An anonymous reader quotes a report from TechCrunch: One of Microsoft's latest AI models can accurately predict air quality, hurricanes, typhoons, and other weather-related phenomena, the company claims. In a paper published in the journal Nature and an accompanying blog post this week, Microsoft detailed Aurora, which the tech giant says can forecast atmospheric events with greater precision and speed than traditional meteorological approaches. Aurora, which has been trained on more than a million hours of data from satellites, radar and weather stations, simulations, and forecasts, can be fine-tuned with additional data to make predictions for particular weather events.

AI weather models are nothing new. Google DeepMind has released a handful over the past several years, including WeatherNext, which the lab claims beats some of the world's best forecasting systems. Microsoft is positioning Aurora as one of the field's top performers -- and a potential boon for labs studying weather science. In experiments, Aurora predicted Typhoon Doksuri's landfall in the Philippines four days in advance of the actual event, beating some expert predictions, Microsoft says. The model also bested the National Hurricane Center in forecasting five-day tropical cyclone tracks for the 2022-2023 season, and successfully predicted the 2022 Iraq sandstorm.

While Aurora required substantial computing infrastructure to train, Microsoft says the model is highly efficient to run. It generates forecasts in seconds compared to the hours traditional systems take using supercomputer hardware. Microsoft, which has made the source code and model weights publicly available, says that it's incorporating Aurora's AI modeling into its MSN Weather app via a specialized version of the model that produces hourly forecasts, including for clouds.

Amazon's decade-old acquisition of Annapurna Labs has emerged as a pivotal element in its AI strategy, with the once-secretive Israeli chip design startup now powering AWS infrastructure. The $350 million deal, struck in 2015 after initial talks between Annapurna co-founder Nafea Bshara and Amazon executive James Hamilton, has equipped the tech giant with custom silicon capabilities critical to its cloud computing dominance.

Annapurna's chips, particularly the Trainium processor for AI model training and Graviton for general-purpose computing, now form the foundation of Amazon's AI infrastructure. The company is deploying hundreds of thousands of Trainium chips in its Project Rainier supercomputer being delivered to AI startup Anthropic this year. Amazon CEO Andy Jassy, who led AWS when the acquisition occurred, described it as "one of the most important moments" in AWS history.

Nvidia has announced plans to manufacture AI supercomputers entirely within the United States, commissioning over 1 million square feet of manufacturing space across Arizona and Texas. Production of Blackwell chips has begun at TSMC's Phoenix facilities, while supercomputer assembly will occur at new Foxconn and Wistron plants in Houston and Dallas respectively.

"The engines of the world's AI infrastructure are being built in the United States for the first time," said Jensen Huang, Nvidia's founder and CEO. "Adding American manufacturing helps us better meet the incredible and growing demand for AI chips and supercomputers, strengthens our supply chain and boosts our resiliency."

The company will deploy its own AI, robotics, and digital twin technologies in these facilities, using Nvidia Omniverse to create digital twins of factories and Isaac GR00T to build manufacturing automation robots. Nvidia projects an ambitious $500 billion in domestic AI infrastructure production over the next four years, with manufacturing expected to create hundreds of thousands of jobs.

"AMD processors were instrumental in achieving a new world record," reports Tom's Hardware, "during a recent Ansys Fluent computational fluid dynamics simulation run on the Frontier supercomputer at the Oak Ridge National Laboratory."

The article points out that Frontier was the fastest supercomputer in the world until it was beaten by Lawrence Livermore Lab's El Capitan — with both computers powered by AMD GPUs: According to a press release by Ansys, it ran a 2.2-billion-cell axial turbine simulation for Baker Hughes, an energy technology company, testing its next-generation gas turbines aimed at increasing efficiency. The simulation previously took 38.5 hours to complete on 3,700 CPU cores. By using 1,024 AMD Instinct MI250X accelerators paired with AMD EPYC CPUs in Frontier, the simulation time was slashed to 1.5 hours. This is more than 25 times faster, allowing the company to see the impact of the changes it makes on designs much more quickly...

Given those numbers, the Ansys Fluent CFD simulator apparently only used a fraction of the power available on Frontier. That means it has the potential to run even faster if it can utilize all the available accelerators on the supercomputer. It also shows that, despite Nvidia's market dominance in AI GPUs, AMD remains a formidable competitor, with its CPUs and GPUs serving as the brains of some of the fastest supercomputers on Earth.

"The number of discrete GPU developers from the U.S. and Western Europe shrank to three companies in 2025," notes Tom's Hardware, "from around 10 in 2000." (Nvidia, AMD, and Intel...) No company in the recent years — at least outside of China — was bold enough to engage into competition against these three contenders, so the very emergence of Bolt Graphics seems like a breakthrough. However, the major focuses of Bolt's Zeus are high-quality rendering for movie and scientific industries as well as high-performance supercomputer simulations. If Zeus delivers on its promises, it could establish itself as a serious alternative for scientific computing, path tracing, and offline rendering. But without strong software support, it risks struggling against dominant market leaders.
This week the Sunnyvale, California-based startup introduced its Zeus GPU platform designed for gaming, rendering, and supercomputer simulations, according to the article. "The company says that its Zeus GPU not only supports features like upgradeable memory and built-in Ethernet interfaces, but it can also beat Nvidia's GeForce RTX 5090 by around 10 times in path tracing workloads, according to slide published by technology news site ServeTheHome." There is one catch: Zeus can only beat the RTX 5090 GPU in path tracing and FP64 compute workloads. It's not clear how well it will handle traditional rendering techniques, as that was less of a focus. In speaking with Bolt Graphics, the card does support rasterization, but there was less emphasis on that aspect of the GPU, and it may struggle to compete with the best graphics cards when it comes to gaming. And when it comes to data center options like Nvidia's Blackwell B200, it's an entirely different matter.

Unlike GPUs from AMD, Intel, and Nvidia that rely on proprietary instruction set architectures, Bolt's Zeus relies on the open-source RISC-V ISA, according to the published slides. The Zeus core relies on an open-source out-of-order general-purpose RVA23 scalar core mated with FP64 ALUs and the RVV 1.0 (RISC-V Vector Extension Version 1.0) that can handle 8-bit, 16-bit, 32-bit, and 64-bit data types as well as Bolt's additional proprietary extensions designed for acceleration of scientific workloads... Like many processors these days, Zeus relies on a multi-chiplet design... Unlike high-end GPUs that prioritize bandwidth, Bolt is evidently focusing on greater memory size to handle larger datasets for rendering and simulations. Also, built-in 400GbE and 800GbE ports to enable faster data transfer across networked GPUs indicates the data center focus of Zeus.

High-quality rendering, real-time path tracing, and compute are key focus areas for Zeus. As a result, even the entry-level Zeus 1c26-32 offers significantly higher FP64 compute performance than Nvidia's GeForce RTX 5090 — up to 5 TFLOPS vs. 1.6 TFLOPS — and considerably higher path tracing performance: 77 Gigarays vs. 32 Gigarays. Zeus also features a larger on-chip cache than Nvidia's flagship — up to 128MB vs. 96MB — and lower power consumption of 120W vs. 575W, making it more efficient for simulations, path tracing, and offline rendering. However, the RTX 5090 dominates in AI workloads with its 105 FP16 TFLOPS and 1,637 INT8 TFLOPS compared to the 10 FP16 TFLOPS and 614 INT8 TFLOPS offered by a single-chiplet Zeus...
The article emphasizes that Zeus "is only running in simulation right now... Bolt Graphics says that the first developer kits will be available in late 2025, with full production set for late 2026."

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

Using the Summit supercomputer at the Department of Energy's Oak Ridge National Laboratory, researchers have modeled a key component of nucleotide excision repair (NER) called the pre-incision complex (PInC), which plays a crucial role in DNA damage repair. Their study, published in Nature Communications, provides new insights into how the PInC machinery orchestrates precise DNA excision, potentially leading to advancements in treating genetic disorders, preventing premature aging, and understanding conditions like xeroderma pigmentosum and Cockayne syndrome. Phys.Org reports: "Computationally, once you assemble the PInC, molecular dynamics simulations of the complex become relatively straightforward, especially on large supercomputers like Summit," [said lead investigator Ivaylo Ivanov, a chemistry professor at Georgia State University]. Nanoscale Molecular Dynamics, or NAMD, is a molecular dynamics code specifically designed for supercomputers and is used to simulate the movements and interactions of large biomolecular systems that contain millions of atoms. Using NAMD, the research team ran extensive simulations. The number-crunching power of the 200-petaflop Summit supercomputer -- capable of performing 200,000 trillion calculations per second -- was essential in unraveling the functional dynamics of the PInC complex on a timescale of microseconds. "The simulations showed us a lot about the complex nature of the PInC machinery. It showed us how these different components move together as modules and the subdivision of this complex into dynamic communities, which form the moving parts of this machine," Ivanov said.

The findings are significant in that mutations in XPF and XPG can lead to severe human genetic disorders. They include xeroderma pigmentosum, which is a condition that makes people more susceptible to skin cancer, and Cockayne syndrome, which can affect human growth and development, lead to impaired hearing and vision, and speed up the aging process. "Simulations allow us to zero in on these important regions because mutations that interfere with the function of the NER complex often occur at community interfaces, which are the most dynamic regions of the machine," Ivanov said. "Now we have a much better understanding of how and from where these disorders manifest."

In an interview with CNBC's Jon Fortt on Wednesday, Nvidia CEO Jensen Huang said next-gen AI will need 100 times more compute than older models as a result of new reasoning approaches that think "about how best to answer" questions step by step. From a report: "The amount of computation necessary to do that reasoning process is 100 times more than what we used to do," Huang told CNBC's Jon Fortt in an interview on Wednesday following the chipmaker's fourth-quarter earnings report. He cited models including DeepSeek's R1, OpenAI's GPT-4 and xAI's Grok 3 as models that use a reasoning process.

Huang pushed back on that idea in the interview on Wednesday, saying DeepSeek popularized reasoning models that will need more chips. "DeepSeek was fantastic," Huang said. "It was fantastic because it open sourced a reasoning model that's absolutely world class." Huang said that company's percentage of revenue in China has fallen by about half due to the export restrictions, adding that there are other competitive pressures in the country, including from Huawei.

Developers will likely search for ways around export controls through software, whether it be for a supercomputer, a personal computer, a phone or a game console, Huang said. "Ultimately, software finds a way," he said. "You ultimately make that software work on whatever system that you're targeting, and you create great software." Huang said that Nvidia's GB200, which is sold in the United States, can generate AI content 60 times faster than the versions of the company's chips that it sells to China under export controls.

According to The Intercept, the IRS is set to purchase an Nvidia SuperPod AI supercomputer to enhance its machine learning capabilities for tasks like fraud detection and taxpayer behavior analysis. From the report: With Elon Musk's so-called Department of Government Efficiency installing itself at the IRS amid a broader push to replace federal bureaucracy with machine-learning software, the tax agency's computing center in Martinsburg, West Virginia, will soon be home to a state-of-the-art Nvidia SuperPod AI computing cluster. According to the previously unreported February 5 acquisition document, the setup will combine 31 separate Nvidia servers, each containing eight of the company's flagship Blackwell processors designed to train and operate artificial intelligence models that power tools like ChatGPT. The hardware has not yet been purchased and installed, nor is a price listed, but SuperPod systems reportedly start at $7 million. The setup described in the contract materials notes that it will include a substantial memory upgrade from Nvidia.

Though small compared to the massive AI-training data centers deployed by companies like OpenAI and Meta, the SuperPod is still a powerful and expensive setup using the most advanced technology offered by Nvidia, whose chips have facilitated the global machine-learning spree. While the hardware can be used in many ways, it's marketed as a turnkey means of creating and querying an AI model. Last year, the MITRE Corporation, a federally funded military R&D lab, acquired a $20 million SuperPod setup to train bespoke AI models for use by government agencies, touting the purchase as a "massive increase in computing power" for the United States.

How exactly the IRS will use its SuperPod is unclear. An agency spokesperson said the IRS had no information to share on the supercomputer purchase, including which presidential administration ordered it. A 2024 report by the Treasury Inspector General for Tax Administration identified 68 different AI-related projects underway at the IRS; the Nvidia cluster is not named among them, though many were redacted. But some clues can be gleaned from the purchase materials. "The IRS requires a robust and scalable infrastructure that can handle complex machine learning (ML) workloads," the document explains. "The Nvidia Super Pod is a critical component of this infrastructure, providing the necessary compute power, storage, and networking capabilities to support the development and deployment of large-scale ML models."

The document notes that the SuperPod will be run by the IRS Research, Applied Analytics, and Statistics division, or RAAS, which leads a variety of data-centric initiatives at the agency. While no specific uses are cited, it states that this division's Compliance Data Warehouse project, which is behind this SuperPod purchase, has previously used machine learning for automated fraud detection, identity theft prevention, and generally gaining a "deeper understanding of the mechanisms that drive taxpayer behavior."

Nearly three years ago a team of Chinese AI engineers working for DeepSeek's parent company unveiled an earlier AI supercomputer that the Washington Post says was constructed from 10,000 A100 GPUs purchased from Nvidia. Roughly six months later "Washington had banned Nvidia from selling any more A100s to China," the article notes.

Remember that number as you read this. 10,000 A100 GPUs... DeepSeek's new chatbot caused a panic in Silicon Valley and on Wall Street this week, erasing $1 trillion from the stock market. That impact stemmed in large part from the company's claim that it had trained one of its recent models on a minuscule $5.6 million in computing costs and with only 2,000 or so of Nvidia's less-advanced H800 chips.

Nvidia saw its soaring value crater by $589 billion Monday as DeepSeek rocketed to the top of download charts, prompting President Donald Trump to call for U.S. industry to be "laser focused" on competing... But a closer look at DeepSeek reveals that its parent company deployed a large and sophisticated chip set in its supercomputer, leading experts to assess the total cost of the project as much higher than the relatively paltry sum that U.S. markets reacted to this week... Lennart Heim, an AI expert at Rand, said DeepSeek's evident access to [the earlier] supercomputer would have made it easier for the company to develop a more efficient model, requiring fewer chips.
That earlier project "suggests that DeepSeek had a major boost..." according to the article, "with technology comparable to that of the leading U.S. AI companies." And while DeepSeek claims it only spent $5.6 million to train one of its advanced models, "its parent company has said that building the earlier supercomputer had cost 1 billion yuan, or $139 million.") Yet the article also cites the latest insights Friday from chip investment company SemiAnalysis, summarizing their finding that DeepSeek "has spent more than half a billion dollars on GPUs, with total capital expenditures of almost $1.3 billion."

The article notes Thursday remarks by OpenAI CEO Sam Altman that DeepSeek's energy-efficiency claims were "wildly overstated... This is a model at a capability level that we had quite some time ago." And Palmer Luckey called DeepSeek "legitimately impressive" on X but called the $5.6 million training cost figure "bogus" and said the Silicon Valley meltdown was "hysteria." Even with these higher total costs in mind, experts say, U.S. companies are right to be concerned about DeepSeek upending the market. "We know two things for sure: DeepSeek is pricing their services very competitively, and second, the performance of their models is comparable to leading competitors," said Kai-Shen Huang, an AI expert at the Research Institute for Democracy, Society and Emerging Technology, a Taipei-based think tank. "I think DeepSeek's pricing strategy has the potential to disrupt the market globally...."

China's broader AI policy push has helped create an environment conducive for a company like DeepSeek to rise. Beijing announced an ambitious AI blueprint in 2017, with a goal to become a global AI leader by 2030 and promises of funding for universities and private enterprise. Local governments across the nation followed with their own programs to support AI.

An anonymous reader quotes a report from MIT Technology Review: A Canadian startup called Xanadu has built a new quantum computer it says can be easily scaled up to achieve the computational power needed to tackle scientific challenges ranging from drug discovery to more energy-efficient machine learning. Aurora is a "photonic" quantum computer, which means it crunches numbers using photonic qubits -- information encoded in light. In practice, this means combining and recombining laser beams on multiple chips using lenses, fibers, and other optics according to an algorithm. Xanadu's computer is designed in such a way that the answer to an algorithm it executes corresponds to the final number of photons in each laser beam. This approach differs from one used by Google and IBM, which involves encoding information in properties of superconducting circuits.

Aurora has a modular design that consists of four similar units, each installed in a standard server rack that is slightly taller and wider than the average human. To make a useful quantum computer, "you copy and paste a thousand of these things and network them together," says Christian Weedbrook, the CEO and founder of the company. Ultimately, Xanadu envisions a quantum computer as a specialized data center, consisting of rows upon rows of these servers. This contrasts with the industry's earlier conception of a specialized chip within a supercomputer, much like a GPU. [...]

Xanadu's 12 qubits may seem like a paltry number next to IBM's 1,121, but Tiwari says this doesn't mean that quantum computers based on photonics are running behind. In his opinion, the number of qubits reflects the amount of investment more than it does the technology's promise. [...] Xanadu's next goal is to improve the quality of the photons in the computer, which will ease the error correction requirements. "When you send lasers through a medium, whether it's free space, chips, or fiber optics, not all the information makes it from the start to the finish," he says. "So you're actually losing light and therefore losing information." The company is working to reduce this loss, which means fewer errors in the first place. Xanadu aims to build a quantum data center, with thousands of servers containing a million qubits, in 2029. The company published its work on chip design optimization and fabrication in the journal Nature.

An anonymous reader quotes a report from Ars Technica: On Thursday, OpenAI announced that it is deepening its ties with the US government through a partnership with the National Laboratories and expects to use AI to "supercharge" research across a wide range of fields to better serve the public. "This is the beginning of a new era, where AI will advance science, strengthen national security, and support US government initiatives," OpenAI said. The deal ensures that "approximately 15,000 scientists working across a wide range of disciplines to advance our understanding of nature and the universe" will have access to OpenAI's latest reasoning models, the announcement said.

For researchers from Los Alamos, Lawrence Livermore, and Sandia National Labs, access to "o1 or another o-series model" will be available on Venado -- an Nvidia supercomputer at Los Alamos that will become a "shared resource." Microsoft will help deploy the model, OpenAI noted. OpenAI suggested this access could propel major "breakthroughs in materials science, renewable energy, astrophysics," and other areas that Venado was "specifically designed" to advance. Key areas of focus for Venado's deployment of OpenAI's model include accelerating US global tech leadership, finding ways to treat and prevent disease, strengthening cybersecurity, protecting the US power grid, detecting natural and man-made threats "before they emerge," and " deepening our understanding of the forces that govern the universe," OpenAI said.

Perhaps among OpenAI's flashiest promises for the partnership, though, is helping the US achieve a "a new era of US energy leadership by unlocking the full potential of natural resources and revolutionizing the nation's energy infrastructure." That is urgently needed, as officials have warned that America's aging energy infrastructure is becoming increasingly unstable, threatening the country's health and welfare, and without efforts to stabilize it, the US economy could tank. But possibly the most "highly consequential" government use case for OpenAI's models will be supercharging research safeguarding national security, OpenAI indicated. "The Labs also lead a comprehensive program in nuclear security, focused on reducing the risk of nuclear war and securing nuclear materials and weapons worldwide," OpenAI noted. "Our partnership will support this work, with careful and selective review of use cases and consultations on AI safety from OpenAI researchers with security clearances." The announcement follows the launch earlier this week of ChatGPT Gov, "a new tailored version of ChatGPT designed to provide US government agencies with an additional way to access OpenAI's frontier models." It also worked with the Biden administration to voluntarily commit to give officials early access to its latest models for safety inspections.

Nvidia has dedicated a supercomputer running thousands of its latest GPUs exclusively to improving its DLSS upscaling technology for the past six years, a company executive revealed at CES 2025. Speaking at the RTX Blackwell Editor's Day in Las Vegas, Brian Catanzaro, Nvidia's VP of applied deep learning research, said the system operates continuously to analyze failures and retrain models across hundreds of games.

"I know, I know: 'Year of the Linux desktop ... yadda, yadda'," writes Steven Vaughan-Nichols, a ZDNet senior contributing editor. "You've heard it all before. But now there's a Linux-powered PC that many people will want..."

He's talking about Nvidia's newly-announced Project Digits, describing it as "a desktop with AI supercomputer power that runs DGX OS, a customized Ubuntu Linux 22.04 distro." Powered by MediaTek and Nvidia's Grace Blackwell Superchip, Project DIGITS is a $3,000 personal AI that combines Nvidia's Blackwell GPU with a 20-core Grace CPU built on the Arm architecture... At CES, Nvidia CEO Jensen Huang confirmed plans to make this technology available to everyone, not just AI developers. "We're going to make this a mainstream product," Huang said. His statement suggests that Nvidia and MediaTek are positioning themselves to challenge established players — including Intel and AMD — in the desktop CPU market. This move to the desktop and perhaps even laptops has been coming for a while. As early as 2023, Nvidia was hinting that a consumer desktop chip would be in its future... [W]hy not use native Linux as the primary operating system on this new chip family?

Linux, after all, already runs on the Grace Blackwell Superchip. Windows doesn't. It's that simple. Nowadays, Linux runs well with Nvidia chips. Recent benchmarks show that open-source Linux graphic drivers work with Nvidia GPUs as well as its proprietary drivers. Even Linus Torvalds thinks Nvidia has gotten its open-source and Linux act together. In August 2023, Torvalds said, "Nvidia got much more involved in the kernel. Nvidia went from being on my list of companies who are not good to my list of companies who are doing really good work." Canonical, Ubuntu Linux's parent company, has long worked closely with Nvidia. Ubuntu already provides Blackwell drivers.
The article strays into speculation, when it adds "maybe you wouldn't pay three grand for a Project DIGITS PC. But what about a $1,000 Blackwell PC from Acer, Asus, or Lenovo? All three of these companies are already selling MediaTek-powered Chromebooks...."

"The first consumer products featuring this technology are expected to hit the market later this year. I'm looking forward to running Linux on it. Come on in! The operating system's fine."

The world's most powerful supercomputer, capable of 2.79 quintillion calculations per second, has been unveiled at Lawrence Livermore National Laboratory in California, designed primarily to maintain the U.S. nuclear weapons stockpile and run other classified simulations. The $600 million system, named El Capitan, consists of 87 computer racks weighing 1.3 million pounds and draws 30 megawatts of power.

Built by Hewlett-Packard Enterprise using AMD chips, it operates alongside a smaller system called Tuolumne, which ranks tenth globally in computing power. "While we're still exploring the full role AI will play, there's no doubt that it is going to improve our ability to do research and development that we need," said Bradley Wallin, a deputy director at the laboratory.

"Nvidia has officially introduced its highly anticipated GeForce 50 Series graphics cards, accompanied by the debut of DLSS 4 technology," writes Slashdot reader jjslash. "The lineup includes four premium GPUs: the RTX 5080 and RTX 5090 are slated for release on January 30, with the RTX 5070 and RTX 5070 Ti following in February. TechSpot recount of the Jensen Huang keynote tries to differentiate between dubious performance claims and actual expected raw output": The new RTX 5090 flagship comes packing significantly more hardware over its predecessor. Not only does this GPU use Nvidia's new Blackwell architecture, but it also packs significantly more CUDA cores, greater memory bandwidth, and a higher VRAM capacity. The SM count has increased from 128 with the RTX 4090 to a whopping 170 with the RTX 5090 -- a 33% increase in the core size. The memory subsystem is overhauled, now featuring GDDR7 technology on a massive 512-bit bus. With this GDDR7 memory clocked at 28 Gbps, memory bandwidth reaches 1,792 GB/s -- a near 80% increase over the RTX 4090's bandwidth. It also includes 32GB of VRAM, the most Nvidia has ever provided on a consumer GPU. [...]

As for the performance claims... Nvidia has - as usual - used its marketing to obscure actual gaming performance. RTX 50 GPUs support DLSS 4 multi-frame generation, which previous-generation GPUs lack. This means RTX 50 series GPUs can generate double the frames of previous-gen models in DLSS-supported games, making them appear up to twice as "fast" as RTX 40 series GPUs. But in reality, while FPS numbers will increase with DLSS 4, latency and gameplay feel may not improve as dramatically. [...] The claim that the RTX 5070 matches the RTX 4090 in performance seems dubious. Perhaps it could match in frame rate with DLSS 4, but certainly not in raw, non-DLSS performance. Based on Nvidia's charts, the RTX 5070 seems 20-30% faster than the RTX 4070 at 1440p. This would place the RTX 5070 slightly ahead of the RTX 4070 Super for about $50 less, or alternatively, 20-30% faster than the RTX 4070 for the same price. These GeForce 50 series wasn't the only announcement Nvidia made at CES 2025. The chipmaker unveiled a $3,000 personal AI supercomputer, capable of running sophisticated AI models with up to 200 billion parameters. It also announced plans to introduce AI-powered autonomous characters in video games this year, starting with a virtual teammate in the battle royale game PUBG.

Nvidia will begin selling a personal AI supercomputer in May that can run sophisticated AI models with up to 200 billion parameters, the chipmaker has announced. The $3,000 Project Digits system is powered by the new GB10 Grace Blackwell Superchip and can operate from a standard power outlet.

The device delivers 1 petaflop of AI performance and includes 128GB of memory and up to 4TB of storage. Two units can be linked to handle models with 405 billion parameters. "AI will be mainstream in every application for every industry," Nvidia CEO Jensen Huang said. The system runs on Linux-based Nvidia DGX OS and supports PyTorch, Python, and Jupyter notebooks.

Dell CEO Michael Dell has acknowledged delays in corporate adoption of AI-enabled PCs but remains confident in their eventual widespread uptake, citing his four decades of industry experience with technology transitions.

The PC maker's chief executive told Fortune that while the current refresh cycle is "definitely delayed," adoption is inevitable once sufficient features drive customer demand. Meanwhile, Dell's infrastructure division saw 80% revenue growth last quarter from AI-server sales. The company is supplying servers for xAI's Colossus supercomputer project in Memphis and sees opportunities in "sovereign AI" systems for nations seeking technological independence. "Pick a country ranked by GDP, the [top] 49 other than the U.S., they all need one," Dell said.

An anonymous reader shared this report from Reuters: Elon Musk's artificial intelligence startup xAI plans to expand its Memphis, Tennessee, supercomputer to house at least one million graphics processing units (GPUs), the Greater Memphis Chamber said on Wednesday, as xAI races to compete against rivals like OpenAI.

The move represents a massive expansion for the supercomputer called Colossus, which currently has 100,000 GPUs to train xAI's chatbot called Grok. As part of the expansion, Nvidia, which supplies the GPUs, and Dell and Super Micro, which have assembled the server racks for the computer, will establish operations in Memphis, the chamber said in a statement.
The Greater Memphis chamber (an economic development organization) called it "the largest capital investment in the region's history," even saying that xAI "is setting the stage for Memphis to become the global epicenter of artificial intelligence." ("To facilitate this massive undertaking, the Greater Memphis Chamber established an xAI Special Operations Team... This team provides round-the-clock concierge service to the company.")

Reuters calls the supercomputer "a critical component of advancing Musk's AI efforts, as the billionaire has deepened his rivalry against OpenAI..." And the Greater Memphis chamber describes the expansion by Nvidia/Dell/Super Micro as "further solidifying the city's position as the 'Digital Delta'... Memphis has provided the power and velocity necessary for not just xAI to grow and thrive, but making way for other companies as well."

Leading technology companies are dramatically expanding their AI capabilities by building multibillion-dollar "super clusters" packed with unprecedented numbers of Nvidia's AI processors. Elon Musk's xAI recently constructed Colossus, a supercomputer containing 100,000 Nvidia Hopper chips, while Meta CEO Mark Zuckerberg claims his company operates an even larger system for training advanced AI models. The push toward massive chip clusters has helped drive Nvidia's quarterly revenue from $7 billion to over $35 billion in two years, making it the world's most valuable public company.

WSJ adds: Nvidia Chief Executive Jensen Huang said in a call with analysts following its earnings Wednesday that there was still plenty of room for so-called AI foundation models to improve with larger-scale computing setups. He predicted continued investment as the company transitions to its next-generation AI chips, called Blackwell, which are several times as powerful as its current chips.

Huang said that while the biggest clusters for training for giant AI models now top out at around 100,000 of Nvidia's current chips, "the next generation starts at around 100,000 Blackwells. And so that gives you a sense of where the industry is moving."

Lawrence Livermore National Laboratory's "El Capitan" supercomputer is now ranked as the world's most powerful, exceeding a High-Performance Linpack (HPL) score of 1.742 exaflops on the latest Top500 list. Engadget reports: El Capitan is only the third "exascale" computer, meaning it can perform more than a quintillion calculations in a second. The other two, called Frontier and Aurora, claim the second and third place slots on the TOP500 now. Unsurprisingly, all of these massive machines live within government research facilities: El Capitan is housed at Lawrence Livermore National Laboratory; Frontier is at Oak Ridge National Laboratory; Argonne National Laboratory claims Aurora. [Cray Computing] had a hand in all three systems.

El Capitan has more than 11 million combined CPU and GPU cores based on AMD 4th-gen EPYC processors. These 24-core processors are rated at 1.8GHz each and have AMD Instinct M1300A APUs. It's also relatively efficient, as such systems go, squeezing out an estimated 58.89 Gigaflops per watt. If you're wondering what El Capitan is built for, the answer is addressing nuclear stockpile safety, but it can also be used for nuclear counterterrorism.