Amazon Web Services has signed a long-term deal with Talen Energy to receive up to 1,920 megawatts of carbon-free electricity from the Susquehanna nuclear plant through 2042 to support AWS's AI and cloud operations. The partnership also includes plans to explore new Small Modular Reactors and expand nuclear capacity amid rising U.S. energy demand. Utility Drive reports: Under the PPA, Talen's existing 300-MW co-location arrangement with AWS will shift to a "front of the meter" framework that doesn't require Federal Energy Regulatory Commission approval, according to Houston-based Talen. The company expects the transition will occur next spring after transmission upgrades are finished. FERC in November rejected an amended interconnection service agreement that would have facilitated expanded power sales to a co-located AWS data center at the Susquehanna plant. The agency is considering potential rules for co-located loads in PJM.

Talen expects to earn about $18 billion in revenue over the life of the contract at its full quantity, according to an investor presentation. The contract, which runs through 2042, calls for delivering 840 MW to 1,200 MW in 2029 and 1,680 MW to 1,920 MW in 2032. Talen will act as the retail power supplier to AWS, and PPL Electric Utilities will be responsible for transmission and delivery, the company said. Amazon on Monday said it plans to spend about $20 billion building data centers in Pennsylvania.

"We are making the largest private sector investment in state history -- $20 billion-- to bring 1,250 high-skilled jobs and economic benefits to the state, while also collaborating with Talen Energy to help power our infrastructure with carbon-free energy," Kevin Miller, AWS vice president of global data centers, said.

An anonymous reader quotes a report from Ars Technica: The robot built by [Yuntao Ma and his team at ETH Zurich] was called ANYmal and resembled a miniature giraffe that plays badminton by holding a racket in its teeth. It was a quadruped platform developed by ANYbotics, an ETH Zurich spinoff company that mainly builds robots for the oil and gas industries. "It was an industry-grade robot," Ma said. The robot had elastic actuators in its legs, weighed roughly 50 kilograms, and was half a meter wide and under a meter long. On top of the robot, Ma's team fitted an arm with several degrees of freedom produced by another ETH Zurich spinoff called Duatic. This is what would hold and swing a badminton racket. Shuttlecock tracking and sensing the environment were done with a stereoscopic camera. "We've been working to integrate the hardware for five years," Ma said.

Along with the hardware, his team was also working on the robot's brain. State-of-the-art robots usually use model-based control optimization, a time-consuming, sophisticated approach that relies on a mathematical model of the robot's dynamics and environment. "In recent years, though, the approach based on reinforcement learning algorithms became more popular," Ma told Ars. "Instead of building advanced models, we simulated the robot in a simulated world and let it learn to move on its own." In ANYmal's case, this simulated world was a badminton court where its digital alter ego was chasing after shuttlecocks with a racket. The training was divided into repeatable units, each of which required that the robot predict the shuttlecock's trajectory and hit it with a racket six times in a row. During this training, like a true sportsman, the robot also got to know its physical limits and to work around them.

The idea behind training the control algorithms was to develop visuo-motor skills similar to human badminton players. The robot was supposed to move around the court, anticipating where the shuttlecock might go next and position its whole body, using all available degrees of freedom, for a swing that would mean a good return. This is why balancing perception and movement played such an important role. The training procedure included a perception model based on real camera data, which taught the robot to keep the shuttlecock in its field of view while accounting for the noise and resulting object-tracking errors.

Once the training was done, the robot learned to position itself on the court. It figured out that the best strategy after a successful return is to move back to the center and toward the backline, which is something human players do. It even came with a trick where it stood on its hind legs to see the incoming shuttlecock better. It also learned fall avoidance and determined how much risk was reasonable to take given its limited speed. The robot did not attempt impossible plays that would create the potential for serious damage -- it was committed, but not suicidal. But when it finally played humans, it turned out ANYmal, as a badminton player, was amateur at best. The findings have been published in the journal Science Robotics.

You can watch a video of the four-legged robot playing badminton on YouTube.

An anonymous reader quotes a report from Reuters: Indirect carbon emissions from the operations of four of the leading AI-focused tech companies rose on average by 150% from 2020-2023, due to the demands of power-hungry data centers, a United Nations report (PDF) said on Thursday. The use of artificial intelligence by Amazon, Microsoft, Alphabet and Meta drove up their global indirect emissions because of the vast amounts of energy required to power data centers, the report by the International Telecommunication Union (ITU), the U.N. agency for digital technologies, said.

Indirect emissions include those generated by purchased electricity, steam, heating and cooling consumed by a company. Amazon's operational carbon emissions grew the most at 182% in 2023 compared to three years before, followed by Microsoft at 155%, Meta at 145% and Alphabet at 138%, according to the report. The ITU tracked the greenhouse gas emissions of 200 leading digital companies between 2020 and 2023. [...] As investment in AI increases, carbon emissions from the top-emitting AI systems are predicted to reach up to 102.6 million tons of carbon dioxide equivalent per year, the report stated.

The data centres that are needed for AI development could also put pressure on existing energy infrastructure. "The rapid growth of artificial intelligence is driving a sharp rise in global electricity demand, with electricity use by data centers increasing four times faster than the overall rise in electricity consumption," the report found. It also highlighted that although a growing number of digital companies had set emissions targets, those ambitions had not yet fully translated into actual reductions of emissions. UPDATE: The headline has been revised to clarify that four leading AI-focused tech companies saw their operational emissions rise to 150% of their 2020 levels by 2023 -- a 50% increase, not a 150% one.

An anonymous reader quotes a report from Tom's Hardware: The head of the Federal Aviation Administration just outlined an ambitious goal to upgrade the U.S.'s air traffic control (ATC) system and bring it into the 21st century. According to NPR, most ATC towers and other facilities today feel like they're stuck in the 20th century, with controllers using paper strips and floppy disks to transfer data, while their computers run Windows 95. While this likely saved them from the disastrous CrowdStrike outage that had a massive global impact, their age is a major risk to the nation's critical infrastructure, with the FAA itself saying that the current state of its hardware is unsustainable.

"The whole idea is to replace the system. No more floppy disks or paper strips," acting FAA administrator Chris Rocheleau told the House Appropriations Committee last Wednesday. Transportation Secretary Sean Duffy also said earlier this week," This is the most important infrastructure project that we've had in this country for decades. Everyone agrees -- this is non-partisan. Everyone knows we have to do it." The aviation industry put up a coalition pushing for ATC modernization called Modern Skies, and it even ran an ad telling us that ATC is still using floppy disks and several older technologies to keep our skies safe. [...]

Currently, the White House hasn't said what this update will cost. The FAA has already put out a Request For Information to gather data from companies willing to take on the challenge of upgrading the entire system. It also announced several 'Industry Days' so companies can pitch their tech and ideas to the Transportation Department. Duffy said that the Transportation Department aims to complete the project within four years. However, industry experts say this timeline is unrealistic. No matter how long it takes, it's high time that the FAA upgrades the U.S.'s ATC system today after decades of neglect.

Apple announced that macOS 26 "Tahoe" will be the final version to support Intel-based Macs, with future macOS releases running exclusively on Apple Silicon devices (that is, 2020 M1 models and newer). They will, however, continue to receive security updates for a few more years. 9to5Mac reports: In some ways, Apple has already stopped supporting some non-Apple Silicon models of its lineup. macOS Tahoe does not work with any Intel MacBook Air or Mac mini for instance. But Tahoe does still support some Intel Macs. That includes compatibility with the 2019 16-inch MacBook Pro, the 2020 Intel 13-inch MacBook Pro, 2020 iMac, and the 2019 Mac Pro.

Based on Apple's warning, you can expect that macOS 27 will drop support for all of these legacy machines, and therefore macOS 26 will be the last compatible version. These devices will continue to receive security updates for another three years, however. Going forward, the minimum support hardware generations will be from 2020 onwards, as that is when Apple began the Apple Silicon transition with the M1. M1 Pro and M1 Max MacBook Pros followed in 2021.

An anonymous reader shares a report: Seagate has announced a new 4TB version of its storage expansion card for the Xbox Series X and S consoles. It's the first time the company has introduced a new capacity since launching 2TB and 512GB versions of the expansion card in late 2021.

The 4TB card is available starting today through Seagate's online store and Best Buy for $499.99, but is discounted to $429.99 as part of a limited-time launch promotion. For comparison, the Xbox Series S starts at $379.99, while the Xbox Series X starts at $599.99.

The U.K. electricity grid "was built to deliver power generated by coal and gas plants near the country's major cities and towns," reports the BBC, "and doesn't always have sufficient capacity in the wires that carry electricity around the country to get the new renewable electricity generated way out in the wild seas and rural areas.

"And this has major consequences." The way the system currently works means a company like Ocean Winds gets what are effectively compensation payments if the system can't take the power its wind turbines are generating and it has to turn down its output. It means Ocean winds was paid £72,000 [nearly $100,000 USD] not to generate power from its wind farms in the Moray Firth during a half-hour period on 3 June because the system was overloaded — one of a number of occasions output was restricted that day. At the same time, 44 miles (70km) east of London, the Grain gas-fired power station on the Thames Estuary was paid £43,000 to provide more electricity.

Payments like that happen virtually every day. Seagreen, Scotland's largest wind farm, was paid £65 million last year to restrict its output 71% of the time, according to analysis by Octopus Energy. Balancing the grid in this way has already cost the country more than £500 million this year alone, the company's analysis shows. The total could reach almost £8bn a year by 2030, warns the National Electricity System Operator (NESO), the body in charge of the electricity network. It's pushing up all our energy bills and calling into question the government's promise that net zero would end up delivering cheaper electricity... the potential for renewables to deliver lower costs just isn't coming through to consumers.

Renewables now generate more than half the country's electricity, but because of the limits to how much electricity can be moved around the system, even on windy days some gas generation is almost always needed to top the system up. And because gas tends to be more expensive, it sets the wholesale price.
The UK government is now considering smaller regional markets, so wind companies "would have to sell that spare power to local people instead of into a national market. The theory is prices would fall dramatically — on some days Scottish customers might even get their electricity for free...

"Supporters argue that it would attract energy-intensive businesses such as data centres, chemical companies and other manufacturing industries."

Britain is moving toward major nuclear power commitments after years of delays, as government officials acknowledge they can no longer postpone critical energy infrastructure decisions. The U.K. Treasury has exhausted options for delaying nuclear power choices, Politico reported this week, citing sources within Whitehall and the nuclear industry.

The urgency stems from Britain's aging nuclear infrastructure, where five power plants currently supply 15% of the country's total energy needs but face shutdown by 2030. This timeline has created significant pressure on policymakers to secure replacement capacity or risk substantial gaps in the nation's electricity supply.

Startup Nord Quantique has demonstrated that a single piece of hardware can host an error-detecting logical qubit by using two quantum frequencies within one resonator. The breakthrough has the potential to slash the hardware demands for quantum error correction and deliver more compact and efficient quantum computing architectures. Ars Technica reports: The company did two experiments with this new hardware. First, it ran multiple rounds of error detection on data stored in the logical qubit, essentially testing its ability to act like a quantum memory and retain the information stored there. Without correcting errors, the system rapidly decayed, with an error probability in each round of measurement of about 12 percent. By the time the system reached the 25th measurement, almost every instance had already encountered an error. The second time through, the company repeated the process, discarding any instances in which an error occurred. In almost every instance, that meant the results were discarded long before they got through two dozen rounds of measurement. But at these later stages, none of the remaining instances were in an erroneous state. That indicates that a successful correction of the errors -- something the team didn't try -- would be able to fix all the detected problems.

Several other companies have already performed experiments in which errors were detected -- and corrected. In a few instances, companies have even performed operations with logical qubits, although these were not sophisticated calculations. Nord Quantique, in contrast, is only showing the operation of a single logical qubit, so it's not even possible to test a two-qubit gate operation using the hardware it has described so far. So simply being able to identify the occurrence of errors is not on the cutting edge. Why is this notable?

All the other companies require multiple hardware qubits to host a single logical qubit. Since building many hardware qubits has been an ongoing challenge, most researchers have plans to minimize the number of hardware qubits needed to support a logical qubit -- some combination of high-quality hardware, a clever error correction scheme, and/or a hardware-specific feature that catches the most common errors. You can view Nord Quantique's approach as being at the extreme end of the spectrum of solutions, where the number of hardware qubits required is simply one. From Nord Quantique's perspective, that's significant because it means that its hardware will ultimately occupy less space and have lower power and cooling requirements than some of its competitors. (Other hardware, like neutral atoms, requires lots of lasers and a high vacuum, so the needs are difficult to compare.) But it also means that, should it become technically difficult to get large numbers of qubits to operate as a coherent whole, Nord Quantique's approach may ultimately help us overcome some of these limits.

Apple's Worldwide Developers Conference 25 (WWDC) kicks off next week, June 9th, showcasing the company's latest software and new technologies. That includes the next version of iOS, which is rumored to have the most significant design overhaul since the introduction of iOS 7. Here's an overview of what to expect: Major Software Redesigns
Apple plans to shift its operating system naming to reflect the release year, moving from sequential numbers to year-based identifiers. Consequently, the upcoming releases will be labeled as iOS 26, macOS 26, watchOS 26, etc., streamlining the versioning across platforms.

iOS 26 is anticipated to feature a glossy, glass-like interface inspired by visionOS, incorporating translucent elements and rounded buttons. This design language is expected to extend across iPadOS, macOS, watchOS, and tvOS, promoting a cohesive user experience across devices. Core applications like Phone, Safari, and Camera are slated for significant redesigns, too. For instance, Safari may introduce a translucent, "glassy" address bar, aligning with the new visual aesthetics.

While AI is not expected to be the main focus due to Siri's current readiness, some AI-related updates are rumored. The Shortcuts app may gain "Apple Intelligence," enabling users to create shortcuts using natural language. It's also possible that Gemini will be offered as an option for AI functionalities on the iPhone, similar to ChatGPT.

Other App and Feature Updates
The lock screen might display charging estimates, indicating how long it will take for the phone to fully charge. There's a rumor about bringing live translation features to AirPods. The Messages app could receive automatic translations and call support; the Music app might introduce full-screen animated lock screen art; and Apple Notes may get markdown support. Users may also only need to log into a captive Wi-Fi portal once, and all their devices will automatically be logged in.

Significant updates are expected for Apple Home. There's speculation about the potential announcement of a "HomePad" with a screen, Apple's competitor to devices like the Nest Hub Mini. A new dedicated Apple gaming app is also anticipated to replace Game Center. If you're expecting new hardware, don't hold your breath. The event is expected to focus primarily on software developments. It may even see discontinued support for several older Intel-based Macs in macOS 26, including models like the 2018 MacBook Pro and the 2019 iMac, as Apple continues its transition towards exclusive support for Apple Silicon devices.

Sources:
Apple WWDC 2025 Rumors and Predictions! (Waveform)
WWDC 2025 Overview (MacRumors)
WWDC 2025: What to expect from this year's conference (TechCrunch)
What to expect from Apple's Worldwide Developers Conference next week (Ars Technica)
Apple's WWDC 2025: How to Watch and What to Expect (Wired)

Intel has implemented a strict new policy requiring all new projects to demonstrate at least a 50% gross margin to move forward. CEO Lip-Bu Tan explained Intel's new risk-averse policy as "something that we probably should have had before," later clarifying that the number is a figure the company is aspiring toward internally. Tom's Hardware reports: Tan is reportedly "laser focused on the fact that we need to get our gross margins back up above 50%." To accomplish this, Tan is also said to be investigating and potentially cancelling or changing unprofitable deals with other companies. Intel's margins have slipped to new lows for the company in recent months. MacroTrends reports Intel's trailing 12 months gross margin for Q1 2025 was as low as 31.67%. Intel's gross margins had hovered around the 60% mark for the ten years leading up to the COVID-19 pandemic, falling beneath 50% in Q2 2022 and continuing to steadily fall ever since.

Holthaus predicts a "tug-of-war" to ensue within Intel in the coming months as engineers and executives reckon with being forced between a rock and a hard place. "We need to be building products that... fit the right competitive landscape and requirements of our customers, but also have the right cost structure in place. It really requires us to do both." [...] Tan is also quoted as wanting to turn Intel into an "engineering-focused company" again under his leadership. To reach this, Tan has committed to investing in recruiting and retaining top talent; "I believe Intel has lost some of this talent over the years; I want to create a culture of innovation empowerment." Maintaining a culture of empowering innovation and top talent seems, on its face, at odds with layoffs and a lock on projects not projected to gross 50% margins, but Tan seemingly has Intel investors on his side in these pursuits.

Amazon is developing software for humanoid robots that could eventually replace hundreds of thousands of delivery workers, [non-paywalled source] The Information reports. The company is building a "humanoid park" obstacle course at its San Francisco office to test robots that would ride in the back of Amazon's Rivian electric vans and deliver packages to customers, the report said. The indoor testing facility, roughly the size of a coffee shop, will house a Rivian van and serve as a controlled environment before Amazon takes the robots on "field trips" to deliver real packages on actual streets.

This summer, Amazon plans to test multiple humanoid models, including a $16,000 unit from China-based Unitree that has gained popularity among robotics developers, the report said. The initiative represents Amazon's most ambitious robotics project yet, extending beyond its existing warehouse automation to tackle the significantly more complex challenge of outdoor package delivery. Amazon currently operates more than 20,000 Rivian vehicles for deliveries and plans to expand its electric fleet to 100,000 vehicles by 2030.

An anonymous reader quotes a report from Ars Technica: On Tuesday, classic computer collector Joe Strosnider announced the availability of a new 3D-printer filament that replicates the iconic "Platinum" color scheme used in classic Macintosh computers from the late 1980s through the 1990s. The PLA filament (PLA is short for polylactic acid) allows hobbyists to 3D-print nostalgic novelties, replacement parts, and accessories that match the original color of vintage Apple computers. Hobbyists commonly feed this type of filament into commercial desktop 3D printers, which heat the plastic and extrude it in a computer-controlled way to fabricate new plastic parts.

The Platinum color, which Apple used in its desktop and portable computer lines starting with the Apple IIgs in 1986, has become synonymous with a distinctive era of classic Macintosh aesthetic. Over time, original Macintosh plastics have become brittle and discolored with age, so matching the "original" color can be a somewhat challenging and subjective experience. Strosnider said he paid approximately $900 to develop the color. "Rather than keeping the formulation proprietary, he arranged for Polar Filament to make the color publicly available [for $21.99 per kilogram]," adds Ars.

Ars Technica's Benj Edwards reports: In 1979, Atari released the Atari 400 and 800, groundbreaking home computers that included custom graphics and sound chips, four joystick ports, and the ability to run the most advanced home video games of their era. These machines, which retailed for $549 and $999, respectively, represented a leap in consumer-friendly personal computing, with their modular design and serial I/O bus that presaged USB. Now, 46 years later, a hobbyist has shrunk down the system hardware to a size that would have seemed like science fiction in the 1970s.

Polish engineer Piotr "Osa" Ostapowicz recently unveiled "Atarino," which may be the world's smallest 8-bit Atari computer re-creation, according to retro computing site Atariteca. The entire system -- processor, graphics chips, sound hardware, and memory controllers -- fits on a module measuring just 2x1.5 centimeters (about 0.79x0.59 inches), which is roughly the size of a postage stamp.

Ostapowicz's creation reimplements the classic Atari XL/XE architecture using modern FPGA (field-programmable gate array) technology. Unlike software emulators that simulate old hardware (and modern recreations that run them, like the Atari 400 Mini console) on a complete computer system of another architecture, Atarino reproduces the original Atari components faithfully at the logic level, allowing it to run vintage software while maintaining compatibility with original peripherals. [...] The project, which began over a decade ago and was first publicly demonstrated in December 2023, includes a 6502C processor, ANTIC and GTIA graphics chips, POKEY sound chip, and memory controllers onto a single Lattice UP5K FPGA chip. Despite its tiny size, the system can run at clock speeds up to 31 MHz -- far faster than the original hardware's 1.79 MHz. While the Atarino can run vintage software and work with the original peripherals, it brings several key improvements -- including a modernized 6502 core with added instructions, a more efficient memory architecture, enhanced video output via VGA and HDMI, extended graphics modes, refined sound chip emulation, modular hardware design, support for modern connectivity like Wi-Fi and Ethernet, and compatibility with contemporary development tools like CC65 and Visual Studio Code.

Ostapowicz "plans to release complete kits with documentation, inviting the retrocomputing community to experiment with the hardware," adds Edwards.

Meta has struck a 20-year deal with energy company Constellation to keep the Clinton Clean Energy Center nuclear plant in Illinois operational, the social media giant's first nuclear power purchase agreement as it seeks clean energy sources for AI data centers. The aging facility, which was slated to close in 2017 after years of financial losses and currently operates under a state tax credit reprieve until 2027, will receive undisclosed financial support that enables a 30-megawatt capacity expansion to 1,121 MW total output.

The arrangement preserves 1,100 local jobs while generating electricity for 800,000 homes, as Meta purchases clean energy certificates to offset a portion of its growing carbon footprint driven by AI operations.

Texas is poised to become the first state with a Republican-controlled government to pass a right to repair law, as its Senate unanimously approved HB 2963. The bill requires manufacturers to provide parts, manuals, and tools for equipment sold or used in the state. The Verge reports: A press release from the United States Public Interest Research Group (PIRG), which has pushed for repairability laws nationwide, noted that this would make Texas the ninth state with a right to repair rule, and the seventh with a version that includes consumer electronics. It follows New York, Colorado, Minnesota, California, Oregon, Maine, and most recently, Washington [...]. "More repair means less waste. Texas produces some 621,000 tons of electronic waste per year, which creates an expensive and toxic mess. Now, thanks to this bipartisan win, Texans can fix that," said Environment Texas executive director Luke Metzger.

Laurene Powell Jobs has publicly endorsed the secretive AI hardware device being developed by Jony Ive and OpenAI, expressing admiration for his design process and investing in his ventures. Ive says the project is an attempt to address the unintended harms of past tech like the iPhone, and Powell Jobs stands to benefit financially if the device succeeds. The Verge reports: In a new interview published by The Financial Times, the two reminisce about Jony Ive's time working at Apple alongside Powell Jobs' late husband, Steve, and trying to make up for the "unintentional" harms associated with those efforts. [...] Powell Jobs, who has remained close friends with Ive since Steve Jobs passed in 2011, echoes his concerns, saying that "there are dark uses for certain types of technology," even if it "wasn't designed to have that result." Powell Jobs has invested in both Ive's LoveFrom design and io hardware startups following his departure from Apple. Ive notes that "there wouldn't be LoveFrom" if not for her involvement. Ive's io company is being purchased by OpenAI for almost $6.5 billion, and with her investment, Powell Jobs stands to gain if the secretive gadget proves anywhere near as successful as the iPhone.

The pair gives away no extra details about the device that Ive is building with OpenAI, but Powell Jobs is expecting big things. She says she has watched "in real time how ideas go from a thought to some words, to some drawings, to some stories, and then to prototypes, and then a different type of prototype," Powell Jobs said. "And then something that you think: I can't imagine that getting any better. Then seeing the next version, which is even better. Just watching something brand new be manifested, it's a wondrous thing to behold."

"We've officially entered the age of watching robots clobber each other in fighting rings," writes Vice.com.

A kick-boxing competition was staged Sunday in Hangzhou, China using four robots from Unitree Robotics, reports Futurism. (The robots were named "AI Strategist", "Silk Artisan", "Armored Mulan", and "Energy Guardian".) "However, the robots weren't acting autonomously just yet, as they were being remotely controlled by human operator teams."

Although those ringside human controllers used quick voice commands, according to the South China Morning Post: Unlike typical remote-controlled toys, handling Unitree's G1 robots entails "a whole set of motion-control algorithms powered by large [artificial intelligence] models", said Liu Tai, deputy chief engineer at China Telecommunication Technology Labs, which is under research institute China Academy of Information and Communications Technology.
More from Vice: The G1 robots are just over 4 feet tall [130 cm] and weigh around 77 pounds [35 kg]. They wear gloves. They have headgear. They throw jabs, uppercuts, and surprisingly sharp kicks... One match even ended in a proper knockout when a robot stayed down for more than eight seconds. The fights ran three rounds and were scored based on clean hits to the head and torso, just like standard kickboxing...
Thanks to long-time Slashdot reader AmiMoJo for sharing the news.

An anonymous reader quotes a report from SiliconANGLE: Hugging Face has open-sourced the blueprints of two internally developed robots called HopeJR and Reachy Mini. The company debuted the machines on Thursday. Hugging Face is backed by more than $390 million in funding from Nvidia Corp., IBM Corp. and other investors. It operates a GitHub-like platform for sharing open-source artificial intelligence projects. It says its platform hosts more than 1 million AI models, hundreds of thousands of datasets and various other technical assets.

The company started prioritizing robotics last year after launching LeRobot, a section of its platform dedicated to autonomous machines. The portal provides access to AI models for powering robots and datasets that can be used to train those models. Hugging Face released its first hardware blueprint, a robotic arm design called the SO-100, late last year. The SO-100 was developed in partnership with a startup called The Robot Studio. Hugging Face also collaborated with the company on the HopeJR, the first new robot that debuted this week. According to TechCrunch, it's a humanoid robot that can perform 66 movements including walking.

HopeJR is equipped with a pair of robotic arms that can be remotely controlled by a human using a pair of specialized, chip-equipped gloves. HopeJR's arms replicate the movements made by the wearer of the gloves. A demo video shared by Hugging Face showed that the robot can shake hands, point to a specific text snippet on a piece of paper and perform other tasks. Hugging Face's other new robot, the Reachy Mini, likewise features an open-source design. It's based on technology that the company obtained through the acquisition of a venture-backed startup called Pollen Robotics earlier this year. Reachy Mini is a turtle-like robot that comes in a rectangular case. Its main mechanical feature is a retractable neck that allows it to follow the user with its head or withdraw into the case. This case, which is stationary, is compact and lightweight enough to be placed on a desk. Hugging Face will offer pre-assembled versions of its open-source Reach Mini and HopeJR robots for $250 and $3,000, with the first units starting to ship by the end of the year.

Artificial intelligence could soon outpace Bitcoin mining in energy consumption, according to Alex de Vries-Gao, a PhD candidate at Vrije Universiteit Amsterdam's Institute for Environmental Studies. His research estimates that by the end of 2025, AI could account for nearly half of all electricity used by data centers worldwide -- raising significant concerns about its impact on global climate goals.

"While companies like Google and Microsoft disclose total emissions, few provide transparency on how much of that is driven specifically by AI," notes DIGIT. To fill this gap, de Vries-Gao employed a triangulation method combining chip production data, corporate disclosures, and industry analyst estimates to map AI's growing energy footprint.

His analysis suggests that specialized AI hardware could consume between 46 and 82 terawatt-hours (TWh) in 2025 -- comparable to the annual energy usage of countries like Switzerland. Drawing on supply chain data, the study estimates that millions of AI accelerators from NVIDIA and AMD were produced between 2023 and 2024, with a potential combined power demand exceeding 12 gigawatts (GW). A detailed explanation of his methodology is available in his commentary published in Joule.