Microsoft is extending the repairability program it introduced for its Surface PC products to include replacement parts for its Xbox Wireless Controller and Xbox Elite Controller Series 2 products. Neowin reports: The page on the Microsoft Store site shows that replacement parts are available for the top case for both versions of the controller, along with replacement buttons. In addition, Microsoft is selling Replacement Input PCBA boards for those Xbox controllers, along with the Replacement PCBA and Motor Assembly parts as well. The parts do come with a one-year warranty.

In a new support page, Microsoft makes it clear that these parts should only be purchased and used for Xbox controllers that are out of their normal warranty period. On another support page, the company adds: "These types of repairs require moderate technical skill, and are suited for enthusiasts, professionals, or those with prior experience in electronic disassembly. If this is your first attempt at performing a repair, use caution and follow our safety recommendations and step-by-step instructions."

Microsoft also says that certain tools, which are not directly sold by the company, will be needed to replace and repair the controllers. They include a plastic pry tool, TR8 and T5 Torx screwdriver bits and plastic tweezers. Microsoft does offer PDF files (PDF) and even offers YouTube video tutorials for repairing the Xbox Wireless Controller and the Xbox Elite Wireless Controller Series 2.

Long-time Slashdot reader nairnr shares a report from The Register: Cloud giant AWS will start charging customers for public IPv4 addresses from next year, claiming it is forced to do this because of the increasing scarcity of these and to encourage the use of IPv6 instead. It is now four years since we officially ran out of IPv4 ranges to allocate, and since then, those wanting a new public IPv4 address have had to rely on address ranges being recovered, either from from organizations that close down or those that return addresses they no longer require as they migrate to IPv6.

If Amazon's cloud division is to be believed, the difficulty in obtaining public IPv4 addresses has seen the cost of acquiring a single address rise by more than 300 percent over the past five years, and as we all know, the business is a little short of cash at the moment, so is having to pass these costs on to users. "This change reflects our own costs and is also intended to encourage you to be a bit more frugal with your use of public IPv4 addresses and to think about accelerating your adoption of IPv6 as a modernization and conservation measure,' writes AWS Chief Evangelist Jeff Barr, on the company news blog.

The update will come into effect on February 1, 2024, when AWS customers will see a charge of $0.005 (half a cent) per IP address per hour for all public IPv4 addresses. These charges will apparently apply whether the address is attached to a service or not, and like many AWS charges, appear inconsequential at first glance but can mount up over time if a customer is using many of them. These charges will apply to all AWS services including EC2, Relational Database Service (RDS) database instances, Elastic Kubernetes Service (EKS) nodes, and will apply across all AWS regions, the company said. However, customers will not be charged for IP addresses that they own and bring to AWS using Amazon's BYOIP feature. AWS offers a free tier for EC2, and this will include 750 hours of public IPv4 address usage per month for the first 12 months, starting from the same date the charges do.

An anonymous reader quotes a report from TechCrunch: Amazon achieved its "fastest Prime speeds ever" last quarter, the online retail giant announced on Monday. The company says it has delivered more than 1.8 billion units to U.S. Prime members so far this year, nearly four times what it delivered at those speeds by this point in 2019. Across the top 60 largest U.S. metro areas, more than half of Prime member orders arrived the same or next day. Same-day delivery is currently available on millions of items for customers across more than 90 U.S. metro areas, and Amazon plans to double the number of delivery sites in the coming years.

Amazon notes that the average time from picking a customer's items to positioning the customer's package on the outbound dock is 11 minutes in same-day facilities, more than an hour faster than its traditional fulfillment centers. For context, same-day facilities are stocked with what customers in those areas are purchasing, while traditional facilities are larger and include a more vast assortment of items. "As we make these changes, we are seeing that they have the added benefit of reducing costs, too," Doug Herrington, CEO of Worldwide Amazon Stores, wrote in a blog post. "Regionalizing our network reduces miles traveled and handoffs. Since the beginning of this year alone, the distance between our sites and the customer decreased by 15%, with 12% fewer touchpoints within our middle mile network. Improved product placement gets items even closer to customers, making our delivery system more efficient. And our Same-Day Delivery network is not only our fastest way to get products to customers, it is also one of our lowest cost ways."

"A new group of automotive super friends is banding together," reports the Verge, "promising to build the next big North American electric vehicle charging network." These worldwide automakers — BMW, General Motors, Honda, Hyundai, Kia, Mercedes-Benz, and Stellantis — announced a planned joint venture Wednesday to erect easy-to-activate DC fast chargers along US and Canadian highways and in urban environments.

The grand plan for the currently unnamed partnership is to install "at least" 30,000 high-speed EV chargers by 2030, with the first ones to open summer 2024 in the US. The collective plans to leverage National Electric Vehicle Infrastructure (NEVI) funding in the US and will also use other private and public funding from state and federal sources to build out the network... The new stations will connect and charge EV models made by the partnered automakers without having to fumble with another charging station app. The companies also plan to integrate the developing "Plug and Charge" standard that the Federal Highway Administration is attempting to standardize... All stations will include the standardized Tesla North American Charging Standard (NACS) ports and also the current widely used Combined Charging System (CCS) plugs.
"The new joint venture is also planned to be entirely powered by renewable energy," the article adds.

But "It's not known if renewable energy will directly power them or if the companies plan to buy credits like Rivian announced Tuesday."

After Red Hat's decision to only share RHEL source code with subscribers, AlmaLinux asked their bug report submitters to "attempt to test and replicate the problem in CentOS Stream as well, so we can focus our energy on correcting it in the right place."

Red Hat told Ars Technica they are "eager to collaborate" on their CentOS Stream distro, "even if we ultimately compete in a business sense. Differentiated competition is a sign of a healthy ecosystem."

But Red Hat still managed to ruffled some feathers, reports ZDNet: AlmaLinux Infrastructure Team Leader Jonathan Wright recently posted a CentOS Stream fix for CVE-2023-38403, a memory overflow problem in iperf3. Iperf3 is a popular open-source network performance test. This security hole is an important one, but not a huge problem.

Still, it's better by far to fix it than let it linger and see it eventually used to crash a server. That's what I and others felt anyway. But, then, a senior Red Hat software engineer replied, "Thanks for the contribution. At this time, we don't plan to address this in RHEL, but we will keep it open for evaluation based on customer feedback."

That went over like a lead balloon.

The GitLab conversation proceeded:

AlmaLinux: "Is customer demand really necessary to fix CVEs?"

Red Hat: "We commit to addressing Red Hat defined Critical and Important security issues. Security vulnerabilities with Low or Moderate severity will be addressed on demand when [a] customer or other business requirements exist to do so."

AlmaLinux: "I can even understand that, but why reject the fix when the work is already done and just has to be merged?"

At this point, Mike McGrath, Red Hat's VP of Core Platforms, AKA RHEL, stepped in. He explained, "We should probably create a 'what to expect when you're submitting' doc. Getting the code written is only the first step in what Red Hat does with it. We'd have to make sure there aren't regressions, QA, etc. ... So thank you for the contribution, it looks like the Fedora side of it is going well, so it'll end up in RHEL at some point."

Things went downhill rapidly from there...

On Reddit, McGrath said, "I will admit that we did have a great opportunity for a good-faith gesture towards Alma here and fumbled."

Finally, though the Red Hat Product Security team rated the CVE as "'Important,' the patch was merged.
Coincidentally, last month AlmaLinux announced that its move away from 1:1 compatibility with RHEL meant "we can now accept bug fixes outside of Red Hat's release cycle."

This Thursday AlmaLinux also reiterated that they're "fully committed to delivering the best possible experience for the community, no matter where or what you run." And in an apparent move to beef up compatibility testing, they announced they'd be bringing openQA to the RHEL ecosystem. (They describe openQA as a tool using virtual machines that "simplifies automated testing of the whole installation process of an operating system in a wide combination of software and hardware configurations.")

An anonymous reader quotes a report from TorrentFreak: Italy's brand new anti-piracy law has just received full approval from telecoms regulator AGCOM. In a statement issued Thursday, AGCOM noted its position "at the forefront of the European scene in combating online piracy." The new law comes into force on August 8 and authorizes nationwide ISP blocking of live events and enables the state to issue fines of up to 5,000 euros to users of pirate streams .

In a statement published Thursday, AGCOM welcomed the amendments to Online Copyright Enforcement regulation 680/13/CONS, which concern measures to counter the illegal distribution of live sports streams, as laid out in Resolution 189/23/CONS. The new provisions grant AGCOM the power to issue "dynamic injunctions" against online service providers of all kinds, a privilege usually reserved for judges in Europe's highest courts. The aim is to streamline blocking measures against unlicensed IPTV services, with the goal of rendering them inaccessible across all of Italy.

"With such measures, it will be possible to disable access to pirated content in the first 30 minutes of the event broadcast by blocking DNS resolution of domain names and blocking the routing of network traffic to IP addresses uniquely intended for illicit activities," AGCOM says. "With this amendment, in perfect synchrony with the changes introduced by Parliament, AGCOM is once again at the forefront of the European scene in combating online piracy activity," says AGCOM Commissioner Massimiliano Capitanio.

A routine sequence of commands has triggered a 2-degree change in Voyager 2's antenna orientation, preventing the iconic spacecraft from receiving commands or transmitting data back to Earth, NASA announced earlier today. Mission controllers transmitted the commands to Voyager 2 on July 21. Gizmodo reports: Voyager 2, one of two twin probes launched in the 1970s to explore planets in the outer solar system, is located some 12.4 billion miles (19.9 billion kilometers) from Earth and is continually moving deeper into interstellar space. The glitch has disrupted the probe's ability to communicate with ground antennas operated by the Deep Space Network (DSN), and it's unable to receive commands from the mission team on Earth, NASA explained.

The communications pause is expected to be just that -- a pause. Voyager 2 is "programmed to reset its orientation multiple times each year to keep its antenna pointing at Earth," the space agency says. This procedure should -- fingers crossed -- re-establish the lost connection and allow routine communications to resume. The next reset is scheduled for October 15, which is 79 days from now. Undoubtedly, this will be 79 agonizing days for NASA and the Voyager team. Despite the current communication hiatus, the mission team remains confident that Voyager 2 will stay on its planned trajectory. Voyager 1, situated nearly 15 billion miles (24 billion kilometers) away from Earth, "continues to operate normally," NASA added.

sciencehabit shares a report from Science: Water is a near-universal solvent, able to dissolve substances ranging from limestone to the sugar in your coffee. That chemical superpower originates, oddly enough, in water's electrical properties. It can oppose and almost entirely cancel electric fields -- including attractions among dissolved ions that might otherwise pull them together. Curiously, dissolving salt in water weakens that electrical response. Now, a team of physicists has figured out exactly why this happens, using state-of-the-art computer simulations bolstered by artificial intelligence (AI).

'This is a fundamental property of water and one can finally do a calculation in which this can be entirely predicted from first principles,' says Roberto Car, a physicist at Princeton University who was not involved in the work. The AI-aided approach should allow physicists to probe in other settings, he says, such as batteries and fuel cells. [...] The results show that most of the salinity effect comes from the disruption of the clustering and correlations produced by hydrogen bonding, the team reports in a paper in press at Physical Review Letters. The researchers can pull out even more detail, explaining exactly how disruptions propagating through the network of water molecules make the dielectric constant vary with the salt concentration in a complex, nonlinear way.

"They can distinguish all the different contributions and identify which effect is dominant over the other," Car says. Yuki Nagata, a physicist at the Max Planck Institute for Polymer Research, says, "This is more or less conclusive." More important than this specific result may be the AI-based method, Nagata adds. It could be used for more practical problems, such as analyzing the interactions of water with membranes or surfaces. Zhang says she's doing just that, analyzing the splitting of water into hydrogen and oxygen along the surface of a titanium dioxide catalyst, one potential way to generate hydrogen for fuel.

Katie Malone writes via Engadget: "Tor" evokes an image of the dark web; a place to hire hitmen or buy drugs that, at this point, is overrun by feds trying to catch you in the act. The reality, however, is a lot more boring than that -- but it's also more secure. The Onion Router, now called Tor, is a privacy-focused web browser run by a nonprofit group. You can download it for free and use it to shop online or browse social media, just like you would on Chrome or Firefox or Safari, but with additional access to unlisted websites ending in .onion. This is what people think of as the "dark web," because the sites aren't indexed by search engines. But those sites aren't an inherently criminal endeavor.

"This is not a hacker tool," said Pavel Zoneff, director of strategic communications at The Tor Project. "It is a browser just as easy to use as any other browser that people are used to." That's right, despite common misconceptions, Tor can be used for any internet browsing you usually do. The key difference with Tor is that the network hides your IP address and other system information for full anonymity. This may sound familiar, because it's how a lot of people approach VPNs, but the difference is in the details. VPNs are just encrypted tunnels hiding your traffic from one hop to another. The company behind a VPN can still access your information, sell it or pass it along to law enforcement. With Tor, there's no link between you and your traffic, according to Jed Crandall, an associate professor at Arizona State University. Tor is built in the "higher layers" of the network and routes your traffic through separate tunnels, instead of a single encrypted tunnel. While the first tunnel may know some personal information and the last one may know the sites you visited, there is virtually nothing connecting those data points because your IP address and other identifying information are bounced from server to server into obscurity.

Accessing unindexed websites adds extra perks, like secure communication. While a platform like WhatsApp offers encrypted conversations, there could be traces that the conversation happened left on the device if it's ever investigated, according to Crandall. Tor's communication tunnels are secure and much harder to trace that the conversation ever happened. Other use cases may include keeping the identities of sensitive populations like undocumented immigrants anonymous, trying to unionize a workplace without the company shutting it down, victims of domestic violence looking for resources without their abuser finding out or, as Crandall said, wanting to make embarrassing Google searches without related targeted ads following you around forever.

Intel reported second-quarter earnings on Thursday, including a return to profitability after two straight quarters of losses, and a stronger-than-expected forecast. CNBC reports: For the third quarter, Intel expects earnings of $0.20 per share, adjusted, on revenue of $13.4 billion at the midpoint, versus analyst expectations of 16 cents per share on $13.23 billion in sales. Intel posted net income of $1.5 billion, or earnings of $0.35 per share, versus a net loss of $454 million, or a loss of 11 cents per share, in the same quarter last year.

Intel CFO David Zinsner said in a statement that part of the reason that Intel's report was stronger than expected was because of the progress it has made towards slashing $3 billion in costs this year. Earlier this year, Intel slashed its dividend and announced plans to save $10 billion per year by 2025, including through layoffs. Revenue fell to $12.9 billion from $15.3 billion a year ago, marking the sixth consecutive quarter of declining sales for the company.

Here's how Intel's business units performed:
- Intel's Client Computing group, which includes the company's laptop and desktop processor shipments, fell 12% annually to $6.8 billion.The overall PC market has been slumping for over a year.
- Intel's server chip division, which is reported as Data Center and AI, declined 15% to $4.0 billion in sales.
- Intel's Network and Edge division, which sells networking products for telecommunications, declined 28% to $1.4 billion.
- Mobileye, a publicly-traded Intel subsidiary focusing on self-driving cars, saw sales down 1% on an annual basis to $454 million.
- It reported $232 million in revenue for its foundry business, Intel Foundry Services, that makes chips for other companies.

An anonymous reader quotes a report from Ars Technica: When Google announced that trackers would be able to tie in to its 3 billion-device Bluetooth tracking network at its Google I/O 2023 conference, it also said that it would make it easier for people to avoid being tracked by trackers they don't know about, like Apple AirTags. Now Android users will soon get these "Unknown Tracker Alerts." Based on the joint specification developed by Google and Apple, and incorporating feedback from tracker-makers like Tile and Chipolo, the alerts currently work only with AirTags, but Google says it will work with tag manufacturers to expand its coverage.

For now, if an AirTag you don't own "is separated from its owner and determined to be traveling with you," a notification will tell you this and that "the owner of the tracker can see its location." Tapping the notification brings up a map tracing back to where it was first seen traveling with you. Google notes that this location data "is always encrypted and never shared with Google." Further into the prompts, you can make the tracker play a sound, "without the owner of the tracker knowing," Google says. If you bring the tracker to the back of your phone (presumably within NFC range), some trackers may provide their serial number and information about their owner, "like the last four digits of their phone number." Google indicates it will also link to information about how to physically disable a tracker. Finally, Google is offering a manual scan feature, if you're suspicious that your Android phone isn't catching a tracker or want to see what's nearby. The alerts are rolling out through a Google Play services update to devices on Android 6.0 and above over the coming weeks. Google is working to finish the joint tracking specification "by the end of this year."

The company added: "At this time, we've made the decision to hold the rollout of the Find My Device network until Apple has implemented protections for iOS."

LinkedIn appears to be developing a new AI tool that can help ease the effectively robotic task of looking for and applying to jobs. From a report: According to a new leak, the Microsoft-owned company seems to have a new "LinkedIn Coach" assistant in testing that could support you through the application processes, teach you new skills, and help you network on your LinkedIn network. The news comes from app researcher Nima Owji, who uncovers features from various developers that haven't been deployed yet. In an email, LinkedIn spokesperson Amanda Purvis tells The Verge the company is "always exploring" new ways to improve user experience on the platform. Purvis adds that the company "will have more to share soon."

OpenSignal blog: While eSIM adoption in the mobile market has been arriving for some time, Apple's move to make eSIM the only option for iPhone 14 range in the U.S. is propelling the worldwide shift towards eSIM technology. Opensignal's latest analysis reveals a significant surge in the proportion of users switching their operator among those who use an eSIM across seven examined markets -- Brazil, Indonesia, Singapore, South Korea, Taiwan, the U.K. and the U.S.

The switch from physical to embedded SIM cards threatens to alter how consumers switch operators and encourages operators to adopt new tactics to retain and acquire users, for example operators can offer network trials from within an app that provisions an eSIM immediately. eSIM also means the risks to operators of dual SIM devices that have long been common in many international markets are arriving in operator-controlled markets too, such as the U.S. and South Korea. Even on smartphones sold by operators, eSIM support is usually present in addition to a physical SIM, making them dual-SIM devices.

Google added eSIM-support to the Pixel range in 2017, Samsung added eSIM support to 2019's Galaxy S20 flagship. While Apple first added eSIM to their phones in 2018 with the iPhone Xs, it switched to selling exclusively eSIM models in the U.S. with the iPhone 14 range in late 2022. South Korea is also a special case -- eSIM support for domestic customers only began in mid-2022, before this point it was only available to international travelers. Notably, Samsung responded by introducing eSIM to a selection of its flagship devices in the home market, which had not been previously available there.

A team of Korean scientists claims to have created a room-temperature superconductor that also works at standard, ambient pressure. The work, however, is yet to be peer-reviewed. You can read their paper on Arxiv. Its abstract: For the first time in the world, we succeeded in synthesizing the room-temperature superconductor working at ambient pressure with a modified lead-apatite (LK-99) structure. The superconductivity of LK-99 is proved with the Critical temperature (TC), Zero-resistivity, Critical current (IC), Critical magnetic field (HC), and the Meissner effect. The superconductivity of LK-99 originates from minute structural distortion by a slight volume shrinkage (0.48 %), not by external factors such as temperature and pressure. The shrinkage is caused by Cu2+ substitution of Pb2+(2) ions in the insulating network of Pb(2)-phosphate and it generates the stress. It concurrently transfers to Pb(1) of the cylindrical column resulting in distortion of the cylindrical column interface, which creates superconducting quantum wells (SQWs) in the interface. The heat capacity results indicated that the new model is suitable for explaining the superconductivity of LK-99. The unique structure of LK-99 that allows the minute distorted structure to be maintained in the interfaces is the most important factor that LK-99 maintains and exhibits superconductivity at room temperatures and ambient pressure.

The Council of the EU has adopted new rules intended to make it much easier for EV owners to travel across Europe, while simultaneously helping to reduce the output of harmful greenhouse gases. From a report: The new regulation is set to benefit owners of electric cars and vans in three ways: It reduces range anxiety by expanding the EV charging infrastructure along Europe's main highways, it makes payments "at the pump" easier without requiring an app or subscription, and ensures pricing and availability is clearly communicated to avoid surprises. From 2025 onward, the new regulation requires fast charging stations offering at least 150kW of power to be installed every 60km (37mi) along the EU's Trans-European Transport Network, or (TEN-T) system of highways, the bloc's main transport corridor. The fast charging network along European highways is already pretty robust, I discovered on a recent 3,000km (2,000 mile) roadtrip with a VW ID Buzz. This new law could all but eliminate range anxiety for those sticking to TEN-T roads.

Google's newest proposed web standard is... DRM? Over the weekend the Internet got wind of this proposal for a "Web Environment Integrity API. " From a report: The explainer is authored by four Googlers, including at least one person on Chrome's "Privacy Sandbox" team, which is responding to the death of tracking cookies by building a user-tracking ad platform right into the browser. The intro to the Web Integrity API starts out: "Users often depend on websites trusting the client environment they run in. This trust may assume that the client environment is honest about certain aspects of itself, keeps user data and intellectual property secure, and is transparent about whether or not a human is using it."

The goal of the project is to learn more about the person on the other side of the web browser, ensuring they aren't a robot and that the browser hasn't been modified or tampered with in any unapproved ways. The intro says this data would be useful to advertisers to better count ad impressions, stop social network bots, enforce intellectual property rights, stop cheating in web games, and help financial transactions be more secure. Perhaps the most telling line of the explainer is that it "takes inspiration from existing native attestation signals such as [Apple's] App Attest and the [Android] Play Integrity API." Play Integrity (formerly called "SafetyNet") is an Android API that lets apps find out if your device has been rooted.

Root access allows you full control over the device that you purchased, and a lot of app developers don't like that. So if you root an Android phone and get flagged by the Android Integrity API, several types of apps will just refuse to run. You'll generally be locked out of banking apps, Google Wallet, online games, Snapchat, and some media apps like Netflix. [...] Google wants the same thing for the web. Google's plan is that, during a webpage transaction, the web server could require you to pass an "environment attestation" test before you get any data. At this point your browser would contact a "third-party" attestation server, and you would need to pass some kind of test. If you passed, you would get a signed "IntegrityToken" that verifies your environment is unmodified and points to the content you wanted unlocked. You bring this back to the web server, and if the server trusts the attestation company, you get the content unlocked and finally get a response with the data you wanted.

Forbes' senior writer on cybersecurity writes on the "warrantless monitoring of citizens en masse" in the United States.

Here's how county police armed with a "powerful new AI tool" identified the suspicious driving pattern of a grey Chevy owned by David Zayas: Searching through a database of 1.6 billion license plate records collected over the last two years from locations across New York State, the AI determined that Zayas' car was on a journey typical of a drug trafficker. According to a Department of Justice prosecutor filing, it made nine trips from Massachusetts to different parts of New York between October 2020 and August 2021 following routes known to be used by narcotics pushers and for conspicuously short stays. So on March 10 last year, Westchester PD pulled him over and searched his car, finding 112 grams of crack cocaine, a semiautomatic pistol and $34,000 in cash inside, according to court documents. A year later, Zayas pleaded guilty to a drug trafficking charge.

The previously unreported case is a window into the evolution of AI-powered policing, and a harbinger of the constitutional issues that will inevitably accompany it... Westchester PD's license plate surveillance system was built by Rekor, a $125 million market cap AI company trading on the NASDAQ. Local reporting and public government data reviewed by Forbes show Rekor has sold its ALPR tech to at least 23 police departments and local governments across America, from Lauderhill, Florida to San Diego, California. That's not including more than 40 police departments across New York state who can avail themselves of Westchester County PD's system, which runs out of its Real-Time Crime Center... It also runs the Rekor Public Safety Network, an opt-in project that has been aggregating vehicle location data from customers for the last three years, since it launched with information from 30 states that, at the time, were reading 150 million plates per month. That kind of centralized database with cross-state data sharing, has troubled civil rights activists, especially in light of recent revelations that Sacramento County Sheriff's Office was sharing license plate reader data with states that have banned abortion...

The ALPR market is growing thanks to a glut of Rekor rivals, including Flock, Motorola, Genetec, Jenoptik and many others who have contracts across federal and state governments. They're each trying to grab a slice of a market estimated to be worth at least $2.5 billion... In pursuit of that elusive profit, the market is looking beyond law enforcement to retail and fast food. Corporate giants have toyed with the idea of tying license plates to customer identities. McDonalds and White Castle have already begun using ALPR to tailor drive-through experiences, detecting returning customers and using past orders to guide them through the ordering process or offer individualized promotion offers. The latter restaurant chain uses Rekor tech to do that via a partnership with Mastercard.
A senior staff attorney at the ACLU tells Forbes that "The scale of this kind of surveillance is just incredibly massive."

Thanks to long-time Slashdot reader Geek_Cop for sharing the article.

Cerebras Systems and G42 have introduced the Condor Galaxy project, a network of nine interconnected supercomputers designed for AI model training with a combined performance of 36 FP16 ExaFLOPs. The first supercomputer, CG-1, located in California, offers 4 ExaFLOPs of FP16 performance and 54 million cores, focusing on Large Language Models and Generative AI without the need for complex distributed programming languages. AnandTech reports: CG-2 and CG-3 will be located in the U.S. and will follow in 2024. The remaining systems will be located across the globe and the total cost of the project will be over $900 million. The CG-1 supercomputer, situated in Santa Clara, California, combines 64 Cerebras CS-2 systems into a single user-friendly AI supercomputer, capable of providing 4 ExaFLOPs of dense, systolic FP16 compute for AI training. Based around Cerebras's 2.6 trillion transistor second-generation wafer scale engine processors, the machine is designed specifically for Large Language Models and Generative AI. It supports up to 600 billion parameter models, with configurations that can be expanded to support up to 100 trillion parameter models. Its 54 million AI-optimized compute cores and massivefabric network bandwidth of 388 Tb/s allow for nearly linear performance scaling from 1 to 64 CS-2 systems, according to Cerebras. The CG-1 supercomputer also offers inherent support for long sequence length training (up to 50,000 tokens) and does not require any complex distributed programming languages, which is common in case of GPU clusters.

This supercomputer is provided as a cloud service by Cerebras and G42 and since it is located in the U.S., Cerebras and G42 assert that it will not be used by hostile states. CG-1 is the first of three 4 FP16 ExaFLOP AI supercomputers (CG-1, CG-2, and CG-3) created by Cerebras and G42 in collaboration and located in the U.S. Once connected, these three AI supercomputers will form a 12 FP16 ExaFLOP, 162 million core distributed AI supercomputer, though it remains to be seen how efficient this network will be. In 2024, G42 and Cerebras plan to launch six additional Condor Galaxy supercomputers across the world, which will increase the total compute power to 36 FP16 ExaFLOPs delivered by 576 CS-2 systems. The Condor Galaxy project aims to democratize AI by offering sophisticated AI compute technology in the cloud. "Delivering 4 exaFLOPs of AI compute at FP16, CG-1 dramatically reduces AI training timelines while eliminating the pain of distributed compute," said Andrew Feldman, CEO of Cerebras Systems. "Many cloud companies have announced massive GPU clusters that cost billions of dollars to build, but that are extremely difficult to use. Distributing a single model over thousands of tiny GPUs takes months of time from dozens of people with rare expertise. CG-1 eliminates this challenge. Setting up a generative AI model takes minutes, not months and can be done by a single person. CG-1 is the first of three 4 ExaFLOP AI supercomputers to be deployed across the U.S. Over the next year, together with G42, we plan to expand this deployment and stand up a staggering 36 exaFLOPs of efficient, purpose-built AI compute."

An anonymous reader quotes a report from Ars Technica: AT&T's legacy telephone network may have nearly 200,000 miles of lead-covered cables, according to an estimate by AT&T submitted in a court filing. "Based on its records, AT&T estimates that lead-clad cables represent less than 10 percent of its copper footprint of roughly two million sheath miles of cable, the overwhelming majority of which remains in active service," AT&T wrote in a court filing yesterday in US District Court for the Eastern District of California. "More than two thirds of its lead-clad cabling is either buried or in conduit, followed by aerial cable, and with a very small portion running underwater. There are varying costs of installation, maintenance, and removal by cable type (aerial, buried, buried in conduit, underwater)."

Reacting to the court filing, financial analyst firm Raymond James & Associates wrote in a research note, "AT&T is telling us that the total exposure is 200,000 route miles or less." With about two-thirds of the lead cables either buried or installed inside conduit, "We believe the implication for AT&T's data is that the route miles that should be addressed most immediately is about 3.3 percent (or less)," the analyst firm wrote. AT&T's new court filing came in a case filed against AT&T subsidiary Pacific Bell by the California Sportfishing Protection Alliance (CSPA) in January 2021. The sportfishing group sued AT&T over cables that are allegedly "damaged and discharging lead into Lake Tahoe."

The two underwater cables run along the bottom of the western side of Lake Tahoe for a total of eight miles. AT&T "contends that it stopped using the Cables in or around the 1980s or earlier, that the Easements therefore have terminated, and that Defendant no longer owns the Cables," according to a November 2021 settlement. AT&T agreed in that settlement to remove the cables but now says it is at an "impasse" with the CSPA regarding removal. "In this matter, AT&T has always maintained that its lead-clad telecommunications cables pose no danger to those who work and play in the waters of Lake Tahoe, but in 2021, AT&T agreed to remove them simply to avoid the expense of litigation," an AT&T lawyer at the firm Paul Hastings wrote yesterday in a letter to the plaintiff that was attached to the court filing. [...]

AT&T's stance that it won't remove the Lake Tahoe cables any time soon is apparently a surprise to the plaintiff. The CSPA said in a court filing last week that in a Zoom meeting on July 10, "AT&T confirmed that it is prepared to commence the removal process on September 6, 2023, as long as the new permit request that AT&T submitted to State Parks in May is approved by State Park." AT&T's filing said the company never "confirmed" that it is prepared to start the cable removal process on September 6. The CSPA argues that the lead-covered cables "have leached, are leaching, and will continue to leach lead into the waters of Lake Tahoe, and that such leaching may present an imminent and substantial endangerment to human health or the environment." Last week, the Wall Street Journal published an investigative report that found evidence of more than 2,000 lead-covered telephone cables installed across the U.S. Teleco stock prices plummeted as a result, since the remediation could cost the telecom industry $60 billion.

While members of Congress are putting pressure on telecom companies to act, AT&T does appear to be taking new actions as a result of the investigation. "AT&T is working with union partners to add a voluntary testing program for any employee who works with or has worked with lead-clad cables," which "expands on AT&T's previous practice of providing blood-lead testing for technicians involved in lead-clad cable removal," the company said in its letter to the plaintiff yesterday. AT&T said it is also conducting new testing and site visits at certain locations.

AT&T's stock slid to a 30-year low following the news.

The U.S. Federal Reserve has launched a long-awaited service which will aim to modernize the country's payment system by eventually allowing everyday Americans to send and receive funds in seconds, 24 hours a day, seven days a week, the central bank announced on Thursday. From a report: The "FedNow" service, which has been in the works since 2019, will seek to eliminate the several-day lag it commonly takes cash transfers to settle, bringing the U.S. in line with countries including the United Kingdom, India, Brazil, as well as the European Union, where similar services have existed for years. FedNow is launching with 41 banks and 15 service providers certified to use the service, including community banks and large lenders like JPMorgan Chase, Bank of New York Mellon, and US Bancorp, but the Fed plans to onboard more banks and credit unions this year.

The Fed said on Thursday in a statement that 35 banks and credit unions were currently utilizing the service, as well as the Treasury Department's Bureau of Fiscal Service. The service will compete with private sector real-time payments systems, including The Clearing House's RTP network, and was initially opposed by big banks who said it was redundant. But many have since agreed to participate on the basis FedNow will allow them to expand the services they can offer clients. "For us, FedNow really is a wonderful way of expanding reach," said Anu Somani, head of global payables and embedded payments at U.S. Bank. Unlike peer-to-peer payments services like Venmo or PayPal, which act as intermediaries between banks, payments made via FedNow will settle directly in central bank accounts.