Get ONMINE

Stay Ahead, Stay ONMINE

Major network vendors team to advance Ethernet for scale-up AI networking

Another multivendor development group, the Ultra Accelerator Link (UALink) consortium, recently published its first specification aimed at delivering an open standard interconnect for AI clusters. The UALink 200G 1.0 Specification was crafted by many of the group’s 75 members — which include AMD, Broadcom, Cisco, Google, HPE, Intel, Meta, Microsoft and Synopsys — and lays […]

Another multivendor development group, the Ultra Accelerator Link (UALink) consortium, recently published its first specification aimed at delivering an open standard interconnect for AI clusters. The UALink 200G 1.0 Specification was crafted by many of the group’s 75 members — which include AMD, Broadcom, Cisco, Google, HPE, Intel, Meta, Microsoft and Synopsys — and lays out the technology needed to support a maximum data rate of 200 Giga transfers per second (GT/s) per channel or lane between accelerators and switches between up to 1,024 AI computing pods, UALink stated. 

ESUN will leverage the work of IEEE and UEC for Ethernet when possible, stated Arista’s CEO Jayshree Ullal and chief development officer Hugh Holbrook in a blog post about ESUN. To that end, Ullal and Holbrook described a modular framework for Ethernet scale-up with three key building blocks:

  1. Common Ethernet headers for Interoperability: ESUN will build on top of Ethernet to enable the widest range of upper-layer protocols and use cases.
  2. Open Ethernet data link layer: Provides the foundation for AI collectives with high-performance at XPU cluster scale. By selecting standards-based mechanisms (such as Link-Layer Retry (LLR), Priority-based Flow Control (PFC) and Credit-based Flow Control (CBFC), ESUN enables cost-efficiency and flexibility with performance for these networks. Even minor delays can stall thousands of concurrent operations.
  3. Ethernet PHY layer: By relying on the ubiquitous Ethernet physical layer, interoperability across multiple vendors and a wide range of optical and copper interconnect options is assured.

“ESUN is designed to support any upper layer transport, including one based on SUE-T. SUE-T (Scale-Up Ethernet Transport) is a new OCP workstream, seeded by Broadcom’s contribution of SUE (Scale-Up Ethernet) to OCP. SUE-T looks to define functionality that can be easily integrated into an ESUN-based XPU for reliability scheduling, load balancing, and transaction packing, which are critical performance enhancers for some AI workloads,” Ullal and Holbrook wrote.

“In essence, the ESUN framework enables a collection of individual accelerators to become a single, powerful AI super computer, where network performance directly correlates to the speed and efficiency of AI model development and execution,” Ullal and Holbrook wrote. “The layered approach of ESUN and SUE-T over Ethernet promotes innovation without fragmentation. XPU accelerator developers retain flexibility on host-side choices such as access models (push vs. pull, and memory vs streaming semantics), transport reliability (hop-by-hop vs. end-to-end), ordering rules, and congestion control strategies while retaining system design choices. The ESUN initiative takes a practical approach for iterative improvements.”

Gartner expects gains in AI networking fabrics

Scale-up AI fabrics (SAIF) have captured a lot of industry attention lately, according to Gartner. The research firm is forecasting massive growth in SAIF to support AI infrastructure initiatives through 2029. The vendor landscape will remain dynamic over the next two years, with multiple technology ecosystems emerging, Gartner wrote in its report, What are “Scale-Up” AI Fabrics and Why Should I Care?

“Scale-Up” AI fabrics (SAIF) provide high-bandwidth, low-latency physical network interconnectivity and enhanced memory interaction between nearby AI processors,” Garter wrote. “Current implementations of SAIF are vendor-proprietary platforms, and there are proximity limitations (typically, SAIF is confined to only a rack or row). In most scenarios, Gartner recommends using Ethernet when connecting multiple SAIF systems together. We believe the scale, performance and supportability of Ethernet is optimal.”

Shape
Shape
Stay Ahead

Explore More Insights

More Articles

Stay ahead with more perspectives on cutting-edge power, infrastructure, energy,  bitcoin and AI solutions. Explore these articles to uncover strategies and insights shaping the future of industries.

Shape

Intel details new efficient Xeon processor line

The new chips will be able to support up to 12-channel DDR5 memory with speeds of up to 8000 MT/s, a substantial increase over the 8 channels of 6400MT/s in the prior generation. In addition to that, the platform will support up to 6 UPI 2.0 links with up to

Read More »

IEA Raises Estimate for Record Oil Glut in 2026

A record oversupply of oil will be bigger than previously estimated and the excess is already starting to build up on ocean going tankers, the International Energy Agency said. World oil supply will exceed demand by almost 4 million barrels a day next year, an unprecedented overhang in annual terms, the IEA said in its latest monthly report. Its predicted surplus is up roughly 18% from last month’s estimate, as the OPEC+ alliance continues to revive output and the outlook for the group’s rivals in 2026 strengthens.  While inventories have piled at a brisk clip of 1.9 million barrels a day this year, their impact on prices has been mitigated by China scooping up the majority, according to the report. That’s beginning to change as a surge in Middle East exports pushes the volume of oil on the water to the highest level in years, the IEA said.  “Looking ahead, as the significant volumes of crude oil on water move onshore to major oil hubs, crude stocks look set to surge,” the Paris-based agency adviser to major economies said. It trimmed consumption growth estimates slightly for this year, and boosted non-OPEC supply estimates for this year and next. The surplus is building as demand growth in China and other key consumers cools, while the OPEC+ alliance revives halted production and the group’s rivals in the Americas continue to expand rapidly. Crude futures are trading near $63 a barrel in London, down 15% for the year. While Wall Street firms including Goldman Sachs Group Inc. and JPMorgan Chase & Co. predict further losses, the market has so far been spared the crash that some anticipated when Saudi Arabia and its partners starting opening the taps earlier this year. That’s partly because much of the supply excess has been in the form of natural

Read More »

Crude Near $59 on Surplus Fears

Oil fell as escalating trade tensions between China and the US diminished demand for risky assets while the International Energy Agency increased its estimate of a record crude surplus. West Texas Intermediate dropped 1.3% on Tuesday to settle near $59 a barrel, the lowest since May, while Brent hovered near $62. In the latest tit-for-tat between Beijing and Washington, China placed limits on five US entities of one of South Korea’s biggest shipbuilders, and threatened further retaliatory measures. The Paris-based IEA on Tuesday increased its forecast for an unprecedented oversupply of oil for 2026. Worldwide crude supplies will exceed demand by almost 4 million barrels a day next year, a record overhang in annual terms, the agency said. Also on Tuesday, US Federal Reserve Chair Jerome Powell reinforced speculation that officials are on track to cut rates in October amid a weakening labor market. “The intraday bounce off the lows in oil today was in reaction of a turnaround in risk tone from Powell’s dovish comments on quantitative easing,” said Frank Monkam, head of macro trading at Buffalo Bayou Commodities. Even so, with oil fundamentals still skewed bearish, “the 60-62 level on WTI is likely to constitute a firm resistance level should the bounce extend.” The projected supply surplus is up roughly 18% from last month’s estimate, as the OPEC+ alliance continues to revive output and the outlook for the group’s rivals strengthens. Several executives from major oil trading houses speaking in London said they see crude prices falling from here. Ben Luckock, global head of oil at Trafigura Group, warned that the onset of a long-awaited oil market surplus is “just about here,” while Gunvor Chief Executive Officer Torbjorn Tornqvist said gasoline and diesel demand may have plateaued. Oil posted losses in August and September, and WTI has shed

Read More »

New York City could face power reliability issues beginning next year: ISO

Dive Brief: The New York electric grid faces increased risk of power shortages over the next five years unless planned projects, including new transmission and offshore wind resources, are brought online, the state’s grid operator said Tuesday. Starting next summer, the Independent System Operator anticipates its reliability margins in New York City will be dangerously thin, making the grid more vulnerable to failures. In addition to potentially losing anticipated wind power, the system is strained by generator deactivations, increasing consumer demand, transmission limitations and difficulties in developing new resources, the ISO said in a statement highlighting a pair of reliability reports.  “The NYISO’s findings should be alarming to residents and serve as another wake up call for the state,” Gavin Donohue, president of the Independent Power Producers of New York, said in a statement. “Electric demand is continuing to drastically rise, and the state needs to look at all possible resources.” Dive Insight: The ISO’s Short-Term Assessment of Reliability examined the five-year period from July 2025 to July 2030 and identified reliability weaknesses beginning in New York City in 2026, in Long Island in 2027 and in the Lower Hudson Valley region in 2030.  The New York City area will be deficient “through the entire five-year horizon without the completion and energization of future planned projects,” the STAR report concluded. Those projects include the 816-MW Empire Wind offshore project, which was expected to be online by 2027. That timeline was complicated when the Trump administration halted the project in April before allowing it to resume a month later. Since then, it has faced additional complications and delays. The STAR report also cited the importance of the 1,250-MW Champlain Hudson Power Express transmission line, which is slated bring power to the city from Quebec beginning next year. “Once CHPE, Empire Wind, and

Read More »

California to invest $226M in offshore wind ports amid federal cuts

Dive Brief: The California Energy Commission authorized $42.75 million in grants for offshore wind port development on Wednesday, in what commission staff described as the first appropriation of the state’s Climate Bond funds for offshore wind. Last month, California state lawmakers also authorized $225.7 million in spending for offshore wind ports and related facilities through June 2030, with funds coming from a $10 billion Climate Bond approved by voters in 2024. The construction of supportive ports and transmission systems in California is critical to deploying commercial floating offshore wind by the mid 2030s, according to a survey by offshore wind industry trade association Oceantic. Dive Insight: Despite stop-work orders and project cancellations on the East Coast, California’s status as a global leader in floating offshore wind development remains unchanged, according to Oceantic Senior Policy Director Nancy Kirshner-Rodriguez. “States have always been the driving force of our industry, and will continue to be, regardless of who is in power,” she said. “They are creating the demand signals that will pull in investment, and they are underwriting the critical enabling infrastructure investments that make the market move forward.” The Port San Luis Harbor District — one of five entities that received grants from the CEC last week to fund offshore wind port planning, engineering and design — is on track to become the state’s first dedicated offshore wind hub, according to Reid Boggiano, a CEC offshore wind program specialist. The Port of Long Beach is also well on its way to building Pier Wind, a dedicated 400-acre offshore wind terminal that the CEC awarded $20 million to wrap up planning and engineering and to complete environmental assessments. The Humboldt Bay Harbor and the cities of Oakland and Richmond also received grants for feasibility studies to determine if those communities could host offshore wind

Read More »

Interior denies canceling largest solar project in U.S. after axing review

The U.S. Department of the Interior has canceled its broad environmental review for the seven individual projects that make up the 6.2-GW Esmeralda 7 solar project located on federal land in Nevada and will review and permit each project individually, according to a spokesperson. Esmeralda 7 is set to be the largest solar project in the U.S. by capacity. The project’s National Environmental Policy Act status is listed as canceled on the Bureau of Land Management’s website. In a Tuesday email, the spokesperson said the BLM had not canceled the project. “During routine discussions prior to the lapse in appropriations, the proponents and BLM agreed to change their approach for the Esmeralda 7 Solar Project in Nevada,” they said. “Instead of pursuing a programmatic level environmental analysis, the applicants will now have the option to submit individual project proposals to the BLM to more effectively analyze potential impacts.” According to a draft programmatic environmental impact statement for the project from July of last year, the individual projects are: Lone Mountain Solar, Nivloc Solar, Smoky Valley Solar, Red Ridge 1 Solar, Red Ridge 2 Solar, Esmeralda Energy Center and Gold Dust Solar, which “would be geographically contiguous and encompass approximately 62,300 acres of BLM-administered lands approximately 30 miles west of Tonopah, Nevada.” Developers of the seven projects include Invenergy, Avantus, and NextEra. A spokesperson for NextEra told The Guardian, “We are in the early stage of development and remain committed to pursuing our project’s comprehensive environmental analysis by working closely with the Bureau of Land Management.” All of the projects have pending right-of-way applications before the Bureau of Land Management, the draft PEIS said. As the project’s previous NEPA review process has been canceled, it’s unclear how long it will now take each project to secure approvals. The original draft PEIS said that

Read More »

Losing power, losing billions: How offshoring grid materials weakens America

Jim Welsh is CEO of Peak Nano. National security requires self-reliance and independence. Today, America’s energy infrastructure and supply chain face a critical test of both.  Electrification, AI, data centers and extreme weather are driving unprecedented demand for energy and grid reliability. The U.S. Department of Energy’s latest Grid Reliability Report makes one thing clear: the time for “business-as-usual” is over. The U.S. can’t keep up, let alone grow, on old infrastructure. We need a radical shift to expand, modernize and intelligently manage the grid.  Success depends on bold investments in digital technologies, enhanced equipment and advanced grid management. Together, these will enable real-time monitoring, rapid integration of renewables and far less energy lost in transmission, which today can range from 8-12% of power generation. That’s how America will meet surging demand and protect its global leadership in the AI era.   The U.S. has an Achilles’ heel: critical materials. At the heart of grid reliability are magnets, rare-earth minerals and dielectric film. These materials are often overlooked but are vital for our power infrastructure, grid reliability and every part of our energy security. Dielectric films, for example, are used in capacitors that condition power, convert AC/DC power, keep power flowing steadily and even help manage spikes in demand to keep the grid stable and secure.   The problem? The U.S. has outsourced our ability to replace failing transformers, capacitors and other critical grid components. Capacitor film — a highly engineered, ultra-thin plastic that enables power stability and distribution for our grid — is almost entirely made overseas, and 75% is made in China, which dominates the global supply.  Every year, we spend nearly $200 billion overseas for this film and other critical materials. Historically, we’ve migrated our capacity to manufacture it overseas. No American manufacturer even builds the equipment to make dielectric films.  This isn’t

Read More »

Florida’s Data Center Moment: Power, Policy, and Potential

Florida is rapidly positioning itself as one of the next major frontiers for data center development. With extended tax incentives, proactive utilities, and a strategic geographic advantage, the state is aligning power, policy, and economic development in ways that echo the early playbook of Northern Virginia. In the latest episode of The Data Center Frontier Show, Buddy Rizer, Executive Director of Loudoun County Economic Development, and Lila Jaber, Founder of the Florida’s Women in Energy Leadership Forum and former Chair of the Florida Public Service Commission, join DCF to explore the opportunities and lessons shaping Florida’s emergence as a data center powerhouse. Energy and Infrastructure: A Strong Starting Position Unlike regions grappling with grid strain, Florida begins its data center growth story with energy abundance. While Loudoun County, Virginia—home to the world’s largest concentration of data centers—faced a 600 MW power deficit last year and could reach 12 GW of demand by 2030, Florida maintains excess generation capacity and robust renewable energy integration. Utilities like Florida Power & Light (FPL) and Duke Energy are already preparing for hyperscale and AI-driven loads, filing new large-load tariff structures to balance growth with ratepayer protection. Over the past decade, Florida utilities have also invested billions to harden their grids against hurricanes and extreme weather, resulting in some of the most resilient energy infrastructure in the country. Florida’s 10-year generation planning requirement, which ensures a diverse portfolio including nuclear, solar, and battery storage, further positions the state to meet growing digital infrastructure needs through hybrid on-site generation and demand-response capabilities. Economic and Workforce Advantages The state’s renewed sales tax exemptions for data centers through 2037—and the raised 100 MW IT load threshold—signal a strong bid to attract hyperscale operators and large-scale AI campuses. Florida also offers a competitive electricity rate structure comparable to Virginia’s

Read More »

Inside Blackstone’s Electrification Push: From Shermco to the Power Backbone of AI Data Centers

According to the National Electrical Manufacturers Association (NEMA), U.S. energy demand is projected to grow 50% by 2050. Electrical manufacturers have invested more than $10 billion since 2021 in new technologies to expand grid and manufacturing capacity, also reducing reliance on materials from China by 32% since 2018. Power access, sustainable infrastructure, and land acquisition have become critical factors shaping where and how data center facilities are built. As we previously reported in Data Center Frontier, investors realized this years ago, viewing these facilities both as technology assets and a unique convergence of real estate, utility infrastructure, and mission-critical systems that can also generate revenue. One of those investors is global asset manager Blackstone, which through its Energy Transition Partners private equity arm, recently acquired Shermco Industries for $1.6 billion. Announced August 21, the deal is part of Blackstone’s strategy to invest in companies that support the growing demand for electrification and a more reliable power grid. The goal is to strengthen data center infrastructure reliability and expand critical electrical services. Founded in 1974, Texas-based Shermco is one of the largest electrical testing organizations accredited by the InterNational Electrical Testing Association (NETA). The company operates in a niche yet important space: providing lifecycle electrical services, including maintenance, testing, commissioning, repair, and design, in support of data centers, utilities, and industrial clients. It has more than 40 service centers in the U.S. and Canada. In addition to helping Blackstone support its electrification and power grid reliability goals, the Shermco purchase is also part of Blackstone’s strategy to increase scale and resources—revenue increases without a substantial increase in resources—thus expanding its footprint and capabilities within the essential energy services sector.  As data centers expand globally, become more energy intensive, and are pressured to incorporate renewables and modernize grids, Blackstone’s leaders plan to leverage Shermco’s

Read More »

Cooling, Compute, and Convergence: How Strategic Alliances Are Informing the AI Data Center Playbook

Schneider Electric and Compass Datacenters: Prefabrication Meets the AI Frontier “We’re removing bottlenecks and setting a new benchmark for AI-ready data centers.” — Aamir Paul, Schneider Electric In another sign of how collaboration is accelerating the next wave of AI infrastructure, Schneider Electric and Compass Datacenters have joined forces to redefine the data center “white space” build-out: the heart of where power, cooling, and compute converge. On September 9, the two companies unveiled the Prefabricated Modular EcoStruxure™ Pod, a factory-built, fully integrated white space module designed to compress construction timelines, reduce CapEx, and simplify installation while meeting the specific demands of AI-ready infrastructure. The traditional fit-out process for the IT floor (i.e. integrating power distribution, cooling systems, busways, cabling, and network components) has long been one of the slowest and most error-prone stages of data center construction. Schneider and Compass’ new approach tackles that head-on, shifting the entire workflow from fragmented on-site assembly to standardized off-site manufacturing. “The traditional design and approach to building out power, cooling, and IT networking equipment has relied on multiple parties installing varied pieces of equipment,” the companies noted. “That process has been slow, inefficient, and prone to errors. Today’s growing demand for AI-ready infrastructure makes traditional build-outs ripe for improvement.” Inside the EcoStruxure Pod: White Space as a Product The EcoStruxure Pod packages every core element of a high-performance white space environment (power, cooling, and IT integration) into a single prefabricated, factory-tested unit. Built for flexibility, it supports hot aisle containment, InRow cooling, and Rear Door Heat Exchanger (RDHx) configurations, alongside high-power busways, complex network cabling, and a technical water loop for hybrid or full liquid-cooled deployments. By manufacturing these pods off-site, Schneider Electric can deliver a complete, ready-to-install white space module that arrives move-in ready. Once delivered to a Compass Datacenters campus, the

Read More »

Inside Microsoft’s Global AI Infrastructure: The Fairwater Blueprint for Distributed Supercomputing

Microsoft’s newest AI data center in Wisconsin, known as “Fairwater,” is being framed as far more than a massive, energy-intensive compute hub. The company describes it as a community-scale investment — one that pairs frontier-model training capacity with regional development. Microsoft has prepaid local grid upgrades, partnered with the Root-Pike Watershed Initiative Network to restore nearby wetlands and prairie sites, and launched Wisconsin’s first Datacenter Academy in collaboration with Gateway Technical College, aiming to train more than 1,000 students over the next five years. The company is also highlighting its broader statewide impact: 114,000 residents trained in AI-related skills through Microsoft partners, alongside the opening of a new AI Co-Innovation Lab at the University of Wisconsin–Milwaukee, focused on applying AI in advanced manufacturing. It’s Just One Big, Happy AI Supercomputer… The Fairwater facility is not a conventional, multi-tenant cloud region. It’s engineered to operate as a single, unified AI supercomputer, built around a flat networking fabric that interconnects hundreds of thousands of accelerators. Microsoft says the campus will deliver up to 10× the performance of today’s fastest supercomputers, purpose-built for frontier-model training. Physically, the site encompasses three buildings across 315 acres, totaling 1.2 million square feet of floor area, all supported by 120 miles of medium-voltage underground cable, 72.6 miles of mechanical piping, and 46.6 miles of deep foundation piles. At the rack level, each NVL72 system integrates 72 NVIDIA Blackwell GPUs (GB200), fused together via NVLink/NVSwitch into a single high-bandwidth memory domain capable of 1.8 TB/s GPU-to-GPU throughput and 14 TB of pooled memory per rack. This creates a topology that may appear as independent servers but can be orchestrated as a single, giant accelerator. Microsoft reports that one NVL72 can process up to 865,000 tokens per second. Future Fairwater-class deployments (including those under construction in the UK and Norway)

Read More »

Powering the AI Era: Innovations in Data Center Power Supply Design and Infrastructure

Recently, Data Center Frontier sister publication Electronic Design (ED) released an eBook curated by ED Senior Editor James Morra titled In the Age of AI, A New Playbook for Power Supply Design, with a collection of detailed technology articles focused on understanding the nuts and bolts of delivering power to AI-centric data centers. This compendium explores how the surge in artificial intelligence (AI) workloads is transforming data center power architectures and includes suggestions for addressing the issues. Breaking the Power Barrier As GPUs like NVIDIA’s Blackwell B100 and B200 cross the 1,000-watt threshold per chip, rack power densities are soaring beyond 100 kW, and in some projections, approaching 1 MW per rack. This unprecedented demand is exposing the limits of legacy 12-volt and 48-volt architectures, where inefficient conversion stages and high I²R losses drive up both energy waste and cooling load. Powering the Next Era of AI Infrastructure As AI data centers scale toward multi-megawatt clusters and rack densities approach one megawatt, traditional power architectures are straining under the load. The next frontier of efficiency lies in rethinking how electricity is distributed, converted, and protected inside the rack. From high-voltage DC distribution to wide-bandgap semiconductors and intelligent eFuses, a new generation of technologies is reshaping power delivery for AI. The articles in this report drill down into five core themes driving that transformation: Electronic Fuses (eFuses) for Power Protection Texas Instruments and others are introducing 48-volt-rated eFuses that integrate current sensing, control, and switching into a single device. These allow hot-swapping of AI servers without dangerous inrush currents, enable intelligent fault detection, and can be paralleled to support rack loads exceeding 100 kW. The result: simplified PCB design, improved reliability, and robust support for AI’s steep and dynamic current requirements. The Shift from 48 V to 400–800 V High-Voltage DC (HVDC)

Read More »

Fusion Energy Moves Toward Reality: Strategic Investments by CFS, Google, and Eni Signal Commercial Readiness

Global Fusion Momentum: France, Europe, and a New Competitive Context As CFS, Google, Eni, and Helion press ahead, other fusion efforts worldwide are also making waves, reminding us this is a global race, not a U.S.-exclusive pursuit. In France, the CEA’s WEST tokamak recently achieved a new benchmark by sustaining plasma for more than 22 minutes (1,337 seconds) at ~50 million °C, breaking previous records and demonstrating improved plasma control and stability. That milestone underscores the incremental but essential progress in continuous operation, one of the key prerequisites for any commercially viable fusion system. Meanwhile, ITER, the international flagship built in southern France, continues its slow-but-steady assembly. Despite years of delays and cost overruns, ITER remains central to global fusion ambitions. It’s not expected to produce significant fusion output until the 2030s, but its role in validating large-scale superconducting magnet systems, remote maintenance, tritium breeding, plasma control, and heat management is essential to de-risking downstream commercial fusion designs. Elsewhere in Europe, Proxima Fusion (Germany) is gaining attention. The company is developing a quasi-isodynamic stellarator design and has recently raised €130 million in its Series A, showing that alternative confinement geometries are earning investor support. While that path is more speculative, it adds needed diversity to the fusion technology portfolio. Germany’s Wendelstein 7-X Raises the Bar Germany added another major milestone to the fusion timeline this fall. At the Max Planck Institute for Plasma Physics, researchers operating the Wendelstein 7-X stellarator sustained a high-performance plasma for 43 seconds, setting a new world record for continuous fusion confinement. The run demonstrated stability and control at temperatures exceeding 30 million °C, proving that stellarators, once viewed mainly as scientific curiosities, can now compete head-to-head with tokamaks in performance. Unlike tokamaks, which rely on strong external currents to confine plasma, stellarators use a twisted

Read More »

Microsoft will invest $80B in AI data centers in fiscal 2025

And Microsoft isn’t the only one that is ramping up its investments into AI-enabled data centers. Rival cloud service providers are all investing in either upgrading or opening new data centers to capture a larger chunk of business from developers and users of large language models (LLMs).  In a report published in October 2024, Bloomberg Intelligence estimated that demand for generative AI would push Microsoft, AWS, Google, Oracle, Meta, and Apple would between them devote $200 billion to capex in 2025, up from $110 billion in 2023. Microsoft is one of the biggest spenders, followed closely by Google and AWS, Bloomberg Intelligence said. Its estimate of Microsoft’s capital spending on AI, at $62.4 billion for calendar 2025, is lower than Smith’s claim that the company will invest $80 billion in the fiscal year to June 30, 2025. Both figures, though, are way higher than Microsoft’s 2020 capital expenditure of “just” $17.6 billion. The majority of the increased spending is tied to cloud services and the expansion of AI infrastructure needed to provide compute capacity for OpenAI workloads. Separately, last October Amazon CEO Andy Jassy said his company planned total capex spend of $75 billion in 2024 and even more in 2025, with much of it going to AWS, its cloud computing division.

Read More »

John Deere unveils more autonomous farm machines to address skill labor shortage

Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More Self-driving tractors might be the path to self-driving cars. John Deere has revealed a new line of autonomous machines and tech across agriculture, construction and commercial landscaping. The Moline, Illinois-based John Deere has been in business for 187 years, yet it’s been a regular as a non-tech company showing off technology at the big tech trade show in Las Vegas and is back at CES 2025 with more autonomous tractors and other vehicles. This is not something we usually cover, but John Deere has a lot of data that is interesting in the big picture of tech. The message from the company is that there aren’t enough skilled farm laborers to do the work that its customers need. It’s been a challenge for most of the last two decades, said Jahmy Hindman, CTO at John Deere, in a briefing. Much of the tech will come this fall and after that. He noted that the average farmer in the U.S. is over 58 and works 12 to 18 hours a day to grow food for us. And he said the American Farm Bureau Federation estimates there are roughly 2.4 million farm jobs that need to be filled annually; and the agricultural work force continues to shrink. (This is my hint to the anti-immigration crowd). John Deere’s autonomous 9RX Tractor. Farmers can oversee it using an app. While each of these industries experiences their own set of challenges, a commonality across all is skilled labor availability. In construction, about 80% percent of contractors struggle to find skilled labor. And in commercial landscaping, 86% of landscaping business owners can’t find labor to fill open positions, he said. “They have to figure out how to do

Read More »

2025 playbook for enterprise AI success, from agents to evals

Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More 2025 is poised to be a pivotal year for enterprise AI. The past year has seen rapid innovation, and this year will see the same. This has made it more critical than ever to revisit your AI strategy to stay competitive and create value for your customers. From scaling AI agents to optimizing costs, here are the five critical areas enterprises should prioritize for their AI strategy this year. 1. Agents: the next generation of automation AI agents are no longer theoretical. In 2025, they’re indispensable tools for enterprises looking to streamline operations and enhance customer interactions. Unlike traditional software, agents powered by large language models (LLMs) can make nuanced decisions, navigate complex multi-step tasks, and integrate seamlessly with tools and APIs. At the start of 2024, agents were not ready for prime time, making frustrating mistakes like hallucinating URLs. They started getting better as frontier large language models themselves improved. “Let me put it this way,” said Sam Witteveen, cofounder of Red Dragon, a company that develops agents for companies, and that recently reviewed the 48 agents it built last year. “Interestingly, the ones that we built at the start of the year, a lot of those worked way better at the end of the year just because the models got better.” Witteveen shared this in the video podcast we filmed to discuss these five big trends in detail. Models are getting better and hallucinating less, and they’re also being trained to do agentic tasks. Another feature that the model providers are researching is a way to use the LLM as a judge, and as models get cheaper (something we’ll cover below), companies can use three or more models to

Read More »

OpenAI’s red teaming innovations define new essentials for security leaders in the AI era

Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More OpenAI has taken a more aggressive approach to red teaming than its AI competitors, demonstrating its security teams’ advanced capabilities in two areas: multi-step reinforcement and external red teaming. OpenAI recently released two papers that set a new competitive standard for improving the quality, reliability and safety of AI models in these two techniques and more. The first paper, “OpenAI’s Approach to External Red Teaming for AI Models and Systems,” reports that specialized teams outside the company have proven effective in uncovering vulnerabilities that might otherwise have made it into a released model because in-house testing techniques may have missed them. In the second paper, “Diverse and Effective Red Teaming with Auto-Generated Rewards and Multi-Step Reinforcement Learning,” OpenAI introduces an automated framework that relies on iterative reinforcement learning to generate a broad spectrum of novel, wide-ranging attacks. Going all-in on red teaming pays practical, competitive dividends It’s encouraging to see competitive intensity in red teaming growing among AI companies. When Anthropic released its AI red team guidelines in June of last year, it joined AI providers including Google, Microsoft, Nvidia, OpenAI, and even the U.S.’s National Institute of Standards and Technology (NIST), which all had released red teaming frameworks. Investing heavily in red teaming yields tangible benefits for security leaders in any organization. OpenAI’s paper on external red teaming provides a detailed analysis of how the company strives to create specialized external teams that include cybersecurity and subject matter experts. The goal is to see if knowledgeable external teams can defeat models’ security perimeters and find gaps in their security, biases and controls that prompt-based testing couldn’t find. What makes OpenAI’s recent papers noteworthy is how well they define using human-in-the-middle

Read More »

Can we repair the internet?

From addictive algorithms to exploitative apps, data mining to misinformation, the internet today can be a hazardous place. Books by three influential figures—the intellect behind

Read More »
More Articles