In mid-2021 Ameria's National Security Advisor set up a new directorate focused on "advanced chips, quantum computing, and other cutting-edge tech," reports Wired. And the next year as Congress was working on boosting America's semiconductor sector, he was "closing in on a plan to cripple China's... In October 2022, the Commerce Department forged ahead with its new export controls."

So what happened next? In a phone call with President Biden this past spring, Xi Jinping warned that if the US continued trying to stall China's technological development, he would not "sit back and watch." And he hasn't. Already, China has answered the US export controls — and its corresponding deals with other countries — by imposing its own restrictions on critical minerals used to make semiconductors and by hoovering up older chips and manufacturing equipment it is still allowed to buy. For the past several quarters, in fact, China was the top customer for ASML and a number of Japanese chip companies. A robust black market for banned chips has also emerged in China. According to a recent New York Times investigation, some of the Chinese companies that have been barred from accessing American chips through US export controls have set up new corporations to evade those bans. (These companies have claimed no connection to the ones who've been banned.) This has reportedly enabled Chinese entities with ties to the military to obtain small amounts of Nvidia's high-powered chips.

Nvidia, meanwhile, has responded to the US actions by developing new China-specific chips that don't run afoul of the US controls but don't exactly thrill the Biden administration either. For the White House and Commerce Department, keeping pace with all of these workarounds has been a constant game of cat and mouse. In 2023, the US introduced the first round of updates to its export controls. This September, it released another — an announcement that was quickly followed by a similar expansion of controls by the Dutch. Some observers have speculated that the Biden administration's actions have only made China more determined to invest in its advanced tech sector.

And there's clearly some truth to that. But it's also true that China has been trying to become self-sufficient since long before Biden entered office. Since 2014, it has plowed nearly $100 billion into its domestic chip sector. "That was the world we walked into," [NSA Advisor Jake] Sullivan said. "Not the world we created through our export controls." The United States' actions, he argues, have only made accomplishing that mission that much tougher and costlier for Beijing. Intel CEO Pat Gelsinger estimated earlier this year that there's a "10-year gap" between the most powerful chips being made by Chinese chipmakers like SMIC and the ones Intel and Nvidia are working on, thanks in part to the export controls.

If the measure of Sullivan's success is how effectively the United States has constrained China's advancement, it's hard to argue with the evidence. "It's probably one of the biggest achievements of the entire Biden administration," said Martijn Rasser, managing director of Datenna, a leading intelligence firm focused on China. Rasser said the impact of the US export controls alone "will endure for decades." But if you're judging Sullivan's success by his more idealistic promises regarding the future of technology — the idea that the US can usher in an era of progress dominated by democratic values — well, that's a far tougher test. In many ways, the world, and the way advanced technologies are poised to shape it, feels more unsettled than ever.

Four years was always going to be too short for Sullivan to deliver on that promise. The question is whether whoever's sitting in Sullivan's seat next will pick up where he left off.

The PC market is not showing many signs of a rebound, despite the hype around AI PCs, with market watchers split over whether unit shipments are up or down slightly. From a report: Those magical AI PC boxes were supposed to fire up buyer enthusiasm and spur the somewhat listless market for desktop and laptop systems into significant growth territory, but that doesn't appear to be happening. According to the latest figures from Gartner, global PC shipments totaled 62.9 million units during Q3 of this year, representing a 1.3 percent decline compared with the same period last year. However, this does follow three consecutive quarters of modest growth.

"Even with a full line-up of Windows-based AI PCs for both Arm and x86 in the third quarter of 2024, AI PCs did not boost the demand for PCs since buyers have yet to see their clear benefits or business value," commented Gartner Director Analyst Mikako Kitagawa. This is perhaps understandable when AI PCs are largely just a marketing concept, and vendors can't agree on exactly what the the definition of an AI PC should be. Even worse, some buyers of Arm-based Copilot+ machines discovered that their performance isn't actually very good with some applications.

Ecovacs robot vacuums have been hacked across the U.S. to shout racial slurs at unsuspecting people. VICE News reports: The issue is specifically with Ecovacs' Deebot X2 model. The hackers gained control of the devices and used the onboard speakers to blast racial slurs at anyone within earshot. One such person was a lawyer from Minnesota named Daniel Swenson. He was watching TV when he heard some odd noises coming from the direction of his vacuum. He changed the password and restarted it. But then the odd sounds started up again. And then it started shouting racial slurs at him like a surly disgruntled maid.

There were multiple reports of similar incidents across the United States and around the same time. One of them happened in Los Angeles, where a vacuum chased a dog while spewing hate. Another happened in El Paso, where the vac spewed slurs until it's owner turned it off. The attacks are apparently quite easy to pull off thanks to several known security vulnerabilities in Ecovacs, like a bad Bluetooth connector and a defective PIN system that is intended to safeguard video feeds and remote access but actually doesn't do any of that at all. A pair of cybersecurity researchers released a report on Ecovacs detailing the brand's multiple security flaws earlier this year.

At Tesla's "We, Robot" event at Warner Bros. Studios tonight, Elon Musk unveiled the Tesla Cybercab, Robovan, and an updated version of the Optimus robot. Slashdot is at the event capturing photos and getting demos of everything announced. You can follow along on X. Below is a summary of each of the offerings.

Tesla Cybercab: The Tesla Cybercab is a futuristic, fully autonomous robotaxi designed without a steering wheel or pedals, positioned to revolutionize mass transit with extremely low operating costs. It features a sleek design with upward-opening butterfly doors and a compact cabin that seats two passengers. Musk said the Cybercab uses inductive charging instead of a traditional plug-in. "Something we're also doing is and it's really high time we did this is inductive charging. So the robotaxi has no plug it just goes over the inductive charger and charges so yeah, it's kind of how it should be." The vehicle is expected to cost under $30,000. Regulatory approval will be needed before it can go into production, which is projected to begin by 2026 or 2027. Tesla Robovan: The Tesla Robovan is a dustbuster-shaped electric passenger van featuring sliding glass doors, a bright interior, and carriage-style seating for up to 20 passengers. "One of the things we want to do and we've seen this with the CyberTruck is we want to change the look of the roads the future should look like the future," said Musk. Musk also claimed that autonomy will "turn parking lots into parks," as fewer cars will be needed and they won't sit idle for most of the day. Pricing and release details were not disclosed. Tesla Optimus: The updated Tesla Optimus robot is a humanoid designed to handle everyday tasks, such as retrieving packages or serving drinks. Optimus walked on stage and interacted with attendees, though its current capabilities are still limited. Elon Musk envisions the robot as a transformative product, with plans to produce millions of units at a price of around $20,000. "It'll be able to do anything you want. So it can be a teacher, babysit your kids, it can walk your dog, mow your lawn, get the groceries, just be your friend, serve drinks. Whatever you can think of, it will do." Optimus is expected to start performing useful tasks by the end of the year, with broader availability projected by the end of next year. In closing, Musk said: "I think this will be the biggest product ever of any kind. Because I think everyone of the 8 billion people of Earth, I think everyone's going to want their Optimus buddy." Developing...

Researchers at BitEnergy AI, Inc. have developed Linear-Complexity Multiplication (L-Mul), a technique that reduces AI model power consumption by up to 95% by replacing energy-intensive floating-point multiplications with simpler integer additions. This method promises significant energy savings without compromising accuracy, but it requires specialized hardware to fully realize its benefits. Decrypt reports: L-Mul tackles the AI energy problem head-on by reimagining how AI models handle calculations. Instead of complex floating-point multiplications, L-Mul approximates these operations using integer additions. So, for example, instead of multiplying 123.45 by 67.89, L-Mul breaks it down into smaller, easier steps using addition. This makes the calculations faster and uses less energy, while still maintaining accuracy. The results seem promising. "Applying the L-Mul operation in tensor processing hardware can potentially reduce 95% energy cost by element wise floating point tensor multiplications and 80% energy cost of dot products," the researchers claim. Without getting overly complicated, what that means is simply this: If a model used this technique, it would require 95% less energy to think, and 80% less energy to come up with new ideas, according to this research.

The algorithm's impact extends beyond energy savings. L-Mul outperforms current 8-bit standards in some cases, achieving higher precision while using significantly less bit-level computation. Tests across natural language processing, vision tasks, and symbolic reasoning showed an average performance drop of just 0.07% -- a negligible tradeoff for the potential energy savings. Transformer-based models, the backbone of large language models like GPT, could benefit greatly from L-Mul. The algorithm seamlessly integrates into the attention mechanism, a computationally intensive part of these models. Tests on popular models such as Llama, Mistral, and Gemma even revealed some accuracy gain on certain vision tasks.

At an operational level, L-Mul's advantages become even clearer. The research shows that multiplying two float8 numbers (the way AI models would operate today) requires 325 operations, while L-Mul uses only 157 -- less than half. "To summarize the error and complexity analysis, L-Mul is both more efficient and more accurate than fp8 multiplication," the study concludes. But nothing is perfect and this technique has a major achilles heel: It requires a special type of hardware, so the current hardware isn't optimized to take full advantage of it. Plans for specialized hardware that natively supports L-Mul calculations may be already in motion. "To unlock the full potential of our proposed method, we will implement the L-Mul and L-Matmul kernel algorithms on hardware level and develop programming APIs for high-level model design," the researchers say.

Global semiconductor sales recorded a 20.6% year-on-year increase in August to $53.1 billion, according to the Semiconductor Industry Association (SIA). The Register reports: The Americas led the way, with sales up 43.9 percent to $15.4 billion over last year to notch up what may be the highest on record for August, the SIA said. This comes on the back of swelling demand from sectors such as AI, cloud computing, and automotive. Over in Asia-Pacific sales grew year-on-year by 17.1 percent to $10.95 billion, according to the World Semiconductor Trade Statistics organization, which compiles these stats for the SIA. China was up 19.2 percent to $13 billion and Japan grew two percent to $4 billion.

Europe was the outlier, recording a nine percent drop to $4.7 billion. No reason was given for this decline. However, on a worldwide basis, all continents returned positive month-on-month numbers in August for the first time since October 2023, indicating that the semiconductor industry is on a path to recovery.

Tech giants Amazon and Microsoft have inked major deals with U.S. nuclear power plants to fuel their energy-hungry data centers, marking a shift in the industry's power sourcing strategy. The move comes as AI-driven facilities strain companies' climate goals, pushing them towards carbon-free electricity sources.

Microsoft plans to revive the shuttered Three Mile Island plant by 2028, while Amazon secured power from Pennsylvania's Susquehanna Nuclear facility in a $650 million deal. Google is also exploring nuclear options, including small modular reactors still under development. This trend could potentially triple U.S. nuclear capacity by 2050, according to a Department of Energy report.

"China Telecom, one of the largest wireless carriers in mainland China, says that it has developed two large language models (LLMs) relying solely on domestically manufactured AI chips..." reports Tom's Hardware. "If the information is accurate, this is a crucial milestone in China's attempt at becoming independent of other countries for its semiconductor needs, especially as the U.S. is increasingly tightening and banning the supply of the latest, highest-end chips for Beijing in the U.S.-China chip war." Huawei, which has mostly been banned from the U.S. and other allied countries, is one of the leaders in China's local chip industry... If China Telecom's LLMs were indeed fully trained using Huawei chips alone, then this would be a massive success for Huawei and the Chinese government.
The project's GitHub page "contains a hint about how China Telecom may have trained the model," reports the Register, "in a mention of compatibility with the 'Ascend Atlas 800T A2 training server' — a Huawei product listed as supporting the Kunpeng 920 7265 or Kunpeng 920 5250 processors, respectively running 64 cores at 3.0GHz and 48 cores at 2.6GHz. Huawei builds those processors using the Arm 8.2 architecture and bills them as produced with a 7nm process."

The South China Morning Post says the unnamed model has 1 trillion parameters, according to China Telecom, while the TeleChat2t-115B model has over 100 billion parameters.

Thanks to long-time Slashdot reader hackingbear for sharing the news.

Days after a hurricane struck America's southeast, Florida's state's fire marshall "confirmed 16 lithium-ion battery fires related to storm surge," according to local news reports. "Officials said six of those fires are associated with electric vehicles and they are working with fire departments statewide to gather more data." (Earlier this year America's federal transportation safety agency estimated that after a 2022 hurricane "about 36 EVs caught on fire. In several instances, the fire erupted while the impacted EVs were being towed on their flatbed trailers.")

But Tuesday, when over 1 million Americans were without electricity, the Atlantic pointed out the other side of the story. "EV owners are using their cars to keep the lights on." When Hurricane Helene knocked out the power in Charlotte, North Carolina, on Friday, Dustin Baker, like many other people across the Southeast, turned to a backup power source. His just happened to be an electric pickup truck. Over the weekend, Baker ran extension cords from the back of his Ford F-150 Lightning, using the truck's battery to keep his refrigerator and freezer running. It worked so well that Baker became an energy Good Samaritan. "I ran another extension cord to my neighbor so they could run two refrigerators they have," he told me.

Americans in hurricane territory have long kept diesel-powered generators as a way of life, but electric cars are a leap forward. An EV, at its most fundamental level, is just a big battery on wheels that can be used to power anything, not only the car itself. Some EVs pack enough juice to power a whole home for several days, or a few appliances for even longer. In the aftermath of Helene, as millions of Americans were left without power, many EV owners did just that. A vet clinic that had lost power used an electric F-150 to keep its medicines cold and continue seeing patients during the blackout. One Tesla Cybertruck owner used his car to power his home after his entire neighborhood lost power.
One Louisiana man just ran cords straight from the outlets in the bed of his Tesla Cybertruck, according to the article. "We were able to run my internet router and TV, [plus] lamps, refrigerator, a window AC unit, and fans, as well as several phone, watch, and laptop chargers." Over the course of about 24 hours, he said, all of this activity ran his Cybertruck battery down from 99 percent to 80 percent...

Bidirectional charging may prove to be the secret weapon that sells electrification to the South, which has generally remained far behind the West and the Northeast in electric-vehicle purchases. If EVs become widely seen as the best option for blackouts, they could entice not just the climate conscious but also the suburban dads in hurricane country with a core belief in prepping for anything. It will take a lot to overcome the widespread distrust of EVs and anxiety about a new technology, but our loathing of power outages just might do the trick.
The article notes that Tesla has confirmed all its electric vehicles will support bidirectional charging by 2025.

An anonymous reader quotes a report from TechCrunch: The European Union's top court has sided with a privacy challenge to Meta's data retention policies. It ruled on Friday that social networks, such as Facebook, cannot keep using people's information for ad targeting indefinitely. The judgement could have major implications on the way Meta and other ad-funded social networks operate in the region. Limits on how long personal data can be kept must be applied in order to comply with data minimization principles contained in the bloc's General Data Protection Regulation (GDPR). Breaches of the regime can lead to fines of up to 4% of global annual turnover -- which, in Meta's case, could put it on the hook for billions more in penalties (NB: it is already at the top of the leaderboard of Big Tech GDPR breachers). [...]

The original challenge to Meta's ad business dates back to 2014 but was not fully heard in Austria until 2020, per noyb. The Austrian supreme court then referred several legal questions to the CJEU in 2021. Some were answered via a separate challenge to Meta/Facebook, in a July 2023 CJEU ruling -- which struck down the company's ability to claim a "legitimate interest" to process people's data for ads. The remaining two questions have now been dealt with by the CJEU. And it's more bad news for Meta's surveillance-based ad business. Limits do apply. Summarizing this component of the judgement in a press release, the CJEU wrote: "An online social network such as Facebook cannot use all of the personal data obtained for the purposes of targeted advertising, without restriction as to time and without distinction as to type of data."

The ruling looks important on account of how ads businesses, such as Meta's, function. Crudely put, the more of your data they can grab, the better -- as far as they are concerned. Back in 2022, an internal memo penned by Meta engineers which was obtained by Vice's Motherboard likened its data collection practices to tipping bottles of ink into a vast lake and suggested the company's aggregation of personal data lacked controls and did not lend itself to being able to silo different types of data or apply data retention limits. Although Meta claimed at the time that the document "does not describe our extensive processes and controls to comply with privacy regulations." How exactly the adtech giant will need to amend its data retention practices following the CJEU ruling remains to be seen. But the law is clear that it must have limits. "[Advertising] companies must develop data management protocols to gradually delete unneeded data or stop using them," noyb suggests. The court also weighed in a second question that concerns sensitive data that has been "manifestly made public" by the data subject, "and whether sensitive characteristics could be used for ad targeting because of that," reports TechCrunch. "The court ruled that it could not, maintaining the GDPR's purpose limitation principle."

An anonymous reader shares a report: The arrival of energy-assisted magnetic recording (EAMR) technologies like Seagate's HAMR will play a crucial role in accelerating HDD capacity growth in the coming years. According to the new IEEE International Roadmap for Devices and Systems Mass Data Storage, we will see 60 TB hard disk drives in 2028. If the prediction is accurate, we will see HDD storage capacity doubling in just four years, something that did not happen for a while. Also, IEEE believes that HDD unit sales will increase.

IEEE's latest HDD development roadmap spans 2022 to 2037 and covers 15 years of hard drive evolution. The arrival of HAMR in 2024 will play a pivotal role in the increase in HDD capacity (even though Western Digital has managed to stay competitive with Seagate's HAMR HDDs using a set of its technologies) over the next few years. IEEE engineers expect HDDs to leapfrog to 40TB in 2025 and 60TB in 2028, doubling capacity from 30TB in 2024. By 2037, there will be 100TB of storage space, according to IEEE.

To get to those extreme capacities, HDD makers will have to increase the areal density of their platters steadily. To get to 40TB per drive, they will have to get to 2 TB/inch^2 in 2025 and then to over 4 TB/inch^2 in 2028 to build 60TB HDDs. By 2037, areal density will grow to over 10 Tb/inch^2. Increasing areal density will necessitate the use of new media, magnetic films, and all-new write and read heads.

New submitter ae4ax writes: North American buyers of JuiceBox EVSEs (chargers) received an email today declaring the imminent closure of Enel X Way USA, LLC, the maintainers of the software infrastructure behind their EVSEs. Customer support has already shut down, and apps will be deactivated and removed by October 11, 2024. The company claims economic headwinds from lackluster EV sales and high interest rates as the motivation for the closure. Enel X Way properties outside North America are not affected, they say. "An experienced third-party firm will be appointed to manage the company's affairs and ensure that the closure is handled with the utmost care and professionalism," the company said in a statement. "The appointed firm will be responsible for managing the remaining obligations and communicating directly with customers and partners regarding the closure."

Customers will still be able to charge vehicles but all their connectivity features -- the Enel X Way app and all other Enel e-mobility apps in North America -- will stop working. Commercial charging stations will also lose functionality. "So If you own a JuiceBox, you just got nine days' warning that your home charger can no longer be configured," reports Electrek.

Electrek's Michael Bower, who uses a JuiceBox to charge his Chevy Bolt, said: "I'm disappointed that Enel X Way is removing their apps -- and thus the ability to change the amperage -- for their EVSEs. I live in a condo with a 100A panel, so the ability to lower the amperage from 40 to 32 or 16 was beneficial when charging my EV while drawing power for laundry or the central A/C in the summer. It just shows how 'smart' EVSEs are too reliant on their respective apps."

California Governor Gavin Newsom signed a law giving the California Energy Commission the authority to require bidirectional charging in electric vehicles (EVs) in the future -- although no timeline is set. Bidirectional charging allows EVs to not only charge from the grid but also supply electricity back to the grid, potentially enhancing grid resiliency, supporting renewable energy, and reducing peak electricity demand. Electrek reports: The idea started in 2023 when state Senator Nancy Skinner introduced a bill which would require EVs to have bidirectional charging by 2027. As this bill made its way through the legislative process, it got watered down from that ambitious timeline. So the current form of the bill, which is now called SB 59, took away that timeline and instead gave the California Energy Commission (CEC) the go-ahead to issue a requirement whenever they see it fit. The bill directs the CEC, the California Air Resources Board, and the California Public Utilities Commission to examine the use cases of bidirectional charging and give them the power to require specific weight classes of EVs to be bidirectional-capable if a compelling use case exists.

The state already estimates that integrating EVs into the grid could save $1 billion in costs annually, so there's definitely a use case there, but the question is the cost and immediacy of building those vehicles into the grid. The reason this can't be done immediately is that cars take time to design, and while adding bidirectional charging to an EV isn't the most difficult process, it also only really becomes useful with a whole ecosystem of services around the vehicle.

And that ecosystem has been a bit of a hard sell so far. It's all well and good to tell someone they can make $500/year by selling energy to the grid, but then you have to convince them to buy a more expensive charging unit and keep their car plugged in all the time, with someone else managing its energy storage. Some consumers might push back against that, so part of CEC's job is to wait to pull the trigger until it becomes apparent that people are actually interested in the end-user use case for V2G -- otherwise, no sense in requiring a feature that nobody is going to use.

Google is introducing a new "Quick Insert" button on Chromebooks, offering contextual AI tools across the operating system. The feature debuts on Samsung's Galaxy Chromebook Plus, replacing the traditional Caps Lock key. Older Chromebooks can access Quick Insert via a keyboard shortcut. The button opens an overlay providing access to emojis, GIFs, Google's Help Me Write AI feature, and recent web links. Future updates will include AI-generated image creation.

Google is also rolling out new AI features to Chromebook Plus devices, including automatic transcription, real-time translation, and voice isolation for video calls. Standard Chromebooks will receive updates like Welcome Recap and Focus mode. Lenovo and Samsung are launching new Chromebook models to coincide with these software updates. The Lenovo Duet, a detachable 2-in-1, features an 11-inch 2K screen and starts at $349. Samsung's Galaxy Chromebook Plus boasts a 15.6-inch OLED display in a lightweight 2.58-pound package.

An anonymous reader quotes a report from TechCrunch: Raspberry Pi, the company that sells tiny, cheap, single-board computers, is releasing an add-on that is going to open up several use cases -- and yes, because it's 2024, there's an AI angle. Called the Raspberry Pi AI Camera, this image sensor comes with on-board AI processing and is going to cost $70. In more technical terms, the AI Camera is based on a Sony image sensor (the IMX500) paired with a RP2040, Raspberry Pi's own microcontroller chip with on-chip SRAM. Like the rest of the line-up, the RP2040 follows Raspberry Pi's overall philosophy -- it is inexpensive yet efficient. In other words, AI startups aren't going to replace their Nvidia GPUs with RP2040 chips for inference. But when you pair it with an image sensor, you get an extension module that can capture images and process those images through common neural network models. As an added benefit, on-board processing on the camera module means that the host Raspberry Pi isn't affected by visual data processing. The Raspberry Pi remains free to perform other operations -- you don't need to add a separate accelerator. The new module is compatible with all Raspberry Pi computers.

This isn't Raspberry Pi's first camera module. The company still sells the Raspberry Pi Camera Module 3, a simple 12-megapixel image sensor from Sony (IMX708) mounted on a small add-on board that you can pair with a Raspberry Pi with a ribbon cable. As Raspberry Pi promises to keep production running for many years, the Camera Module 3 will remain available for around $25. The AI Camera is the same size as the Camera Module 3 (25mm x 24mm) but slightly thicker due to the structure of the optical sensor. It comes pre-loaded with the MobileNet-SSD model, an object detection model that can run in realtime.

U.S. real estate developers "are having a hard time keeping up with demand," reports the Los Angeles Times, "as businesses in search of secure spots for their servers rent nearly every square foot that becomes available..." Construction of new data centers is at "extraordinary levels" driven by "insatiable demand," a recent report on the industry by real estate brokerage JLL found. "Never in my career of 25 years in real estate have I seen demand like this on a global scale," said JLL real estate broker Darren Eades, who specializes in data centers...

The biggest drivers are AI and cloud service providers that include some of the biggest names in tech, such as Amazon, Microsoft, Google and Oracle. With occupancy in conventional office buildings still down sharply following the impact of the COVID-19 pandemic and property values falling, data centers represent a rare ripe opportunity for real estate developers, who are pursuing opportunities in major markets like Los Angeles and less urban locales that are served by plentiful and preferably cheap power needed to run data centers. "If you can find a cluster of power to build a site, they'll come," Eades said of developers. Construction is taking place at an "extraordinary" pace nationwide and still not keeping up, the JLL data center report said. [Data center] "Vacancy declined to a record low of 3% at midyear due to insatiable demand and despite rampant construction."

Development increased more than sevenfold in two years, with the pipeline of new projects leveling off in the first half of 2024, a potential signal that the U.S. power grid cannot support development at a faster pace. But when projects currently under construction or planned are complete, the U.S. colocation market, in which businesses rent space in a data center owned by another company for their servers and other computing hardware, will triple in size from current levels... Real estate investors and landlords are being drawn into the market because demand from tenants is high and they are likely to renew their leases after shouldering the costs of setting up data centers. "They invest in their space and in your space and they tend to stick around longer," said Mark Messana, president of Downtown Properties, which owns offices in Los Angeles and San Francisco. "As we all know, the office market is struggling a little bit, so it's nice to be able to have some data customers in the mix..."

Power demand for computing is growing so intense that it threatens to strain the nation's electrical grid, sending users to remote locations where power is plentiful and preferably cheap. Data center developers are working in Alabama, the Dakotas and Indiana, "traditionally states that wouldn't have data centers," Eades said.
The article includes "the mother of all data centers" in the western U.S. — a 30-story building where "thousands of miles of undersea fiber-optic cables disappear into an ordinary-looking office tower." Once a prestigious location for businesses, "The recent departure of a law firm that had been in the building more than 50 years cleared out five floors that will quickly be re-leased to data tenants, said Eades, who represents the landlord..."

To retrofit the building for data centers, "two elevators were removed so the empty shafts could hold water pipes used to help keep the temperature cool enough for the heat-producing servers" — and developers are happy rents "can be double what they are at newer downtown office high-rises, according to real estate data provider CoStar...

"By 2030, data centers could account for as much as 11% of U.S. power demand — up from 3% now, according to analysts at Goldman Sachs."

"For the first time, scientists have created a flexible programmable chip that is not made of silicon..." reports IEEE Spectrum — opening new possibilities for implantable devices, on-skin computers, brain-machine interfaces, and soft robotics.

U.K.-based Pragmatic Semiconductor produced an "ultralow-power" 32-bit microprocessor, according to the article, and "The microchip's open-source RISC-V architecture suggests it might cost less than a dollar..." This shows potential for inexpensive applications like wearable healthcare electronics and smart package labels, according to the chip's inventors: For example, "we can develop an ECG patch that has flexible electrodes attached to the chest and a flexible microprocessor connected to flexible electrodes to classify arrhythmia conditions by processing the ECG data from a patient," says Emre Ozer, senior director of processor development at Pragmatic, a flexible chip manufacturer in Cambridge, England. Detecting normal heart rhythms versus an arrhythmia "is a machine learning task that can run in software in the flexible microprocessor," he says...

Pragmatic sought to create a flexible microchip that cost significantly less to make than a silicon processor. The new device, named Flex-RV, is a 32-bit microprocessor based on the metal-oxide semiconductor indium gallium zinc oxide (IGZO). Attempts to create flexible devices from silicon require special packaging for the brittle microchips to protect them from the mechanical stresses of bending and stretching. In contrast, pliable thin-film transistors made from IGZO can be made directly at low temperatures onto flexible plastics, leading to lower costs...

"Our end goal is to democratize computing by developing a license-free microprocessor," Ozer says... Other processors have been built using flexible semiconductors, such as Pragmatic's 32-bit PlasticARM and an ultracheap microcontroller designed by engineers in Illinois. Unlike these earlier devices, Flex-RV is programmable and can run compiled programs written in high-level languages such as C. In addition, the open-source nature of RISC-V also let the researchers equip Flex-RV with a programmable machine learning hardware accelerator, enabling artificial intelligence applications.

Each Flex-RV microprocessor has a 17.5 square millimeter core and roughly 12,600 logic gates. The research team found Flex-RV could run as fast as 60 kilohertz while consuming less than 6 milliwatts of power... The Pragmatic team found that Flex-RV could still execute programs correctly when bent to a curve with a radius of 3 millimeters. Performance varied between a 4.3 percent slowdown to a 2.3 percent speedup depending on the way it was bent.

An anonymous reader quotes a report from PCWorld, written by Michael Crider: The latest perpetrator of questionable AI branding? HP. The company is introducing "Print AI," what it calls the "industry's first intelligent print experience for home, office, and large format printing." What does that mean? It's essentially a new beta software driver package for some HP printers. According to the press release, it can deliver "Perfect Output" -- capital P capital O -- a branded tool that reformats the contents of a page in order to more ideally fit it onto physical paper.

Despite my skeptical tone, this is actually a pretty cool idea. "Perfect Output can detect unwanted content like ads and web text, printing only the desired text and images, saving time, paper, and ink." That's neat! If the web page you're printing doesn't offer a built-in print format, the software will make one for you. It'll also serve to better organize printed spreadsheets and images, too. But I don't see anything in this software that's actually AI -- or even machine learning, for that matter. This is applying the same tech (functionally, if not necessarily the same code) as the "reader mode" formatting we've seen in browsers for about a decade now. Take the text and images of a page, strip out everything else that's unnecessary, and present it as efficiently as possible. [...]

The press release does mention that support and formatting tasks can be accomplished with "simple conversational prompts," which at least might be leveraging some of the large language models that have become synonymous with AI as consumers understand it. But based on the description, it's more about selling you something than helping you. "Customers can choose to print or explore a curated list of partners that offer unique photo printing capabilities, gift certificates to be printed on the card, and so much more." Whoopee.

An anonymous reader quotes a report from 404 Media: After a horseback riding accident left him paralyzed from the waist down in 2009, former jockey Michael Straight learned to walk again with the help of a $100,000 ReWalk Personal exoskeleton. Earlier this month, that exoskeleton broke because of a malfunctioning piece of wiring in an accompanying watch that makes the exoskeleton work. The manufacturer refused to fix it, saying the machine was now too old to be serviced, and Straight once again couldn't walk anymore. "After 371,091 steps my exoskeleton is being retired after 10 years of unbelievable physical therapy," Straight posted on Facebook on September 16. "The reasons [sic] why it has stopped is a pathetic excuse for a bad company to try and make more money. The reason it stopped is because of a battery in the watch I wear to operate the machine. I called thinking it was no big deal, yet I was told they stopped working on any machine that was 5 years or older. I find it very hard to believe after paying nearly $100,000 for the machine and training that a $20 battery for the watch is the reason I can't walk anymore?"

Straight's experience is a nightmare scenario that highlights what happens when companies decide to stop supporting their products and do not actively support independent repair. It's also what happens without the protection of right to repair legislation that requires manufacturers to make repair parts, guides, and tools available to the general public. Specifically, a connection wire became desoldered from the battery in a watch that connects to the exoskeleton: "It's not the actual battery, but it's the little green connection piece we need to be the right fit and that's been our problem," Straight posted on Facebook. Straight's personal exoskeleton was broken for two months, he said in a video on Facebook. He was eventually able to get the device fixed after attention from an article in the Paulick Report, a website about the horse industry, and a spot on local TV. "It took me two months, and I got no results," he said in the video. With social media and news attention, "it only took you all four days, and look at the results," he said earlier this week while standing in the exoskeleton. "This is the dystopian nightmare that we've kind of entered in, where the manufacturer perspective on products is that their responsibility completely ends when it hands it over to a customer. That's not good enough for a device like this, but it's also the same thing we see up and down with every single product," Nathan Proctor, head of citizen rights group US PIRG's right to repair project told 404 Media. "People need to be able to fix things, there needs to be a plan in place. A $100,000 product you can only use as long as the battery lasts, that's enraging. We should not have to tolerate a society where this happens."

"We have all this technology we release into the wild and it changes people's lives, but there's no long-term thinking. Manufacturers currently have no legal obligation to support the equipment indefinitely and there's no requirements that they publish sufficient documentation to allow others to do it," Proctor said. "We need to set minimum standards for documentation so that, even if a company goes bankrupt or falls off the face of the earth, a technician with sufficient knowledge can fix it."

An anonymous reader quotes a report from CNN: Some McDonald's franchisees -- which own and operate 95% of McDonald's in the United States -- are now rolling out kiosks that can take cash and accept change. But even in these locations, McDonald's is reassigning cashiers to other roles, including new "guest experience lead" jobs that help customers use the kiosks and assist with any issues. "In theory, kiosks should help save on labor, but in reality, restaurants have added complexity due to mobile ordering and delivery, and the labor saved from kiosks is often reallocated for these efforts," said RJ Hottovy, an analyst who covers the restaurant and retail industries at data analytics firm Placer.ai. Kiosks "have created a restaurant within a restaurant." And in some cases, kiosks have even been a flop. Bowling ally chain Bowlero added kiosks in lanes for customers to order food and drinks, but they went unused because staff and customers weren't fully trained on using them. "The unintended consequences have surprised a lot of people," Hottovy said.

Even some of the benefits of kiosks touted by chains -- they upsell customers by suggesting menu items and speed up orders -- don't always play out. A recent study from Temple University researchers found that, when a line forms behind customers using kiosks, they experience more stress when placing their orders and purchase less food. And some customers take longer to order tapping around on kiosks and paying than they do telling a cashier they'd like to order a burger and fries. Not to mention the kiosks can malfunction or break down. "If kiosks really improved speed of service, order accuracy, and upsell, they'd be rolled out more extensively across the industry than they are today," Hottovy said.

Kiosks have also been threatened as a fast-food industry response to higher minimum wage laws. [...] But the quick-service and fast-casual segments of the restaurant industry continue to grow. Staffing levels were nearly 150,000 jobs, or 3%, above pre-pandemic levels, according to the latest Labor Department data. Christopher Andrews, a sociologist at Drew University who studies the effects of technology on work, said the impacts of kiosks were similar to other self-service technology such as ATMs and self-checkout machines in supermarkets. Both technologies were predicted to cause job losses. "The introduction of ATMs did not result in massive technological unemployment for bank tellers," he said. "Instead, it freed them up from low-value tasks such as depositing and cashing checks to perform other tasks that created value." Self-checkout have also not resulted in retail job losses, the report adds. "In some cases, self-checkout backfired for chains because self-checkout leads to higher merchandise losses from customer errors and more intentional shoplifting than when human cashiers are ringing up customers."