An anonymous reader quotes a report from The Associated Press: A nuclear power plant in Georgia has begun splitting atoms in one of its two new reactors, Georgia Power said Monday, a key step toward reaching commercial operation at the first new nuclear reactors built from scratch in decades in the United States. The unit of Atlanta-based Southern Co. said operators reached self-sustaining nuclear fission inside the reactor at Plant Vogtle, southeast of Augusta. That makes the intense heat that will be used to produce steam and spin turbines to generate electricity.

A third and a fourth reactor were approved for construction at Vogtle by the Georgia Public Service Commission in 2009, and the third reactor was supposed to start generating power in 2016. The company now says Unit 3 could begin commercial operation in May or June. Unit 4 is projected to begin commercial operation sometime between this November and March 2024.

The cost of the third and fourth reactors was originally supposed to be $14 billion. The reactors are now supposed to cost more than $30 billion. That doesn't include $3.68 billion that original contractor Westinghouse paid to the owners after going bankrupt, which brings total spending to more than $34 billion. Georgia Power said Unit 3 would continue startup testing to show that its cooling system and steam supply system will work at the intense heat and pressure that a nuclear reactor creates. After that, operators are supposed to link the reactor to the electrical grid and gradually raise it to full power.

Intel has completed development of its Intel 18A (1.8nm-class) and Intel 20A (2nm-class) fabrication processes that will be used to make the company's own products, as well as chips for clients of its Intel Foundry Services (IFS) division, reports UDN. From a report: Wang Rui, president and chairman of Intel China, said at an event that the company had finalized the development of its Intel 18A (18 angstroms-class) and Intel 20A (20 angstroms-class) fabrication processes. This does not mean that the production nodes are ready to be used for commercial manufacturing, but rather that Intel has determined all specifications, materials, requirements, and performance targets for both technologies.

Intel's 20A fabrication technology will rely on gate-all-around RibbonFET transistors and will use backside power delivery. Shrinking metal pitches, introducing all-new transistor structure and adding backside power delivery at the same time is a risky move, but it is expected that 20A will allow Intel to leapfrog the company's competitors -- TSMC and Samsung Foundry. Intel plans to start using this node in the first half of 2024. Intel's 18A manufacturing process will further refine the company's RibbonFET and PowerVia technologies, as well as shrink transistor sizes. Development of this node is apparently going so well that Intel pulled in its introduction from 2025 to the second half of 2024. Intel originally planned to use ASML's Twinscan EXE scanners with a 0.55 numerical aperture (NA) optics for its 1.8 angstroms node, but because it decided to start using the technology sooner, it will have to rely on extensive use of existing Twinscan NXE scanners with 0.33 NA optics, as well as EUV double patterning.

Unprecedented access to space is leading to all sorts of cool new ideas, including the prospect of storing data on the lunar surface. Cloud computing startup Lonestar Data Holdings announced the results of its latest funding round, taking it one step closer to this very goal. Gizmodo reports: The Florida-based company raised $5 million in seed funding to establish lunar data centers, Lonestar announced in a press release on Monday. Lonestar wants to build a series of data centers on the Moon and establish a viable platform for data storage and edge processing (i.e. the practice of processing data near the source, as a means to reduce latency and improve bandwidth) on the lunar surface. "Data is the greatest currency created by the human race," Chris Stott, founder of Lonestar, said in an April 2022 statement. "We are dependent upon it for nearly everything we do and it is too important to us as a species to store in Earth's ever more fragile biosphere. Earth's largest satellite, our Moon, represents the ideal place to safely store our future."

In December 2021, Lonestar successfully ran a test of its data center on board the International Space Station. The company is now ready to launch a small data center box to the lunar surface later this year as part of Intuitive Machines's second lunar mission, IM-2 (the company's first mission, IM-1, is expected to launch in June). Intuitive Machines is receiving funding from NASA's Commercial Lunar Payload Services program for delivering research projects to the Moon as part of the space agency's Artemis program. The lunar data centers will initially be geared towards remote data storage and disaster recovery, allowing companies to back up their data and store it on the Moon. In addition, the data centers could assist with both commercial and private ventures to the lunar environment.

The miniature data center weighs about 2 pounds (1 kilogram) and has a capacity of 16 terabytes, Stott told SpaceNews. He said the first data center will draw power and communications from the lander, but the ones that will follow (pending its success) will be standalone data centers that the company hopes to deploy on the lunar surface by 2026. The test is only supposed to last for the duration of the IM-2 mission, which is expected to be around 11-14 days, an Intuitive Machines spokesperson told SpaceNews.

Formerly known as LinuxBIOS, coreboot is defined by Wikipedia as "a software project aimed at replacing proprietary firmware (BIOS or UEFI) found in most computers with a lightweight firmware."

Phoronix is wondering if there's about to be a big announcement from AMD: AMD dropped a juicy tid-bit of information to be announced next month with "openSIL" [an open-source AMD x86 silicon initialization library], complete with AMD Coreboot support....

While about a decade ago AMD was big into Coreboot and at the time committed to it for future hardware platforms (2011: AMD To Support Coreboot On All Future CPUs) [and] open-source AGESA at the time did a lot of enabling around it, that work had died off. In more recent years, AMD's Coreboot contributions have largely been limited to select consumer APU/SoC platforms for Google Chromebook use. But issues around closing up the AGESA as well as concerns with the AMD Platform Security Processor (PSP) have diminished open-source firmware hopes in recent years....

For the Open Compute Project Regional Summit in Prague, there is a new entry added with a title of OSF on AMD — Enabled by openSIL (yes, folks, OSF as in "Open-Source Firmware").... [H]opefully this will prove to be a monumental shift for open-source firmware in the HPC server space.
From the talk's description: openSIL (AMD open-source x86 Silicon Initialization Library) offers the versatility, scalability, and light weight interface to allow for ease of integration with open-source and/or proprietary host boot solutions such as coreboot, UEFI and others and adds major flexibility to the overall platform design.

In other words, this library-based solution simply allows a platform integrator to scale from feature rich solutions such as UEFI to slim, lightweight, and secure solutions such as coreboot.
The description promises the talk will include demonstrations "highlighting system bring-up using openSIL integrated with coreboot and UEFI Host Firmware stacks on AMD's Genoa based platforms."

Perhaps AI "needs a little more work," writes ZDNet columnist Chris Matyszczyk — noting problems with the automated voice-recognition systems at McDonald's drivethrough lanes. The trouble started when TikTok-er Ren Adams ordered hash browns, sweet tea and a Coke. All seemed fine until, at a second drive-thru lane, another car pulled up. Adams' AI helper seems to have overheard that order and added it to Adams'. Adams tried to make the robot see sense. Or, rather hear it. Instead, the robot removed the errant Diet Coke and replaced it with, oh, nine sweet teas instead of one.

Which suggests something of a problem. When your robot drive-thru employee makes a mistake, to whom can you complain? Complaining to the robot seems to create an extra layer of complication and the potential for even greater misunderstanding.

Adams, indeed, isn't alone. Here's Caitlyn Sykora (not) ordering $254 of McNuggets meals. And here's Madilynn Cameron wanting a large cup of water and a cup of ice cream and discovering butter is included. She seems to have given up.
The customer who'd ordered one sweet tea and instead got nine also drove off in a huff, according to the end of their TikTok video.

Matyszczyk's conclusion? " if you're not so good at fixing ice-cream machines, how good will you be at maintaining thousands of robot order-takers?"

Matt_Bennett (Slashdot reader #79,107) writes: Swedish power company Vattenfall released a study on the interactions of seabirds and offshore wind turbines. They used cameras and radar to record the tracks of the birds during daylight hours at Aberdeen Offshore Wind Farm over peak periods of bird activity in 2020 and 2021.

The study observed no collisions or even narrow escapes between birds and rotor blades. In 97.7% of the recordings, the birds avoided the RSZ (rotor swept zone).
The company (owned by the Swedish government) spent €3 million on the two-year study, according to Electrek, and now has ten thousand videos of birds flying...nowhere near the wind turbines. Herring gulls avoided the rotor blades by a full 90-110 meters (295-361 feet) while kittiwakes flew even further from the blades — 140-160 meters (459-525 feet).

"By way of comparison, each of these human-related sources kill millions or even billions of birds per year: fossil fuels, deforestation, pesticides, windows, and the common housecat."

According to Reuters, Elon Musk's medical device company, Neuralink, was denied permission last year to begin human trials of a revolutionary brain implant to treat intractable conditions such as paralysis and blindness. The U.S. Food and Drug Administration (FDA) outlined dozens of issues the company must address before human testing can begin, according to seven current and former employees. From the report: The agency's major safety concerns involved the device's lithium battery; the potential for the implant's tiny wires to migrate to other areas of the brain; and questions over whether and how the device can be removed without damaging brain tissue, the employees said. A year after the rejection, Neuralink is still working through the agency's concerns. Three staffers said they were skeptical the company could quickly resolve the issues -- despite Musk's latest prediction at a Nov. 30 presentation that the company would secure FDA human-trial approval this spring.

Neuralink has not disclosed details of its trial application, the FDA's rejection or the extent of the agency's concerns. As a private company, it is not required to disclose such regulatory interactions to investors. During the hours-long November presentation, Musk said the company had submitted "most of our paperwork" to the agency, without specifying any formal application, and Neuralink officials acknowledged the FDA had asked safety questions in what they characterized as an ongoing conversation. Such FDA rejections do not mean a company will ultimately fail to gain the agency's human-testing approval. But the agency's pushback signals substantial concerns, according to more than a dozen experts in FDA device-approval processes.

The rejection also raises the stakes and the difficulty of the company's subsequent requests for trial approval, the experts said. The FDA says it has approved about two-thirds of all human-trial applications for devices on the first attempt over the past three years. That total rose to 85% of all requests after a second review. But firms often give up after three attempts to resolve FDA concerns rather than invest more time and money in expensive research, several of the experts said. Companies that do secure human-testing approval typically conduct at least two rounds of trials before applying for FDA approval to commercially market a device.

An anonymous reader quotes a report from CBS News: The U.S. Postal Service's plans for a nationwide fleet of electric vehicles are getting closer to being realized. The service awarded contracts on Tuesday for 9,250 battery electric vehicles and for more than 14,000 charging stations. The vehicles purchased are Ford E-Transit Battery Electric Vehicles (BEVs), which according to USPS are "100 percent electric." It's part of the agency's plans, announced in December, to make 75% of its newly acquired vehicles, known as Next Generation Delivery Vehicles, over the next five years electric. After 2026, NGDV purchases will be 100% electric, the agency said. The goal is to have a fleet of 66,000 electric vehicles deployed by 2028.

Three suppliers were awarded contracts for more than 14,000 charging stations, as well, USPS said, to kick off its Electric Vehicle Supply Equipment (EVSE) inventory. However, the agency did note that the contracts they just awarded will not provide an immediate supply. The vehicles won't be delivered until December, assuming that the agency successfully finishes its Supplemental Environmental Impact Statement and other necessary tasks. It also remains unclear where the vehicles and charging stations will be placed, as those details have yet to be finalized, the agency said.

As such, a contract has also been awarded for the agency to acquire 9,250 commercial-off-the-shelf internal combustion engine vehicles "to fill the urgent need for vehicles." In December, the agency said that 21,000 COTS vehicles will be purchased and are "expected to be battery electric," but said that depends "on market availability and operational feasibility." In this case, the internal combustion engine vehicles will be gas-powered and made by Fiat Chrysler Automobile, a spokesperson for USPS told CBS News. They added that, unlike older USPS vehicles, these will "feature air conditioning and advanced safety technology and are more suited to modern day operational requirements." "We have an urgent need to replace some of our vehicles as soon as possible, and in those instances we will look to obtain vehicles that can be provided to us expeditiously, recognizing that there are a limited amount of BEV options currently available and that the charging infrastructure buildout will also take some time," they said in an email.

"...Today there remain routes and applications which do not support BEVs. As BEV technology matures and capabilities increase, the Postal Service will continue to review its ability to utilize and expand BEV usage."

The first PCIe 5.0 SSDs are slated to ship this year with massive heatsinks and predictably high prices. Tom's Hardware reports: There are multiple M.2 PCIe 5.0 SSDs slated to ship this year, and the first model looks to be the Gigabyte Aorus Gen5 10000, which as the name inventively implies can deliver up to 10,000 MB/s. Earlier rumors suggested the drive would be able to hit 12,000 MB/s reads and 10,000 MB/s writes, so performance was apparently reigned in while getting the product ready for retail. The Gigabyte Aorus SSD uses the Phison E26 controller, which will be common on a lot of the upcoming models. Silicon Motion is working on its new SM2508 controller that may offer higher overall performance, but it's a bit further out and may not ship this year. The other thing to note with the Aorus is the massive heatsink that comes with the drive, which seems to be the case with all the other Gen5 SSD prototypes we've seen as well. Clearly, these new drives are going to get just a little bit warm.

The Gigabyte drive is currently listed on Amazon and Newegg, though the latter is currently sold out while the former is only available via a third-party marketplace seller -- at a whopping $679.89 for the 2TB model. That's almost certainly not the MSRP or a reflection of what MSRP might end up being once the drive becomes more widely available, which should happen in the coming month or two.

The other PCIe 5.0 M.2 SSD that's now available is the Inland TD510 2TB, available at Microcenter for just $349.99 -- assuming you have a Microcenter within driving distance. Inland is Microcenter's own brand of drive, and while the cooler that comes with the SSD isn't quite as large as the Aorus, it does feature a small fan for active cooling. Word is that the fan can be quite loud for something this small, so not a great feature in other words. Like the Aorus 10000, the Inland TD510 uses the Phison E26 controller and has the same 10,000 MB/s reads and 9,500 MB/s writes specification. Where Gigabyte doesn't currently list random read/write speeds, the Microcenter page lists up to 1.5 million IOPS read and 1.25 million IOPS write for the Inland drive. Both drives also have an endurance rating of 1,400 TBW, with read/write power use of around 11W.

An anonymous reader quotes a report from The Verge's Alex Heath: Meta plans to release its first pair of smart glasses with a display in 2025 alongside a neural interface smartwatch designed to control them, The Verge has learned. Meanwhile, its first pair of full-fledged AR glasses, which CEO Mark Zuckerberg has predicted will eventually be as widely used as mobile phones, is planned for 2027. The details were shared with thousands of employees in Meta's Reality Labs division on Tuesday during a roadmap presentation of its AR and VR efforts that was shared with The Verge. Altogether, they show how Meta is planning to keep investing in consumer hardware after a series of setbacks and broader cost cutting across the company.

With regards to the VR roadmap, employees were told that Meta's flagship Quest 3 headset coming later this year will be two times thinner, at least twice as powerful, and cost slightly more than the $400 Quest 2. Like the recently announced Quest Pro, it will prominently feature mixed reality experiences that don't fully immerse the wearer, thanks to front-facing cameras that pass through video of the real world. [...] There will be 41 new apps and games shipping for the Quest 3, including new mixed reality experiences to take advantage of the updated hardware, Rabkin said. In 2024, he said that Meta plans to ship a more "accessible" headset codenamed Ventura. "The goal for this headset is very simple: pack the biggest punch we can at the most attractive price point in the VR consumer market."

During Tuesday's roadmap presentation, Alex Himel, the company's vice president for AR, laid out the plan for a bevy of devices through 2027. The first launch will come this fall with the second generation of Meta's camera-equipped smart glasses it released in 2021 with Luxottica, the parent company of Ray-Ban. In 2025, Himel said the third generation of the smart glasses will ship with a display that he called a "viewfinder" for viewing incoming text messages, scanning QR codes, and translating text from another language in real time. The glasses will come with a "neural interface" band that allows the wearer to control the glasses through hand movements, such as swiping fingers on an imaginary D-pad. Eventually, he said the band will let the wearer use a virtual keyboard and type at the same words per minute as what mobile phones allow. While Meta halted development of its smartwatch with dual cameras,Himel said that the company is still working on another smartwatch to accompany its 2025 glasses. "We don't want people to have to choose between an input device on their wrist and smartwatch functionality that they've come to love," he said. "So we are building a neural interfaces watch. Number one, this device will do input: input to control your glasses, input to control the functionality on your wrist, and input to control the world around you."

The Verge's Alex Heath adds: "Meta's first true pair of AR glasses, which the company has been internally developing for 8 years under the codename Orion, are more technically advanced, expensive, and designed to project high-quality holograms of avatars onto the real world." These glasses will "won't be released to the public until 2027," but an "internal launch" for employees will begin in 2024.

As for advertising, Meta is planning to utilize its existing business model for these future devices. "We should be able to run a very good ads business," he said. "I think it's easy to imagine how ads would show up in space when you have AR glasses on. Our ability to track conversions, which is where there has been a lot of focus as a company, should also be close to 100 percent."

"If we're hitting anything near projections, it will be a tremendous business," he said. "A business unlike anything we've seen on mobile phones before."

Tesla has finally started to open some Supercharger stations to non-Tesla electric cars owners in the US and it explained how it works. Electrek reports: As we previously reported, everything is handled through the app. Non-Tesla EV owners simply have to download the Tesla app, create an account, add a credit card for payment, and then they can roll up to some of the select few Supercharger stations now equipped with a Magic Dock -- primarily in New York for now. In the app, electric car owners can see the station and select the stall where they park. After, they simply have to grab the handle where the CCS adapter will come out of the Magic Dock:

At the moment, it appears that only half a dozen stations in the state of New York are available to non-Tesla EV owners, but the number is expected to grow rapidly as Tesla deploys the Magic Dock (the integrated CCS adapter to work with non-Tesla EVs) at more stations and builds new ones.

After widely rolling out an Energy Saver mode, Google has made four optimizations to Chrome for Mac that allows the browser to match the battery life you get when using Safari. 9to5Google reports: Google conducted testing on a MacBook Pro (13", M2, 2022 with 8 GB RAM running macOS Ventura 13.2.1) with Chrome 110.0.5481.100 in February of 2023. It showed that you can "browse for 17 hours or watch YouTube for 18 hours." For comparison, Apple touts up to 17 hours of wireless web browsing, and up to 20 hours Apple TV app movie playback. Meanwhile, Google uses this open-source benchmarking suite to run tests, and says that users will also "see performance gains on older models." Four changes from waking the CPU less often to tuning memory compression are specifically credited:

- Eliminating unnecessary redraws: "We navigated on real-world sites with a bot and identified Document Object Model (DOM) change patterns that don't affect pixels on the screen. We modified Chrome to detect those early and bypass the unnecessary style, layout, paint, raster and gpu steps. We implemented similar optimizations for changes to the Chrome UI."
- Fine tuning iframes: "...we fine-tuned the garbage collection and memory compression heuristics for recently created iframes. This results in less energy consumed to reduce short-term memory usage (without impact on long-term memory usage)."
- Tweaking timers: "...Javascript timers still drive a large proportion of a Web page's power consumption. As a result, we tweaked the way they fire in Chrome to let the CPU wake up less often. Similarly, we identified opportunities to cancel internal timers when they're no longer needed, reducing the number of times that the CPU is woken up."
- Streamlining data structures: "We identified data structures in which there were frequent accesses with the same key and optimized their access pattern."

MojoKid writes: At CES 2023, AMD unveiled an array of Ryzen 7000 series Zen 4 processors, including new gaming-targeted X3D models that featured integrated 3D V-Cache, similar to the Ryzen 7 5800X3D. The processors go on sale tomorrow, but review embargos for AMD's latest socket AM5 flagship, the Ryzen 9 7950X3D, lifted today. As its name implies, the new Ryzen 9 7950X3D has a similar core configuration to the existing Ryzen 9 7950X (16-cores/32-threads), but this specialized CPU also packs an additional 64MB of 3D V-Cache, fused to one of its 8-core compute core dies (CCD). The CCD without 3D V-Cache operates like a standard AMD Ryzen 9 7950X, while the 3D V-Cache enabled CCD will have a more conservative voltage and frequency curve. Gaming performance received a massive boost with this new CPU, while multi-threaded content creation tests are roughly in-line with the standard 7950X. Power efficiency also shows a large, measurable improvement due to the chip relying less often on system memory.

An explosion in proposed clean energy ventures in America "has overwhelmed the system for connecting new power sources to homes and businesses," reports the New York Times: So many projects are trying to squeeze through the approval process that delays can drag on for years, leaving some developers to throw up their hands and walk away.

More than 8,100 energy projects — the vast majority of them wind, solar and batteries — were waiting for permission to connect to electric grids at the end of 2021, up from 5,600 the year before, jamming the system known as interconnection.... PJM Interconnection, which operates the nation's largest regional grid, stretching from Illinois to New Jersey, has been so inundated by connection requests that last year it announced a freeze on new applications until 2026, so that it can work through a backlog of thousands of proposals, mostly for renewable energy.

It now takes roughly four years, on average, for developers to get approval, double the time it took a decade ago. And when companies finally get their projects reviewed, they often face another hurdle: the local grid is at capacity, and they are required to spend much more than they planned for new transmission lines and other upgrades. Many give up. Fewer than one-fifth of solar and wind proposals actually make it through the so-called interconnection queue, according to research from Lawrence Berkeley National Laboratory. "From our perspective, the interconnection process has become the No. 1 project killer," said Piper Miller, vice president of market development at Pine Gate Renewables, a major solar power and battery developer....

A potentially bigger problem for solar and wind is that, in many places around the country, the local grid is clogged, unable to absorb more power. That means if a developer wants to build a new wind farm, it might have to pay not just for a simple connecting line, but also for deeper grid upgrades elsewhere.... These costs can be unpredictable. In 2018, EDP North America, a renewable energy developer, proposed a 100-megawatt wind farm in southwestern Minnesota, estimating it would have to spend $10 million connecting to the grid. But after the grid operator completed its analysis, EDP learned the upgrades would cost $80 million. It canceled the project.

That creates a new problem: When a proposed energy project drops out of the queue, the grid operator often has to redo studies for other pending projects and shift costs to other developers, which can trigger more cancellations and delays. It also creates perverse incentives, experts said. Some developers will submit multiple proposals for wind and solar farms at different locations without intending to build them all. Instead, they hope that one of their proposals will come after another developer who has to pay for major network upgrades. The rise of this sort of speculative bidding has further jammed up the queue.

Last week ZDNet reported Linux had added upstream support for the Apple M1 Pro, M1 Max, and M1 Ultra chips and then concluded that "newer Mac owners can look forward to running Linux on their M1-powered machines."

Saturday Asahi Linux called ZDNet's story "misleading and borderline false," posting on Twitter that "You will not be able to run Ubuntu nor any other standard distro with 6.2 on any M1 Mac. Please don't get your hopes up." We are continuously upstreaming kernel features, and 6.2 notably adds device trees and basic boot support for M1 Pro/Max/Ultra machines. However, there is still a long road before upstream kernels are usable on laptops. There is no trackpad/keyboard support upstream yet.

While you can boot an upstream 6.2 kernel on desktops (M1 Mac Mini, M1 Max/Ultra Mac Studio) and do useful things with it, that is only the case for 16K page size kernel builds. No generic ARM64 distro ships 16K kernels today, to our knowledge.

Our goal is to upstream everything, but that doesn't mean distros instantly get Apple Silicon support. As with many other platforms, there is some integration work required. Distros need to package our userspace tooling and, at this time, offer 16K kernels. In the future, once 4K kernel builds are somewhat usable, you can expect zero-integration distros to somewhat work on these machines (i.e. some hardware will work, but not all, or only partially). This should be sufficient to add a third-party repo with the integration packages.

But for out-of-the-box hardware support, distros will need to work with us to get everything right. We are already working with some, and we expect to announce official Apple Silicon support for a mainstream distro in the near future. Just not quite yet!

An anonymous reader shares this report from the Washington Post: Sucking carbon dioxide out of the sky — or "direct air capture," as it is known by experts and scientists — is a bit like a time machine for climate change. It removes CO2 from the atmosphere and stores it deep underground, almost exactly the reverse of what humanity has been doing for centuries by burning fossil fuels. Its promise? That it can help run back the clock, undoing some of what we have done to the atmosphere and helping to return the planet to a cooler state.

The problem with direct air capture, however, has been that it takes energy — a lot of energy.... But if the energy powering that comes from fossil fuels, direct air capture starts to look less like a time machine than an accelerator: a way to emit even more CO2. Now, however, a company is working to combine direct air capture with a relatively untapped source of energy: Heat from Earth's crust. Fervo Energy, a geothermal company headquartered in Houston, announced on Thursday that it will design and engineer the first purpose-built geothermal and direct air capture plant. With the help of a grant from the Chan Zuckerberg Initiative, the company hopes to have a pilot facility online in 3 to 5 years.

If it works, it will be a way to produce carbon-free electricity, while reducing CO2 in the atmosphere at the same time. In short, a win-win for the climate. "You have to have your energy from a carbon-free source" for direct air capture to make sense, said Timothy Latimer, the CEO of Fervo Energy. "Geothermal is a great match...." Geothermal wells don't, of course, get anywhere close to Earth's core, but a geothermal well drilled just 1 to 2 miles into hot rocks below the surface can reach temperatures of up to 1,000 degrees. Water is pumped into the well, heated and returned to the surface, where it can be converted into steam and electricity. Even after generating electricity, most geothermal plants have a lot of waste heat — often clocking in around 212 degrees. And conveniently, that happens to be the exact temperature needed to pull carbon dioxide out of an air filter and bury it underground.
The article notes a study which found that if air capture were combined with all the geothermal plants currently in America, the country "could suck up around 12.8 million tons of carbon dioxide every year."

And "Unlike wind and solar, a geothermal plant can be on all of the time, producing electricity even when the wind isn't blowing or the sun isn't shining."

An interesting profile of EV entrepreneur Horace Luke from Rest of World: During his time working for companies like Microsoft and HTC on projects like the Xbox gaming system and Android phones, Luke mulled over the idea of mobility. In 2011, he pitched the idea that would form the core of his company Gogoro: an electric vehicle that didn't have to take up space and time charging its batteries, but instead relied on a network of batteries that could be swapped at roadside stations, like filling up a gas tank. Multiple investors and vehicle makers told him the idea was impossible.

Today, Gogoro battery-swapping stations are as common as gas stations in Taiwan, and the network supports nearly 400,000 battery swaps a day, by over 526,000 riders. Last year, according to the Taiwanese government, 12% of all scooters sold in Taiwan were electric, and over 90% of those relied on Gogoro batteries. But in order to make the battery network a reality, Gogoro didn't have to develop just the batteries but also the vehicles that use them, along with an internal management software that encompasses everything from the supply of vehicle parts to the number of charged batteries at stations to how far riders can go before their next swap.
And the company now has pilot projects in Germany, India, Indonesia, Israel, the Philippines, Singapore, and South Korea.

"We extended the capabilities of ChatGPT to robotics," brags a blog post from Microsoft's Autonomous Systems and Robotics research group, "and controlled multiple platforms such as robot arms, drones, and home assistant robots intuitively with language."

They're exploring how to use ChatGPT to "make natural human-robot interactions possible... to see if ChatGPT can think beyond text, and reason about the physical world to help with robotics tasks." We want to help people interact with robots more easily, without needing to learn complex programming languages or details about robotic systems. The key challenge here is teaching ChatGPT how to solve problems considering the laws of physics, the context of the operating environment, and how the robot's physical actions can change the state of the world.

It turns out that ChatGPT can do a lot by itself, but it still needs some help. Our technical paper describes a series of design principles that can be used to guide language models towards solving robotics tasks. These include, and are not limited to, special prompting structures, high-level APIs, and human feedback via text.... In our work we show multiple examples of ChatGPT solving robotics puzzles, along with complex robot deployments in the manipulation, aerial, and navigation domains....

We gave ChatGPT access to functions that control a real drone, and it proved to be an extremely intuitive language-based interface between the non-technical user and the robot. ChatGPT asked clarification questions when the user's instructions were ambiguous, and wrote complex code structures for the drone such as a zig-zag pattern to visually inspect shelves. It even figured out how to take a selfie! We also used ChatGPT in a simulated industrial inspection scenario with the Microsoft AirSim simulator. The model was able to effectively parse the user's high-level intent and geometrical cues to control the drone accurately....

We are excited to release these technologies with the aim of bringing robotics to the reach of a wider audience. We believe that language-based robotics control will be fundamental to bring robotics out of science labs, and into the hands of everyday users.

That said, we do emphasize that the outputs from ChatGPT are not meant to be deployed directly on robots without careful analysis. We encourage users to harness the power of simulations in order to evaluate these algorithms before potential real life deployments, and to always take the necessary safety precautions. Our work represents only a small fraction of what is possible within the intersection of large language models operating in the robotics space, and we hope to inspire much of the work to come.tics to the reach of a wider audience. We believe that language-based robotics control will be fundamental to bring robotics out of science labs, and into the hands of everyday users.
ZDNet points out that Google Research and Alphabet's Everyday Robots "have also worked on similar robotics challenges using a large language models called PaLM, or Pathways Language Model, which helped a robot to process open-ended prompts and respond in reasonable ways."

More than 126,000 Californians are without electricity, reports ABC News. But Reuters notes that meanwhile "more than 400,000 customers of Detroit based DTE Energy remained without power on Saturday, the Detroit News reported," suffering through "a separate storm that clobbered the U.S. Plains, Midwest and Great Lakes regions earlier this week" that finally moved over the Atlantic.

And ABC News notes that as of Saturday morning, "more than 30 million Americans are under weather alerts in the West" — roughly 1 in 11 Americans — "ranging from blizzard warnings in the mountains near Los Angeles to wind chill alerts in the Northern Plains" near Wyoming. But California's problems came from its own major storm that delivered heavy snow, record rainfall, and damaging winds — a storm that "will be moving from southern California across the entire country over the next few days, eventually moving northeast by Tuesday." The Los Angeles area saw record rainfall on Friday, and it came along with 50- to 70-mile-per-hour winds. Burbank, California, saw 4.6 inches of rain Friday — stranding cars in floods and causing dozens of flight delays and cancellations. Records for daily rainfall were also set at the Los Angeles International Airport and the cities of Fresno, Bakersfield, Modesto and Oxnard.... Multiple stretches of I-5 in Los Angeles County were shuttered on Saturday due to rain and snow.
Snowflakes even fell around the "Hollywood" sign, reports Reuters. But bad weather wasn't just hitting southern California: In Northern California, San Francisco was expected to experience record cold temperatures on Saturday, and the National Weather Service warned residents of the state capital of Sacramento to avoid travel from Sunday through Wednesday as rain and snow started up again after a reprieve on Saturday. "Extreme impacts from heavy snow & winds will cause extremely dangerous to impossible driving conditions & likely widespread road closures & infrastructure impacts!" the agency said on Twitter. The next set of storms, expected to hit on Sunday, will bring wind gusts of up to 50 miles per hour (80 kph) in the Sacramento Valley, and up to 70 miles per hour in the nearby Sierra Nevada mountains....

A massive low-pressure system driven from the Arctic was responsible for the unusual conditions, said Bryan Jackson, a forecaster at the NWS Weather Prediction Center in College Park, Maryland.
This week one political cartoonist suggested a connection between "crazy weather" and climate change.

An anonymous reader shares this story from The New York Times. (Alternate URL for a shorter version here.) Taiwan Semiconductor Manufacturing Company, the world's biggest maker of advanced computer chips, is upgrading and expanding a new factory in Arizona that promises to help move the United States toward a more self-reliant technological future. But to some at the company, the $40 billion project is something else: a bad business decision.

Internal doubts are mounting at the Taiwanese chip maker over its U.S. factory, according to interviews with 11 TSMC employees, who declined to be identified because they were not authorized to speak publicly. Many of the workers said the project could distract from the research and development focus that had long helped TSMC outmaneuver rivals. Some added that they were hesitant to move to the United States because of potential culture clashes.... Its factory expansion in the northern outskirts of Phoenix is meant to bring advanced microchip production closer to the United States and away from any potential standoff with China. Yet the effort has stoked internal apprehension, with high costs and managerial challenges showing how difficult it is to transplant one of the most complicated manufacturing processes known to man halfway across the world.

The pressure for the Arizona factory to succeed is immense. Failure would mean a setback for U.S. efforts to cultivate the advanced chip manufacturing that mostly moved to Asia decades ago. And TSMC would have spent billions on a plant that did not produce enough viable chips to make it worth the effort.