Stay Ahead, Stay ONMINE

GitHub Copilot previews agent mode as market for agentic AI coding tools accelerates

Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More Agentic AI is all the rage today across multiple sectors, including application development and coding. Today at long last, GitHub has joined the agentic AI party with the launch of GitHub Copilot agent mode. The promise […]

Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More


Agentic AI is all the rage today across multiple sectors, including application development and coding.

Today at long last, GitHub has joined the agentic AI party with the launch of GitHub Copilot agent mode. The promise of agentic AI in development is about enabling developers to build more code with just a simple prompt. The new agent mode will enable Copilot to iterate on its own code and fix errors automatically. Looking forward, GitHub is also previewing a fully autonomous software engineering agent, Project Padawan, that can independently handle entire development tasks.

The new agentic AI features mark the latest step in the multi-year evolution of the AI-powered coding development space that GitHub helped to pioneer. The Microsoft-owned GitHub first previewed GitHub Copilot in 2021, with general availability coming in 2022. In the AI world, that’s a long time ago, before ChatGPT became a household name and most people had ever heard the term “generative AI.”

GitHub has been steadily iterating on Copilot. Initially, the service relied on the OpenAI Codex large language model (LLM). In October 2024, users gained the ability to choose from a variety of LLMs, including Anthropic’s Claude, Google’s Gemini 1.5 and OpenAI’s GPT4o. Alongside the agent mode launch, GitHub is now also adding support for Gemini 2.0 Flash and OpenAI’s o3-mini. Microsoft overall has been emphasizing agentic AI, assembling one of the largest AI agent ecosystems in the market.

AI that supports ‘peer programming’

The new GitHub Copilot agent mode service comes as a series of rivals, mostly led by startups, have shaken up the development landscape. Cursor, Replit, Bolt and Lovable are all chasing the growing market for AI-powered development that GitHub helped to create.

When GitHub Copilot first emerged, it was positioned as a pair programming tool, which pairs with a developer. Now, GitHub is leaning into the term peer programming as it embraces agentic AI.

“Developer teams will soon be joined by teams of intelligent, increasingly advanced AI agents that act as peer-programmers for everyday tasks,” said GitHub CEO Thomas Dohmke. “With today’s launch of GitHub Copilot agent mode, developers can generate, refactor and deploy code across the files of any organization’s codebase with a single prompt command.”

Technical breakdown: How GitHub’s new agent architecture works

Since its initial debut, GitHub Copilot has provided a series of core features. Among them is intelligent code completion, which is the ability to suggest code snippets to execute a given function. Copilot also functions as an assistant, allowing developers to input natural language queries to generate code, or get answers about a specific code base. The system, while intelligent, still requires a non-trivial amount of human interaction.

Agent mode goes beyond that. According to GitHub, the platform enables Copilot to iterate on its own output, as well as the results of that output. This can significantly improve results and code output.

Here’s a detailed breakdown of agent mode operation.

Task understanding and planning:

  • When given a prompt, agent mode doesn’t just generate code — it analyzes complete task requirements;
  • According to GitHub, the system can “infer additional tasks that were not specified, but are also necessary for the primary request to work”. 

Iterative execution:

  • The agent iterates on both its own output and the result of that output;
  • It continues iteration until all subtasks are completed.

Self-healing capabilities:

  • Automatically recognizes errors in its output;
  • Can fix identified issues without developer intervention;
  • Analyzes runtime errors and implements corrections;
  • Suggests and executes necessary terminal commands.

Project Padawan brings the ‘force’ to development

While agent mode certainly is more powerful than the basic GitHub Copilot operation, it’s still not quite a fully automated experience.

To get to that full experience, GitHub is previewing Project Padawan. In popular culture, a ‘Padawan’ is a reference to a Jedi apprentice from the Star Wars science fiction franchise. 

Project Padawan builds on the agent mode and extends it with more automation. In a blog post, Dohmke noted that Padawan will allow users to assign an issue to GitHub Copilot, and the agentic AI system will handle the entire task. That task can include code development, setting up a repository and assigning humans to review the final code.

“In a sense, it will be like onboarding Copilot as a contributor to every repository on GitHub,” Dohmke said.

Comparing GitHub’s agent to other agentic AI coding options

GitHub in some respects is a late entrant to the agentic AI coding race.

Cursor AI and Bolt AI debuted their first AI agents in 2023, while Replit released its agent in 2024. Those tools have had over a year to iterate, gain a following and develop brand loyalty.

I personally have been experimenting with Replit agents for the last several months. Just this week, the company brought the technology to its mobile app — which you wouldn’t think is a big deal, but it is. The ability to use a simple prompt, without the need for a full desktop setup to build software, is powerful. Replit’s agent also provides AI prompt tuning to help generate the best possible code. The Replit system runs entirely in the cloud and users like me don’t need to download anything. 

Bolt doesn’t have a mobile app, but it does have a really nice web interface that makes it easy for beginners to get started. Cursor is a bit more bulky in that it involves a download, but it is a powerful tool for professional developers.

So how does GitHub Copilot agent mode compare? GitHub is the de facto standard for code repositories on the internet today. More than 150 million developers, including more than 90% of the Fortune 100 companies, use GitHub. According to the company, more than 77,000 organizations have adopted GitHub Copilot. That makes the technology very sticky. Those organizations already relying heavily on GitHub and Copilot are not going to move away from the technology easily.

In comparison to Replit and Bolt, GitHub Copilot agent mode is not a web-based feature, at least not today. Its preview is currently only available with GitHub Copilot in VS code. That creates a small barrier to entry for absolute newbies for sure, but the reality is also that VS code is arguably the most popular and widely used integrated development environment (IDE).

Developers are a picky bunch. That’s why there are so many different programming languages and frameworks (there seems to be a new JavaScript framework emerging every other month). The bottom line is about comfort and workflow. For existing GitHub Copilot and VS code users, the new agent mode brings a much needed feature that will help improve productivity. For those that aren’t stuck in the GitHub Copilot world, agent mode could very well help bring Github Copilot back into the conversation about which agentic AI-driven coding tool to use.

GitHub Copilot agent mode is currently available in preview and requires VS code insiders, which is intended for early adopters. GitHub has not yet provided any pricing details or a date for general availability.

Shape
Shape
Stay Ahead

Explore More Insights

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

HPE memory server targets compute-heavy and agentic AI workloads

Rather than tying together multiple one- or two- socket servers, the 3250 acts as a large, shared-memory machine, reducing server sprawl and complex data reassembly for target workloads such as SAP HANA, Oracle, and Microsoft SQL Server databases, according to HPE. The server also supports HPE’s Integrated Lights Out (iLO)

Read More »

Lyrie.ai Joins First Batch of Anthropic’s Cyber Verification Program

Dubai-founded OTT Cybersecurity LLC also unveils the Agent Trust Protocol (ATP), the first open cryptographic standard for AI agent identity, scope, and action verification — slated for IETF submission. OTT Cybersecurity LLC, the company behind Lyrie.ai, today announced two milestones that together position the company as foundational infrastructure for the

Read More »

Energy Department Awards Contracts from the Strategic Petroleum Reserve, Advancing President Trump’s Historic Emergency Exchange

WASHINGTON—The U.S. Department of Energy (DOE) announced today the contract awards for the exchange of approximately 53.3 million barrels of crude from the Strategic Petroleum Reserve’s (SPR) Bayou Choctaw, Bryan Mound, Big Hill, and West Hackberry sites. Today’s announcement marks the next phase of DOE’s execution of the United States’ 172-million-barrel contribution to the International Energy Agency’s collective action to stabilize global oil supplies. It follows DOE’s Request for Proposal (RFP) issued at the end of April. Deliveries will begin immediately as the Department continues to move swiftly to address short-term supply disruptions and strengthen U.S. energy security.  “With today’s announcement of contract awards, we are advancing the President’s commitment to carrying out this historic emergency exchange,” said DOE Assistant Secretary of the Hydrocarbons and Geothermal Energy Office Kyle Haustveit. “These actions continue to move oil swiftly into the market, address near-term supply needs, and ensure that the Strategic Petroleum Reserve remains strong through the return of premium barrels.” Under President Trump’s leadership, the Department has executed a historic, record-speed series of SPR exchange solicitations—the largest in the Reserve’s 50-year history—moving critical crude oil supplies quickly to market to address short-term disruptions. President Trump and Secretary Wright are managing the SPR as the critical national security asset it was designed to be, helping stabilize oil markets, protect Americans from supply disruptions, and strengthen energy security at home and abroad. To date, approximately 35 million barrels have been delivered to the market, while President Trump’s historic effort has generated approximately 35 million barrels of additional volume for the SPR at no additional cost to taxpayers.  With these awards, DOE will move forward with an exchange of more than 53.3 million barrels of crude oil, while securing an approximately 28 percent return premium—representing 15.1 million barrels. This action builds on earlier exchange actions, which have already awarded approximately 80 million barrels from the Bayou

Read More »

EIA: US crude inventories down 2.3 million bbl

US crude oil inventories for the week ended May 1, excluding the Strategic Petroleum Reserve, decreased by 2.3 million bbl from the previous week, according to data from the US Energy Information Administration (EIA). At 457.2 million bbl, US crude oil inventories are about 1% above the 5-year average for this time of year, the EIA report indicated. EIA said total motor gasoline inventories decreased by 2.5 million bbl from last week and are about 4% below the 5-year average for this time of year. Finished gasoline inventories increased while blending components inventories decreased last week. Distillate fuel inventories decreased by 1.3 million bbl last week and are about 11% below the 5-year average for this time of year. Propane-propylene inventories decreased by 1.3 million bbl from last week and are 56% above the 5-year average for this time of year, EIA said. US crude oil refinery inputs averaged 16.0 million b/d for the week ended May 1, which was 42,000 b/d less than the previous week’s average. Refineries operated at 90.1% of capacity. Gasoline production decreased, averaging 9.6 million b/d. Distillate fuel production decreased, averaging 4.9 million b/d. US crude oil imports averaged 5.5 million b/d, down 273,000 b/d from the previous week. Over the last 4 weeks, crude oil imports averaged about 5.6 million b/d, 2.4% less than the same 4-week period last year. Total motor gasoline imports averaged 755,000 b/d. Distillate fuel imports averaged 123,000 b/d.

Read More »

Diversified Energy, Carlyle partner in $1.175-billion acquisition in Oklahoma’s Anadarko basin

Diversified Energy Co., in partnership with investment firm Carlyle, have agreed to acquire a portfolio of oil and natural gas properties and related assets in the Anadarko basin of Oklahoma from Camino Natural Resources, Denver, for $1.175 billion. The assets are contiguous with Diversified’s existing Oklahoma assets, creating “an immediate line of sight to future operating efficiencies and general and administrative savings,” the company said in a release May 6. The acquisition is expected to add about 300 MMcfed (51,000 boe/d) of production (about 55% gas, 30% NGLs, 15% oil) for the next 12 and about 1,478 bcfe of reserves. The assets include an approximate 101,000 acres in the SCOOP/STACK/MERGE with over 100 identified, drill-ready locations in an active development area, the company continued. Camino will retain its ownership of the Chickasha development area. The deal will be financed through a bespoke asset-backed securitization (ABS) structured and arranged by Carlyle. A new special purpose vehicle (SPV) will hold the producing assets and issue debt backed by underlying cash flows. Carlyle will hold majority ownership interest in the SPV that issues the ABS, with Diversified serving as asset operator and manager of the ABS. Diversified will retain ownership of the undeveloped assets outside the SPV. Diversified will fund a net amount, inclusive of customary purchase price adjustments, of about $210 million for the acquisition through its senior secured bank facility. The deal is expected to close in this year’s third quarter, subject to customary conditions.

Read More »

Shell increases quarterly earnings amid energy market disruptions

Shell plc reported first-quarter 2026 adjusted earnings of $6.9 billion, up from $5.58 billion in first-quarter 2025 and $3.26 billion in fourth-quarter 2025. “Shell delivered strong results enabled by our relentless focus on operational performance in a quarter marked by unprecedented disruption in global energy markets,” Shell chief executive officer Wael Sawan said in a statement. Shell announced a 5% increase and launched a new $3.0 billion share buyback program expected to be completed by the second-quarter 2026 results announcement. Total shareholder distributions during the quarter amounted to $5.3 billion, including $3.2 billion in share repurchases and $2.1 billion in cash dividends. Net debt increased to $52.6 billion at the end of the first quarter from $45.7 billion at the end of fourth-quarter 2025, reflecting working capital outflow and a $3 billion non-cash net debt increase in variable component of long-term shipping leases in the current macro environment. Shell said commodity price volatility had a significant impact on earnings and working capital during the quarter. The company indicated that unprecedented volatility in commodity prices affected inventory values and receivables, contributing to a working capital impact of $11.2 billion. Gas, LNG operations to reflect declines and Qatar disruptions Integrated Gas adjusted earnings were $1.8 billion in first-quarter 2026, compared with $1.7 billion in fourth-quarter 2025. LNG liquefaction volumes were 7.9 million tonnes, compared with 7.8 million tonnes in the prior quarter, while production averaged 909,000 boe/d, down from 948,000 boe/d in fourth-quarter 2025. QatarEnergy shut in production across all LNG infrastructure on Mar. 2, 2026, and subsequently declared force majeure, although QatarEnergy LNG N(4) was not directly impacted by the attacks. Shell also said damage to Train 2 at the Pearl gas-to-liquids (GTL) plant in Qatar is expected to require about 1 year of repairs.  Stay updated on oil price volatility,

Read More »

Permian Resources leaning on more workovers to ‘kind of hit the gas pedal’

Leaders of Permian Resources Corp., Midland, Tex., plan to continue ramping workover activity to capitalize on higher oil prices and have lifted their full-year oil production guidance by about 2%. Permian Resources roughly doubled its workover rig count during the first quarter and is building on that ramp this quarter, co-chief executive officer (CEO) James Walter told analysts. From replacing hardware on fewer than 40 wells early this year, Permian Resources teams are now completing about 90 workovers per month. That work will lead to a slight increase in spending from early this year but Walter and fellow CEO Will Hickey aren’t committing capital to adding rigs as some of their peers in the Permian basin are adding. “We have the flexibility that, with the existing kind of equipment we’re running today, we can accelerate TILs [turned-in-line] above and beyond what our original base plan highlighted,” Hickey said on a May 7 conference call with analysts and investors. “The general plan today is to really kind of hit the gas pedal but do it within the confines of the equipment we’re running today.” Permian Resources’ average total production totaled nearly 413,000 boe/d (192,300 b/d of oil) in the 3 months ended Mar. 31, which was an increase of more than 10% from first-quarter 2025. Executives credited strong runtime and good performances from new wells for that number, which they’re looking to increase slightly this quarter. The midpoint of the full-year oil production target is now 192,500 b/d versus 189,000 b/d in February. The operator is looking to turn in line 250 wells this year. Hickey said that number could grow by 5% or more by teams “doing the easy things” such as compressing cycle times and drilling more quickly. “Easy is probably understating it,” Hickey joked on the conference call.

Read More »

DOE awards $36 million for Bakken CO₂ EOR program at University of North Dakota

The US Department of Energy’s (DOE) Hydrocarbons and Geothermal Energy Office (HGEO) will invest $36 million to advance carbon dioxide (CO2)–based enhanced oil recovery (EOR) in the Bakken shale through a program led by the University of North Dakota’s Energy & Environmental Research Center. The Bakken Enhanced Oil Recovery–Cracking the Code (Bakken EOR-CC) effort will combine laboratory research, reservoir modeling, artificial intelligence (AI), and field work to evaluate EOR strategies, potentially unlocking billions of additional barrels of incremental oil and extending the life of the state’s coal-fired power plants by utilizing their captured CO2 for EOR, DOE said in a release May 8. Unconventional formations typically yield only about 10% of oil in place, leaving significant volumes unrecovered.  “This program is essential to maximize the full potential of our valuable hydrocarbon resources in the Bakken. By ‘cracking the code,’ these integrated pilot projects will help establish a clear path for the broad commercial deployment of enhanced energy recovery across the nation,” said DOE Assistant Secretary for the Hydrocarbons and Geothermal Energy Office Kyle Haustveit.  Bakken pilot projects  The Bakken EOR-CC program will assess multiple injection strategies, reservoir conditions, and operational approaches across six pilot projects, building a technical basis for commercial-scale EOR in the play.  Bakken EOR-CC will include DOE’s investment of $36 million along with $9 million in cost-share from the University of North Dakota’s Energy & Environmental Research Center and project partners for one of the six pilot projects. State and private sources are providing about $100 million in funding for the remaining five pilot projects. AI and machine-learning tools will be used to analyze performance data, identifying best practices and operational benchmarks across the pilots to accelerate broader deployment. DOE said the integrated pilot effort is intended to create a unified technical foundation that supports wider adoption of EOR

Read More »

Power Takes Center Stage in the AI Infrastructure Race at IMN Data Centers Power Capital 2026

Underwriting the New Risk Stack for AI Power The afternoon session on underwriting power risk widened the lens from whether AI data center demand is real to how investors, operators and suppliers should manage the risks of building into that demand. Moderated by Phil Lookadoo of Haynes Boone, the panel brought together perspectives from Wesco, Generate Capital, King Street and Nuveen Energy Infrastructure Credit. The result was a more granular look at the physical, financial and technology risks now embedded in AI data center development. David Speidelsbach, VP Sales & Strategic Initiatives – Electrical & Electronic Solutions at Wesco, began with the electrical infrastructure implications of AI density. As rack and chip power requirements rise, he said the industry is moving toward higher-voltage DC applications, including 800 VDC distribution. That transition creates new complexity for switchgear, breakers, labor skills and onsite electrical design. Data center power is no longer only about securing megawatts at the site boundary. It is also about safely and reliably distributing far more power inside the building. Sharin Valia, Managing Director of Capital Markets, King Street framed the same issue from inside the data hall. Power density, he said, has changed the nature of future-proofing. A few years ago, 10 kilowatts per cabinet represented a high-density deployment. Now the industry is discussing hundreds of kilowatts per cabinet, with future architectures pushing far beyond legacy assumptions. That creates a basic underwriting question: is a data center being built today already at risk of obsolescence? The answer depends on design flexibility. Valia argued that operators must build thermal and electrical infrastructure capable of supporting mixed-density environments, from lower-density enterprise workloads to high-density liquid-cooled AI deployments. The challenge is not simply building for the highest possible density. It is building a facility that can adapt as tenant requirements change.

Read More »

Supermicro’s New AI Campus Embodies the Industrialization of AI Infrastructure

That positioning places the new San Jose campus at the intersection of three major industry shifts: the localization of critical AI infrastructure manufacturing, the move from server-level integration to rack- and cluster-scale deployment, and the growing importance of liquid cooling as AI systems push beyond conventional enterprise power densities. From Server Manufacturing to AI Infrastructure Integration Supermicro’s historic advantage has been speed. The company built its reputation on a modular “building block” approach, rapidly combining motherboards, chassis, power supplies, processors, GPUs, storage, networking, and cooling into workload-specific systems. That model worked well in the cloud era, when customers prioritized rapid customization. In the AI era, the challenge is larger: integrating scarce GPUs, high-speed networking, liquid cooling, power distribution, and software validation into deployable rack-scale infrastructure. The new campus extends that model beyond individual servers and into the data center itself. Supermicro says the facility will support the full operational chain, including design, manufacturing, testing, service, and global distribution. The result is less a traditional factory than an AI infrastructure staging and validation environment, where liquid-cooled racks can be assembled, tested, and shipped as integrated systems rather than collections of discrete components. According to Supermicro, the San Jose DCBBS campus enables closer collaboration with major customers and suppliers while reducing shipping time and keeping engineering and manufacturing teams tightly aligned. The facility also includes 10 MW of on-campus power capacity, an increasingly important detail as AI rack integration itself becomes power- and cooling-intensive before systems ever reach a customer deployment. That operational shift matters. Traditional server manufacturing relied on factory lines, burn-in rooms, and standardized test environments. AI infrastructure integration increasingly requires something closer to a live data center floor: full-rack validation, coolant loop testing, leak detection, network verification, power sequencing, and thermal performance testing under real operational conditions. In that

Read More »

Utah’s Wonder Valley and the Industrialization of AI Infrastructure

The Power Strategy: Natural Gas as AI’s Fastest Bridge Fuel Recent reporting describes a full buildout approaching 9 GW, with Phase 1 alone projected at roughly 3 GW. That power would reportedly be generated on site through access to the Ruby Pipeline, a 680-mile interstate natural gas transmission system serving the western United States. At that scale, Wonder Valley begins to resemble an independent power development paired with AI infrastructure rather than a conventional data center campus. According to the U.S. Energy Information Administration, Utah’s average electricity demand is roughly 4 GW, meaning the proposed full buildout could eventually rival or exceed the state’s current average load demand. The strategy reflects a broader shift now reshaping hyperscale development: bringing power generation directly to the data center. In some markets that means natural gas turbines. In others it involves fuel cells, geothermal systems, nuclear partnerships, utility-scale battery storage, renewable PPAs, or hybrid models combining on-site generation with grid interconnection. For Utah, natural gas appears to be the most practical near-term option, particularly given the existing Ruby Pipeline infrastructure. Gas generation can typically be deployed faster than new nuclear development, faster than major transmission expansion projects, and at much larger scale than most currently available clean firm-power alternatives. That does not eliminate the risks. Large-scale gas generation carries exposure to fuel-price volatility, emissions scrutiny, air permitting challenges, methane leakage concerns, turbine procurement bottlenecks, and long-term sustainability pressure if hyperscale customers impose increasingly aggressive carbon requirements across their infrastructure supply chains. But across much of the AI infrastructure sector today, speed-to-power has become the overriding priority. Developers have emphasized that the campus would generate its own electricity rather than compete directly with Utah households and businesses for existing utility capacity. That argument carries political weight in a state already grappling with accelerating load

Read More »

PCI group begins work on new spec to support bandwidth-hungry apps like AI, HPC

PCIe is the standard interface for hard drives, networking cards, and graphics cards. That includes GPU based accelerators. PCI-SIG,  the standards body leading the development of the PCIe spec, says it is being designed for workloads like AI/ML, high-speed networking, and edge computing running in hyperscale data centers. “With the increasing data throughput required in AI and other applications, there remains a strong demand for high performance. PCIe technology will continue to deliver a cost-effective, high-bandwidth, and low-latency I/O interconnect to meet industry needs,” said Al Yanes, PCI-SIG president and chairperson in a statement. In theory, PCIe Gen8 NVMe SSDs will be rated for sequential speeds of up to 120,000 MB/s. By contrast, a PCIe 6.0 SSD can reach about 28,000 MB/s sequential read in. Pulse Amplitude Modulation with 4 levels (PAM4) signaling, first added to the PCIe spec in 6.0 as a replacement for non-return-to-zero (NRZ) amplitude, we’ll see further refinement to helpPCIe 8.0 achieve its high transfer rates.

Read More »

AWS hit by US-East-1 outage after data center thermal event

AWS shifted traffic away from the affected zone for most services and warned of longer-than-usual provisioning times. As the evening progressed, the company struggled to bring temperatures down. By 6:47 PM PDT, AWS warned that “Other AWS services that depend on the affected EC2 instances and EBS volumes in this Availability Zone may also experience impairments,” and at 8:06 PM PDT, it conceded that “progress is slower than originally anticipated,” recommending that customers needing immediate recovery restore from EBS snapshots or launch resources in unaffected zones. By 10:11 PM PDT, AWS reported “incremental progress to restore cooling systems” but said users were still “experiencing elevated error rates and latencies for some workflows.” The May 7 incident is not the first time US-EAST-1 has gone down. The region suffered two outages in October 2025, including a 15-hour disruption on October 19 and 20 caused by a race condition in DynamoDB’s automated DNS management system that affected over 70 AWS services and produced cascading failures across Slack, Atlassian, Snapchat, and other dependent services. AWS regions in Ohio have also experienced power-related outages tied to EC2 instances in past years.

Read More »

Lumen advances cloud networking vision with $475M Alkira buy

Lumen puts its total addressable market at approximately $70 billion once Alkira’s international and cloud-to-cloud coverage is included. “Alkira is a bull’s eye in terms of strategic alignment and value creation,” Johnson said. “For Lumen, we expect it to dramatically accelerate our road map execution from years to months.” How the architecture works Alkira operates as a cloud-native, carrier-agnostic control plane. Rather than relying on physical hardware at each interconnection point, it uses a virtual port model that lets enterprises design, deploy and manage network connectivity across clouds, data centers and on-premises environments through a single interface. Alkira is distinct from Lumen’s existing Project Berkeley, which introduces fabric ports for building-to-cloud on-ramp connectivity. “Fabric ports is about enabling building on-prem to be able to connect to the cloud and to be able to grow those services in a cloud economic way,” Johnson said. “The Alkira platform really focuses on the East-West interconnect. So that’s data center-to-data center, cloud-to-cloud, so they operate with more of a virtual port kind of a model, and it’s better together.” Lumen’s Multi-Cloud Gateway bridges the two domains, enabling customers to connect any cloud and any data center over Lumen’s private network. After close, Multi-Cloud Gateway and Alkira together are intended to give customers a single control plane for routing, policy and security across both north-south and east-west connectivity.

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 »