tire?” Stanhouse said, adding that storytelling helps fans feel closer to the team.  

Unlike other sports apps IBM has built, Stanhouse said the Ferrari app’s main focus is storytelling because it wants fans to stay engaged with it all year long, rather than for a few weeks a year, as with tournaments like the Masters. Engagement data for the app has been trending upward since IBM came into the scene, Stanhouse said, citing a 62% increase in engagement over race weekends as an example. 

Pallard said the team then uses AI to analyze engagement signals in the app, such as which content people like to read and the sentiment of the messages fans send.

“That helps us understand what resonates most with the Tifosi [the fan nickname for Ferrari] and it directly informs how we shape our storytelling and how we deliver content,” he said.

The team hopes to dive deeper into personalization and create more immersive fan experiences.  

The app developers also took into account Ferrari’s fanbase, which is much more diverse than it was even five years ago. F1 released stats last year showing that 75% of new fans were women, many of whom were Gen Z. A particular draw for women is the F1 Academy, an all-female racing series that aims to develop the next generation of women drivers. But these new fans, much like the old, are after one thing — more. 

“They are asking for more data, more insight, more features, and we have to be able to deliver that,” Pallard said. “With IBM, the vision for the next five years is to make every fan feel like the experience was built for them, whether they have been with us for 30 years or 30 days. That is how you build loyalty that lasts.”  

An international team of astrophysicists has found evidence that the universe recycles black holes, merging them to form even larger ones. Gravitational waves recorded in recent years show that some of the heaviest black holes within star clusters exhibit clear signs of being “second-generation” black holes—products of past collisions—and therefore could not have originated from the collapse of a massive star.

Impossible Black Holes

The evolutionary theory of stars explains that, at the end of the lives of the most massive stars, their cores compress until they form a point so dense that it curves space-time to infinity. This is the classic black hole, with masses 10 to 40 times that of the sun. There are also supermassive black holes, in the center of galaxies, with millions or billions of solar masses, whose origin is related to processes that occurred in the earliest moments of the universe.

Between these two extremes lies a contested category: black holes with masses between 40 and 100 solar masses. They are too heavy to be born after the death of a star, but they do not reach the necessary dimensions to emerge from the collapse of a gigantic cloud of matter. Conventional stellar physics considers them “impossible,” yet they appear frequently in detections.

A “normal” sized black hole, isolated in space.

Advertisement

Courtesy of Space Telescope Science Institute Office of Public Outreach

Astrophysicists propose that these massive black holes could form by the merging of two or more smaller, ultradense objects. The idea was plausible, but it needed evidence. Until relatively recently, there was no way to obtain it.

Then gravitational wave detectors came on the scene. These instruments use lasers to measure the micro-distortion of space-time generated by the collision of extremely dense objects. The first detection, in 2015, confirmed a merger between black holes. Since then, each new signal has allowed for a better characterization of these structures and revealed that these collisions occur much more frequently than previously imagined.

The Second-Generation Signature

The study, published this month in Nature Astronomy, analyzed a transient catalog of gravitational waves generated by the world’s three leading observatories. The database included 153 reliable detections of black hole mergers. Among them, 34 corresponded to particularly heavy objects.

By comparing all the signals, the team identified two distinct populations. The lighter black holes, up to about 40 solar masses, showed small, aligned spins, as expected for objects born from the collapse of a star. But from a certain point, around 45 solar masses, a completely different population appeared: heavier black holes, spinning rapidly and in chaotic directions—a statistical signature that can arise only when the object has already participated in a previous merger.

“This is the exact signature you would expect if black holes repeatedly merged into dense stellar clusters,” said Isobel M. Romero-Shaw, coauthor of the research, in a statement from Cardiff University.

So far researchers have not directly observed any of these “impossible” black holes. They do not appear in x-rays or in the visible spectrum, unlike supermassive ones. However, their collisions vibrate space-time, and that vibration reveals masses that stellar physics cannot explain.

This study shows that the heaviest black holes are built rather than born. They arise from previous generations of collisions, assembled in the densest environments in the cosmos.

This story originally appeared in WIRED en Español and has been translated from Spanish.

Personal Tech

Switch to premium devices pays off as PC giant post record record, just don’t ask about cheap laptops 

PC buyers may be wincing at memory price hikes, but Lenovo isn’t. The China-based tech biz says it sidestepped much of the industry pain by switching to premium devices and the numbers back it up. 

For Q4 of its fiscal 2026 ended March 31, Lenovo’s Intelligent Devices Group posted revenue of $14.6 billion, up from $11.9 billion a year earlier. It reported operating profit – net profit was not disclosed – of just over $1 billion, up 20.7 percent. PC and smart devices revenues, specifically, grew 26 percent. 

“Last quarter, despite the supply shortages and rising
component costs, we committed to sustaining growth and improving profitability,
leveraging our operational excellence,” CEO Yang Yuanqing said on an earnings call. 

“We promised to maintain our PC revenue momentum despite a
slowdown in PC shipments due to rising costs. We delivered. We shifted our mix
towards premium to improve average unit revenue, and our PC shipment growth
continued to outperform the market,” he stated.

PCs accounted for half of Lenovo’s overall group turnover, shipments were up 20 percent year-on-year and the corporation accounted for 24.4 percent global market share. Servers and services comprised the rest of Lenovo’s revenues.

The memory crunch has been brutal. Some DRAM and NAND flash prices doubled or quadrupled by early this year, as chipmakers chased higher margins on AI server memory and starved the consumer market of supply. 

The Register has previously reported how the price hikes led to a spike
in PC sales, as corporate buyers brought forward purchases before memory costs
climbed any further.

Asked whether this had any effect on Lenovo’s numbers, EVP
for Intelligent Devices Luca Rossi downplayed it. “So in calendar Q1, our last fiscal Q4, we definitely
observed strong demand, which might partially be linked to some pull in, but I
don’t think that it will be a substantial number,” he stated.

“Definitely, we are seeing some tight supply in certain
components, particularly – as you probably know – in the semiconductor area.
However, we feel confident about our ability to procure the parts we need and
we did not adjust our full year target based on supply constraints. Rather, we
will align the shipment target based on the real market and demand in order to
maintain a healthy channel inventory and with the goal of maintaining a solid
premium to market,” Rossi said.

Lenovo expects unit shipments
to decline year-on-year for its fiscal 2027. “But at the same time,
we expect to maintain or very likely grow our revenue linked to the significant
growth of the AUR (average unit revenue),” he added. 

Squeezing more profit from fewer system sales means availability of cheaper PCs will take a hit as Lenovo shifts production to premium boxes.

This isn’t the only impact AI is having on the PC market. CEO Yang pledged to embed the technology across Lenovo’s entire product line, including forthcoming “personal AI super agents” Tianxi and QIRA, plus next-generation AI-native PCs, smartphones, wearables, and “personal computing hubs.” Whether customers want all of that remains, as ever, an open question.

Lenovo AI Now or Tianxi is a personal and private AI
assistant to help with writing, summarizing, and quick settings
for your computer, says Lenovo. QIRA is “your personal
intelligence that’s by your
side across Lenovo and Motorola devices. It moves with you, learns from you,
and helps you get things done.”

For those interested in the total financial figures, Lenovo claimed a fourth quarter revenue record of $21.6 billion, up 27 percent
year-on-year. It recorded  revenue of $83.1 billion and net profit of $1.91 billion for the whole of its fiscal ’26. ®

Nature abhors a vacuum, so the saying goes, but nobody told the universe. Space is filled with cosmic voids—vast regions mostly free of matter that have opened between dense threads of material that make up a cosmic web.

Far from being vacant backwaters with little to study, these voids may hold solutions to some of the most persistent cosmic mysteries, such as the behavior of gravity, the nature of dark energy, and the so-called Hubble tension, an observational mismatch in the expansion rate of the universe that has caused astronomers’ headaches for years.

“With voids, we have the power to tackle most of the interesting cosmological riddles,” says Alice Pisani, a research professor in cosmology working at the Centre for Particle Physics in Marseille (CPPM) of the French National Centre for Scientific Research. She adds that because there’s less interference from matter, there’s a “high signal-to-noise” ratio in terms of what researchers can observe.

The advent of new telescopes and advanced simulations has supercharged this field, inspiring a growing community of scientists worldwide to specialize in voids as unique cosmological laboratories. Some experts argue we may even live inside a colossal void, a position that may alter our view of the universe in consequential ways.

For places defined by sparseness, voids are becoming cosmological heavyweights, where the laws of physics can be observed with unusual clarity.

“From a cosmology perspective, it is a very exciting time,” Pisani says.

What Are Cosmic Voids?

Following the Big Bang, the universe was a uniform soup of subatomic particles. But over millions of years, as matter cooled and stabilized into atoms, the faint outlines of the cosmic web began to emerge.

Over billions of years, the web gravitationally pulled gas clouds, galaxy clusters, and other cosmic objects toward its scaffolding. As more matter is drawn into the web, gaps have widened between its filaments, forming voids.

Small “subvoids” can open between galaxy clusters, where they might be only 10 or 20 million light years across. But voids can get bigger. Much bigger. The Boötes Void, also known as the “Great Nothing,” stretches across more than 300 million light years.

Calling them cosmic voids can be “misleading,” Pisani says, “because we end up thinking that a void means empty. But as a matter of fact, the voids that we look at are never empty. There are very tiny low-mass galaxies inside those under-dense regions.” The Boötes Void, for example, contains a few dozen galaxies— though that’s still far less than the thousands that would be expected in a similarly sized area.

Because they are comparatively bereft of material, cosmic voids remained out of observational view until the late 1970s. Until that point, the positions of galaxies had been mapped as 2D points on the sky, but the development of 3D maps of galaxy distribution revealed the contours of the cosmic web for the first time, exposing the presence of voids.

In recent years, a host of new telescope surveys have kicked off an explosion of new void discoveries, such as the Dark Energy Survey Instrument (DESI) in Arizona, and the European Euclid space telescope. These instruments are expected to map more than 100,000 voids in space, offering an unprecedented glimpse of these structures. Yet these surveys will still only capture a fraction of the many millions of voids that are estimated to exist in the observable universe.

“Just in the last 10 years, the field really evolved significantly with new technologies,” says Nico Schuster, a cosmologist and cosmic void expert at CPPM. “All of that really enables us to observe plenty more galaxies than we could before, and that really allows us to probe, finally, the cosmic web at a much deeper depth, and find more voids and resolve them better.”