jain.com — data centers, connectivity and security, analyzed daily

Author: Deepak Jain

  • Harness Debuts AI Agents to Fix Vulnerabilities at Machine Speed

    Harness Debuts AI Agents to Fix Vulnerabilities at Machine Speed

    On August 19, 2026, San Francisco-based Harness announced six new security capabilities — AI SAST, LLM Scan Orchestration, a Triage Agent, a Remediation Agent, a Zero-Day Agent, and virtual patching — all available now on its AI Software Delivery Platform. The agents are designed to compress the gap between the roughly six hours attackers now need to weaponize a disclosed vulnerability and the 50-plus days enterprises take on average to fix one.

    The launch landed the same day Palo Alto Networks unveiled its multi-vendor Frontier AI Critical Defense Program to protect critical infrastructure from AI-discovered vulnerabilities, and MarketsandMarkets projected the critical infrastructure protection market will grow from $160.28 billion in 2026 to $206.31 billion by 2031.

    Executive Summary

    Harness is betting that the vulnerability-response problem is no longer a detection problem but a speed problem. Frontier AI models — the most capable large language models — are being used by attackers to find and chain vulnerabilities faster than ever, with first exploits appearing as little as six hours after disclosure. Defenders are gaining the same scanning power: Harness cites Project Glasswing partners surfacing roughly 10 times more vulnerabilities with LLM-based scanning. But more findings without faster remediation just means a bigger backlog.

    The new agents cover the full vulnerability lifecycle inside the delivery pipeline itself: AI SAST pairs deterministic scanning with an AI layer that filters false positives and catches complex flaws like IDOR (insecure direct object references, where an attacker manipulates identifiers to access data they shouldn’t); the Triage Agent prioritizes what is actually exploitable; the Remediation Agent writes, validates, and opens a pull request with a fix; the Zero-Day Agent monitors disclosures around the clock and generates validated fixes often within minutes; and virtual patching shields production immediately with no code changes while the real fix is finished.

    Why it matters: as Harness application-security GM Rahul Sood put it, the same AI models helping customers ship software faster are what attackers use to exploit it faster — and the only way to close that gap is to make security a first-class part of the delivery pipeline rather than a disconnected process. The simultaneous Palo Alto Networks program launch suggests the whole industry has reached the same conclusion on the same day.

    The Six-Hour Exploit Window Breaks the Old Security Model

    The economics of vulnerability management were built on a comfortable assumption: defenders had weeks between a disclosure and real-world exploitation. Harness’s numbers — six hours to first exploit versus more than 50 days to an average fix — show that assumption is dead. When AI can read a vulnerability disclosure and generate a working exploit before most security teams have finished their morning stand-up, any process with human handoffs between scanning, ticketing, triage, and deployment is structurally too slow, regardless of how well each step is staffed.

    This reframes what security products have to sell. For two decades, the pitch was visibility: find more vulnerabilities. Harness’s own framing concedes that visibility now makes things worse — Project Glasswing partners finding 10x more vulnerabilities via LLM scanning simply produces a 10x bigger backlog if remediation speed stays flat. The scarce resource is no longer detection; it is validated, deployable fixes. Products will increasingly be judged on time-from-disclosure-to-deployed-patch, a metric most enterprises today cannot even measure.

    Security Is Collapsing Into the Delivery Pipeline

    Strategically, this launch is a land grab by a DevOps platform into application security territory. Harness’s argument is architectural: standalone scanners produce findings that must cross organizational and tooling boundaries to become fixes, and every boundary adds days. By putting scanning, triage, remediation, and deployment on one platform — with every agent working from the same reachability data, meaning analysis of whether vulnerable code is actually invoked in a given application — Harness claims fixes ship in hours without added headcount. The 2025 Traceable merger, July 2026’s Agent DLC governance launch, and the Kong and Google integrations show this has been a multi-year build, not a feature bolted on for a press cycle.

    The winners and losers logic is straightforward. Platform vendors that own the pipeline (Harness, and by extension GitHub, GitLab, and the cloud providers) gain a structural advantage over point-solution SAST and vulnerability-management vendors, whose findings now have to flow into someone else’s remediation loop. For buyers, the trade-off is the classic platform bargain: faster outcomes and fewer tools to manage, in exchange for deeper dependence on a single vendor.

    A Coordinated Industry Response — and a $206 Billion Market

    Harness did not announce alone. The same morning, Palo Alto Networks introduced the Frontier AI Critical Defense Program, described as a collaboration of leading technology providers to protect critical infrastructure against the rapid rise of AI-discovered vulnerabilities. When the largest pure-play security vendor organizes a multi-vendor defense program on the same day a DevOps platform ships machine-speed remediation agents, the signal is clear: AI-discovered vulnerabilities have moved from a research concern to the organizing threat model of the industry.

    The money follows. MarketsandMarkets projects the critical infrastructure protection market growing from $160.28 billion in 2026 to $206.31 billion by 2031, a 5.2% compound annual growth rate. That is steady rather than explosive growth — but the composition of that spend is what matters. Budgets built around perimeter appliances and manual patch cycles will be re-allocated toward automated response, and vendors positioned on the remediation side of the ledger stand to capture a disproportionate share of it.

    The Trust Problem: Machines Propose, Humans Still Approve

    Harness has kept a human in the loop at the critical moment — the Remediation Agent opens a pull request for a developer to review and approve rather than pushing fixes straight to production. That is the right call for adoption, but it also means the last mile of the process still runs at human speed. If AI agents generate 10x more validated fixes, code review becomes the new bottleneck, and enterprises will face pressure to auto-merge low-risk patches — a governance question this launch raises but does not resolve.

    Virtual patching, which shields production immediately without code changes, is the pragmatic hedge: it buys time at machine speed while humans finish the real fix. The risk to watch is complacency — virtual patches that quietly become permanent, accumulating an invisible layer of compensating controls. The enterprises that win with these tools will be the ones that treat machine-speed response as a bridge to actual remediation, not a substitute for it.

    Background

    Harness began as a continuous-delivery company and has grown into what it brands the AI Software Delivery Platform™ — automating the software lifecycle after code is written, from builds and testing through deployment and cost management. Customers such as United Airlines, Morningstar, and Choice Hotels use it to accelerate releases by up to 75% and cut cloud costs by 60%, and the company is backed by Goldman Sachs, Menlo Ventures, IVP, Unusual Ventures, and Citi Ventures. Its security push dates to the early-2025 merger with API-security firm Traceable and continued through 2026 with Agent DLC governance for AI coding agents and integrations with Kong and Google.

    The market backdrop is an arms race: the same frontier AI models that help developers ship faster let attackers find and chain vulnerabilities in hours, and let defenders surface an order of magnitude more findings than their patching processes were built to absorb. That dynamic — visibility outrunning remediation — is driving both vendor consolidation around delivery pipelines and industry-wide efforts like Palo Alto Networks’ new Frontier AI Critical Defense Program.

    Source: Harness Launches AI Agents for Machine-Speed Vulnerability Response — Harness press release via PR Newswire, August 19, 2026, with same-day context from Palo Alto Networks’ Frontier AI Critical Defense Program announcement and MarketsandMarkets’ critical infrastructure protection market forecast.

  • OpenAI’s 8 GW Ohio Campus: Legacy Grid Covers 800 MW, New Power Plants Gate the Rest

    OpenAI’s 8 GW Ohio Campus: Legacy Grid Covers 800 MW, New Power Plants Gate the Rest

    TL;DR · 30-second read

    The Short Version

    • OpenAI, the maker of ChatGPT, will rent space in a huge computer campus planned on the grounds of an old Cold War uranium plant in rural southern Ohio.
    • At full size, the campus would use about as much electricity as eight large nuclear reactors produce.
    • Only about a tenth of that can come from today’s power lines, starting in 2028. The rest needs new power plants, including gas-fired ones, that have not been built yet.
    • The developer promises to pay for the grid upgrades itself, so local electric bills do not cover them.

    OpenAI announced on August 17 that it has agreed to secure approximately 8 gigawatts of IT capacity at the PORTS-Pike Technology Campus in Pike County, Ohio, working with developer SB Energy, NVIDIA and the U.S. Department of Energy. The campus is being developed on private land and on remediated federal land at the former Portsmouth Gaseous Diffusion Plant. SB Energy will build, own and operate the data center under a 20-year lease to OpenAI, and the site will exclusively host NVIDIA AI computing hardware.

    The first 800 megawatts are expected to be available in 2028, largely using existing AEP infrastructure. Later phases will require new grid-connected power plants, including natural gas generation, plus new transmission lines. NVIDIA will invest $1.5 billion in SB Energy and provide credit support for the land, power and shell buildout of the initial 4.25 gigawatts. The companies project 35,000 construction jobs over a six-year buildout through 2032 and 2,500 long-term operating jobs.

    Executive Summary

    The PORTS-Pike agreement is one of the largest single-site AI capacity commitments announced to date: roughly 8 gigawatts of power delivered to computing equipment, on a former uranium enrichment site in southern Ohio. OpenAI is the tenant, SB Energy is the developer and owner, NVIDIA is both the exclusive hardware supplier and a financial backer, and the Department of Energy is contributing federal land and an existing onsite water system.

    The more telling number is the smaller one. Only the first 800 megawatts, about a tenth of the headline figure, is expected to come online in 2028 on largely existing grid infrastructure. Everything after that depends on new power plants and new transmission lines, along with permits, environmental reviews and financing that the companies say must still be in place. In practice, this is a power-development project with a data center attached, and its pace will be set by generation and wires rather than by chips.

    The package also includes a pledge that SB Energy will pay the full cost of grid upgrades and new transmission rather than passing them to Ohio ratepayers, a closed-loop cooling design meant to limit ongoing water use, and more than $160 million in community grants and student software credits.

    Nine-Tenths of the Campus Waits on Power That Does Not Yet Exist

    The headline figure is 8 gigawatts-IT, a measure of the electricity delivered to servers and networking gear rather than the site’s total draw, which is higher once cooling and power conversion are counted. Against that, the companies say the first 800 megawatts should be available in 2028 “largely using existing AEP infrastructure,” meaning the transmission and substation capacity American Electric Power already has in the area. That is roughly 10% of the campus. The remaining 7 gigawatts or so, per OpenAI’s announcement, “will require new power plants connected to the grid, including natural gas generation, as well as new transmission lines and associated infrastructure.”

    That split is the operational core of the deal. Existing wires are the fastest path to power for any large data center, because new generation and new high-voltage lines each require siting, permitting, interconnection studies and equipment procurement before a single rack can be energised. Gas turbines, transformers and high-voltage switchgear have all been in heavy demand as utilities and data center developers compete for the same supply chains. The six-year buildout window through 2032 signals that the companies expect most of the capacity to arrive only as that new infrastructure is completed.

    The parties most exposed to this sequencing are OpenAI, whose compute roadmap depends on the later phases; SB Energy, which carries the development risk; AEP and PJM, the regional grid operator covering Ohio, which must study and integrate the new plants and lines; and NVIDIA, whose hardware sales at the site scale with each tranche of power that comes online.

    Why a Former Uranium Plant Works as a Power-First AI Site

    The Portsmouth Gaseous Diffusion Plant was built in the 1950s to enrich uranium, a process that consumed enormous amounts of electricity. That history left the area with heavy grid infrastructure, a large industrial footprint and federal land that has been under cleanup for years. For an AI campus, those legacy assets matter more than proximity to cities or fiber hubs: they are what allows the first 800 megawatts to arrive in 2028 instead of later.

    The Department of Energy’s role extends beyond land. PORTS-Pike is working with DOE to use the site’s existing onsite water system, and the data center will use closed-loop, air-cooled systems that recirculate water rather than evaporative cooling towers, which continuously consume water. OpenAI says ongoing water use, once the system is filled, should be comparable to an office building supporting a similar number of people, and well below what the enrichment plant historically used. It has committed to publish expected water use once the design is final.

    The model is replicable only where similar conditions exist: large parcels of already-disturbed land, legacy high-voltage infrastructure, and a federal owner willing to redevelop. Retired industrial and federal sites with those traits are a finite set, which is why they are becoming contested real estate for AI developers.

    Who Carries the Financial Risk

    The deal is structured to keep capital off OpenAI’s balance sheet up front. SB Energy builds and owns the facility, and OpenAI begins paying only as completed capacity becomes available for lease, funding those payments from revenue, cash flow and investor capital. That protects OpenAI from paying for capacity that is delayed, but it shifts construction and power-development risk to SB Energy and its financiers.

    NVIDIA fills part of that gap. Its $1.5 billion investment in SB Energy and its credit support, a guarantee-like backing that lowers borrowing costs, cover land, power and shell buildout for the initial 4.25 gigawatts. Because the site will exclusively host NVIDIA hardware, the chipmaker is effectively helping finance demand for its own products. That aligns incentives among the three companies, but it also concentrates exposure: the project’s financing, customer and supplier are tightly linked, and the backing described covers just over half of the 8 gigawatts.

    The Ratepayer Pledge Addresses the Biggest Local Objection

    Data center growth has become a flashpoint over whether households end up paying for grid expansion that serves large corporate loads. SB Energy’s commitment to pay the full cost of grid upgrades and new transmission lines, without shifting those costs to Ohio ratepayers, speaks directly to that concern, and the companies argue the new lines will also strengthen regional reliability.

    The community package, which includes $40 million from OpenAI on top of SB Energy’s earlier $40 million and $84 million in Codex credits for about 844,000 Ohio students, adds up to more than $160 million by OpenAI’s count. Alongside a union memorandum of understanding and annual public reporting on hiring, water and energy use, it is designed to build local consent for a project whose largest phases will require new permits and reviews over the coming years.

    Background

    The Portsmouth Gaseous Diffusion Plant near Piketon, Ohio, was built in the 1950s to enrich uranium for national defense and later for nuclear power fuel. Gaseous diffusion was extremely electricity-intensive, which left the region with substantial grid infrastructure. Enrichment ended in the early 2000s, and the Department of Energy has since overseen decontamination and cleanup, with parts of the site now available for redevelopment.

    SB Energy, a SoftBank-backed energy and infrastructure developer, is leading development of the PORTS-Pike campus. OpenAI has been expanding its computing capacity through long-term agreements with developers and hardware partners to support training of frontier AI models and growing demand for products such as ChatGPT and Codex.

    Sources

    Source: OpenAI joins PORTS-Pike project, OpenAI’s announcement of an approximately 8 gigawatt capacity agreement with SB Energy, NVIDIA and the U.S. Department of Energy in Pike County, Ohio.

  • Quantum-GUARD’s One-Year Deadlines Put Grid Encryption on AI’s Power Risk List

    Quantum-GUARD’s One-Year Deadlines Put Grid Encryption on AI’s Power Risk List

    TL;DR · 30-second read

    The Short Version

    Two senators, one from each party, have proposed a law to prepare the power grid for quantum computers. These are a new kind of machine that could one day crack the digital locks protecting the grid’s control systems.

    The bill gives the Energy Department one year to study the danger and set up a test lab where power companies can try stronger locks.

    Why it matters: the huge computer warehouses that run artificial intelligence depend on the grid. If the grid’s security slips, their electricity is at risk too.

    Legis1 reported that Sens. Chris Coons (D-DE) and Mike Rounds (R-SD) introduced the Quantum Grid Utility Assurance and Resilient Defense Act, or Quantum-GUARD Act, on August 14, 2026. The bill targets a future threat: quantum computers powerful enough to break the encryption that protects the electric grid’s IT and control systems.

    The bill would require the Federal Energy Regulatory Commission (FERC) to weigh quantum risks when it reviews grid reliability standards. It would give the Department of Energy one year to stand up a post-quantum cryptography testing sandbox with utilities, grid operators and vendors, and one year to report to Congress on quantum risks to the bulk-power system. IonQ, American Binary, the Quantum Industry Coalition, the Quantum Economic Development Consortium and the Cyber Threat Alliance endorsed it.

    Executive Summary

    The Quantum-GUARD Act is the first bill aimed specifically at moving the electric grid onto post-quantum cryptography. These are encryption methods designed to withstand attacks from quantum computers. The bill does not order utilities to replace anything. It directs regulators to study the problem, build a joint testing environment and factor quantum risk into the rules that govern grid reliability. Sen. Rounds described it as codifying parts of President Trump’s June executive order on advanced cryptographic attacks, which set a federal policy of moving government systems to NIST-approved post-quantum standards and helping critical infrastructure owners do the same.

    For the data center industry, the significance lies in the dependency, not the quantum computers themselves. AI campuses are among the largest new customers of the grid, and their uptime rests on utility control systems they neither own nor secure. Encryption upgrades to those systems are slow to plan and deploy. A bill that starts one-year federal clocks on assessment and testing marks the point at which grid cryptography becomes something power-hungry operators should track alongside capacity, interconnection and price.

    Why Grid Encryption Belongs on the AI Power Risk List

    A data center’s power is only as dependable as the systems that run the grid behind it. Those systems include substation controls, generator dispatch and the communications between control rooms and field equipment, which engineers call operational technology, or OT. Much of that traffic is protected by public-key cryptography, the same family of math that quantum computers are expected to threaten. The Quantum-GUARD Act names this directly: it would require FERC to consider post-quantum cryptography in both IT and OT systems, and it tasks the Energy Department with identifying obstacles to moving ‘high-value systems’ to new standards.

    The one-year deadlines explain why operators should track this now rather than later. Under the bill, DOE’s Office of Cybersecurity, Energy Security, and Emergency Response would have twelve months to launch a testing sandbox and twelve months to deliver findings to the Senate Energy and Natural Resources and House Energy and Commerce committees. Those are the first steps of a migration, not the end of one. Grid control equipment often stays in service far longer than office IT, so any cryptographic change moves through procurement, testing and field replacement over years. For a company planning an AI campus that depends on a particular utility for a decade or more, how that utility handles the transition is a real long-term risk, even though the bill never mentions data centers.

    In practice, this is a due-diligence item rather than an alarm. Large-load customers already question utilities about capacity, redundancy and restoration times. The bill’s framework makes it reasonable to add a question about cryptographic readiness, and to expect that the answer will be shaped by whatever the DOE study and sandbox produce.

    What the Bill Does, and What It Leaves Open

    The bill’s verbs matter. FERC would have to ‘consider’ quantum risk when reviewing proposed reliability standards. That makes quantum risk a factor in rulemaking but sets no deadline for migration. DOE would ‘study’ and ‘establish’ a sandbox, with no requirement that utilities adopt what it finds. This is a framework bill. It builds the evidence base that later mandates, if any, would rest on.

    That restraint has advantages. Utilities run a very wide mix of legacy and modern equipment, and a mandate issued before anyone has tested post-quantum algorithms on real grid hardware could cause more reliability problems than it solves. The sandbox exists to test that, bringing utilities, vendors, federal agencies and state and local organizations into one environment to find where new cryptography breaks old equipment. The cost is time. By design, the bill’s first concrete output is a report, and the harder questions of who pays, on what schedule and under what enforcement come after it.

    Why Now: The Standards Exist, the Threat Does Not Yet

    The timing follows the standards. NIST finalized its first three post-quantum cryptography standards in August 2024 and has encouraged organizations to begin transitioning, which means there is now a defined target to migrate toward. Before that, a grid-specific migration bill would have had nothing concrete to point at.

    The threat, by contrast, is still prospective. The bill is framed around protecting the grid ‘before sufficiently powerful quantum computers can undermine existing encryption,’ and it does not say when that might happen. That is the right way to frame it. The argument for starting early rests on how slowly large infrastructure changes its cryptography, not on any fixed date for when quantum machines will arrive. Readers should treat claims of an imminent quantum break, from any side, with the same caution the bill’s own wording shows.

    A Quantum Industry Coalition Behind a Grid Bill

    The endorsements come mainly from the quantum and cybersecurity sectors: IonQ, a quantum technology company; American Binary; the Quantum Industry Coalition; the Quantum Economic Development Consortium; and the Cyber Threat Alliance. Support from companies that build quantum and security technology is expected and does not weaken the bill’s reasoning, which rests on NIST’s published standards and an existing executive order. It does mean that the people who would carry out the migration, meaning utilities, grid operators and state regulators whose ratepayers ultimately fund utility upgrades, will shape how practical the plan turns out to be, and their public positions on the bill are worth watching.

    Background

    The US electric grid runs under mandatory reliability standards, including cybersecurity rules. These are largely developed by the North American Electric Reliability Corporation and approved by FERC. The Department of Energy’s Office of Cybersecurity, Energy Security, and Emergency Response (CESER) leads the department’s work on energy-sector security. Grid cybersecurity has historically focused on present-day threats such as intrusions and ransomware. Quantum computing adds a longer-horizon concern: that a sufficiently powerful machine could defeat the public-key encryption that protects much of the grid’s communications.

    NIST’s finalization of its first three post-quantum cryptography standards in August 2024 gave organizations a concrete target for migration. Federal policy has since moved toward requiring that transition for government systems and supporting it for critical infrastructure. Meanwhile, rapid data center construction, driven largely by AI, has made grid reliability a central concern for the digital infrastructure industry.

    Sources

    Source: Bipartisan Bill Targets Cyber Threats to Electric Grid, Legis1’s report on the Quantum-GUARD Act introduced by Sens. Coons and Rounds.

  • Nvidia’s 4.25-GW Ohio Guarantee Puts Chipmaker Credit Behind AI Land and Power

    Nvidia’s 4.25-GW Ohio Guarantee Puts Chipmaker Credit Behind AI Land and Power

    TL;DR · 30-second read

    The Short Version

    Nvidia, the company whose chips run most of today’s artificial intelligence, has agreed to financially vouch for the land, power and buildings of a huge computing campus in southern Ohio. The campus is on the site of a closed Cold War uranium plant.

    In return, only Nvidia’s equipment can go inside. OpenAI, the maker of ChatGPT, has signed a 20-year lease to use it.

    At full size, the computers would use roughly the output of eight large nuclear reactors. The first sections are due in 2028. The unusual part is the financing: a chip seller is helping pay for the buildings its chips will sit in.

    Nvidia announced on August 17, 2026 that it will provide credit support for the land, power and shell buildout at SB Energy’s PORTS-Pike Technology Campus in Pike County, Ohio. In this context, credit support means Nvidia’s financial backing stands behind the project’s obligations. The support secures an initial 4.25 gigawatts of IT capacity, meaning power delivered to computing equipment, and Nvidia holds an option on the remaining 3.75 IT-GW. In exchange, Nvidia will be the exclusive AI compute provider on the site. OpenAI will be the customer for the full 8 IT-GW. SB Energy will build, own and operate the data center under a 20-year lease to OpenAI.

    Nvidia will also invest $1.5 billion in SB Energy, joining existing investors SoftBank Group and OpenAI. SB Energy and SoftBank committed to build at least 10 gigawatts of new generation and to invest at least $4.2 billion in regional grid infrastructure with AEP Ohio. Capacity is expected to come online in phases beginning in 2028.

    Executive Summary

    The headline numbers are 8 gigawatts of AI capacity, a 20-year lease and a $1.5 billion equity check. The structural news is who is standing behind the real estate. Data center developers typically finance land, power hookups and building shells against the creditworthiness of the tenant. At PORTS-Pike, Nvidia is supplying credit support for the first 4.25 IT-GW. That makes the chipmaker a financial backer of the physical site as well as the vendor of the hardware inside it.

    What Nvidia gets in return is explicit: exclusivity. Every AI system on the campus will be Nvidia’s, deployed on its DSX AI factory platform of GPUs, CPUs and networking. Jensen Huang framed it as securing “long-lived infrastructure for NVIDIA compute” that can be upgraded with each new chip generation. That implies a site designed to take several future Nvidia generations over the life of the lease.

    The deal matters beyond Ohio for a simple reason. If a chip supplier’s balance sheet can make a gigawatt-scale site financeable, then access to land, power and shells becomes something hardware vendors compete for directly, not just something their customers arrange.

    Why a Chipmaker Is Backing Buildings

    A data center campus has three expensive layers that come before any chip is installed: land, power (generation, substations and grid connections) and the shell, meaning the building, cooling plant and electrical rooms. The industry abbreviates these as LPS. Whoever finances that layer wants confidence that someone creditworthy will keep paying for the space for the life of the investment. At PORTS-Pike, SB Energy owns the asset and OpenAI is the 20-year tenant. Nvidia is adding a third layer of assurance: credit support on the LPS buildout covering the first 4.25 IT-GW.

    That is the mechanism behind the headline. Nvidia is using its own credit to reduce the financing risk of a site it does not own. In return, it gets a guaranteed home for its hardware at a scale of more than four gigawatts, with an option on 3.75 more. Nvidia’s statement makes the logic explicit: “land, power and shell have become vital in the age of AI.” A supplier that is worried about customers finding buildings and electricity for its systems fast enough is now helping underwrite the buildings and electricity itself.

    Several groups are affected. Developers gain a new kind of counterparty whose backing may unlock project finance. Lenders get another large name standing behind the obligations. Competing accelerator vendors face a campus that is closed to them by contract. For OpenAI, the arrangement ties a very large block of future capacity to a single supplier’s roadmap. This is one deal, not yet a market pattern. It does show that chipmaker credit can serve as a financing input for AI sites, and other developers will take note.

    Exclusivity, Upgrades and Concentration

    The exclusivity clause is the price of the guarantee, and it cuts both ways. Nvidia describes the site as upgradeable “with each new generation delivering more intelligence and better economics.” A 20-year lease spans many hardware cycles, so the value to Nvidia is not a single sale. It is a standing claim on refresh after refresh. For OpenAI, a site built around one vendor’s full stack can simplify design and speed deployment. It also narrows negotiating leverage over future generations at this location.

    A fair question is how tightly interlinked the parties now are. SoftBank Group and OpenAI were already investors in SB Energy. Nvidia is now investing $1.5 billion and backing the site’s buildout. OpenAI is the tenant and will run Nvidia hardware. Arrangements like this are not unusual in early infrastructure markets, and nothing in the announcement suggests anything improper. Still, it means the campus’s economics rest on a small group of counterparties whose fortunes are correlated with continued AI demand. The announcement asserts that demand is “growing at an extraordinary pace.” Investors will want to see it in contracted revenue as phases come online.

    Power-First, With Overhead Built In

    SB Energy and SoftBank committed to at least 10 gigawatts of new generation to support 8 IT-GW of AI capacity. The gap between those numbers is instructive. Roughly a fifth of the generation is not delivered to computing equipment. It covers cooling, power conversion and other facility overhead, plus any margin the developers are holding. That is a useful reminder that AI capacity figures quoted in IT-GW understate the electricity a campus actually requires.

    The grid commitment is at least $4.2 billion in regional infrastructure through a partnership with AEP Ohio that the companies say is “designed to protect ratepayers.” The release does not explain how that protection works: who pays for which upgrades, and what happens to those costs if demand falls short. Those questions matter to Ohio households as much as to investors. Reusing the decommissioned Portsmouth Gaseous Diffusion Plant site, with the Department of Energy and Department of Commerce involved, may offer existing industrial land and transmission advantages. Spanning federal and private land also adds approval steps that the companies have not detailed.

    Background

    SB Energy describes itself as an integrated data center and power infrastructure company that develops, builds and operates gigawatt-scale data center campuses and utility-scale power generation. Its investors before this deal included SoftBank Group and OpenAI. Its pitch is a “power-first” model: securing electricity before building compute, on the view that power is the industry’s primary bottleneck.

    The PORTS-Pike site sits on the former Portsmouth Gaseous Diffusion Plant in Appalachian Ohio. The plant was a Cold War-era uranium enrichment facility that has since been decommissioned. Redeveloping it as an AI campus places some of the country’s largest planned computing capacity in a region that SB Energy says has long shaped American industry, and it involves both federal and private land.

    Sources

    Source: NVIDIA Guarantees SB Energy’s PORTS-Pike Technology Campus in Ohio to Exclusively Host NVIDIA AI Compute. Nvidia Newsroom, August 17, 2026: Nvidia’s announcement of credit support, exclusivity and a $1.5 billion investment in SB Energy for OpenAI’s 8 IT-GW Ohio campus.

  • Synergy: Neocloud Revenues Growing 200%+ a Year, Headed for $180B by 2030

    Synergy: Neocloud Revenues Growing 200%+ a Year, Headed for $180B by 2030

    Synergy Research Group reported on August 17, 2026 that “neoclouds” — the emerging tier of specialized GPU cloud providers built for AI workloads — are currently growing revenues at more than 200% per year. On that trajectory, Synergy forecasts the segment will reach $180 billion in annual revenues by 2030.

    Executive Summary

    Synergy Research Group, a market intelligence firm that has tracked cloud and data center markets for decades, put a striking pair of numbers on one of the fastest-moving corners of the infrastructure industry: neocloud providers are more than tripling their revenues each year, and the category is projected to become a $180 billion market by 2030.

    The forecast matters because it treats neoclouds not as a temporary arbitrage on scarce GPUs, but as a durable market tier alongside the hyperscale clouds. If Synergy is right, a business model that barely existed three years ago will, within four years, rival the size of the entire global colocation industry — with all the capital, power, and data center demand that implies. It is worth noting the syndicated item we reviewed carries the headline figures but not Synergy’s full methodology, so the underlying assumptions deserve scrutiny alongside the projection itself.

    What a Neocloud Is — and Why the Category Exists

    “Neocloud” is the industry’s shorthand for cloud providers built specifically around GPU compute for artificial intelligence — renting out clusters of accelerators for model training and inference rather than offering the sprawling general-purpose service catalogs of AWS, Microsoft Azure, or Google Cloud. Commonly cited players in the category include CoreWeave, Lambda, Nebius, and Crusoe, though Synergy’s specific inclusion list is not visible in the syndicated item.

    The category exists because AI demand outran what the traditional clouds could supply. Training frontier models requires dense, tightly networked GPU clusters, exotic power and cooling footprints, and pricing models closer to industrial capacity contracts than to on-demand virtual machines. Specialists that could secure GPUs, power, and data center space quickly found a seller’s market waiting for them.

    The Economics Behind 200% Growth

    Growth above 200% per year is extraordinary, but the arithmetic behind it is straightforward: the segment started from a small base, and demand for AI compute currently exceeds supply. When capacity sells out before it is built, revenue growth tracks how fast a provider can energize new data center capacity — which is why the neocloud story is inseparable from the power and data center construction booms.

    The harder question is margin durability. Neocloud economics rest on expensive, fast-depreciating hardware, heavy debt financing in many cases, and — for several prominent players — revenue concentrated in a small number of very large AI customers. A $180 billion revenue projection says the market will be big; it does not by itself say the businesses in it will be uniformly profitable. Investors should distinguish between the size of the pie and the quality of any individual slice.

    Winners, Losers, and the Hyperscaler Question

    For data center operators, utilities, and connectivity providers, the forecast is almost unambiguously bullish: neoclouds are among the largest lessees of wholesale data center capacity and the most aggressive buyers of power. A tier growing toward $180 billion in revenue implies sustained demand for the physical layer beneath it — sites, substations, fiber, and cooling.

    For the hyperscalers, the picture is more nuanced. Neoclouds are simultaneously competitors for AI workloads and, in some well-publicized arrangements across the industry, suppliers of capacity to the hyperscalers themselves. Whether the big clouds ultimately reabsorb this demand as their own GPU fleets scale, or the neocloud tier keeps a permanent structural advantage in speed and specialization, is the central competitive question the next few years will answer.

    Can the Curve Hold to 2030?

    Extending any 200% growth rate for years produces implausible numbers, and Synergy’s own forecast implies significant deceleration: a market compounding at 200% would blow far past $180 billion by 2030 from almost any plausible base. Read properly, the projection assumes today’s hypergrowth cools into merely strong growth — a reasonable but assumption-laden path.

    The risks to the curve are the familiar ones for AI infrastructure: whether enterprise AI spending keeps converting into paid compute at current rates, whether power availability constrains buildouts, how quickly GPU generations depreciate, and whether customer concentration turns any single buyer’s pullback into a segment-wide shock. None of these invalidate the forecast; all of them are the difference between the projection and the outcome.

    Background

    The neocloud category rose to prominence after 2023, when generative AI demand created acute scarcity in GPU compute and a wave of specialists — several of them former cryptocurrency miners repurposing power-rich sites — pivoted to renting AI capacity. The segment has since attracted tens of billions of dollars in capital and become one of the largest sources of demand in the data center leasing market. Synergy Research Group, which has long published the benchmark market-share data for cloud infrastructure services, tracking the rise of AWS, Microsoft, and Google, now treats this GPU-specialist tier as a distinct market worth forecasting in its own right — itself a signal of how the AI buildout is restructuring cloud economics.

    Source: Neoclouds Currently Growing by Over 200% per Year; Will Reach $180 Billion in Revenues by 2030 — Synergy Research Group, a market forecast for the GPU-specialist cloud segment published August 17, 2026.

  • Iloilo’s Stalled 50 MW AI Campus Shows Community Consent Is a Siting Input

    Iloilo’s Stalled 50 MW AI Campus Shows Community Consent Is a Siting Input

    TL;DR · 30-second read

    The Short Version

    A company calling itself Norway Green Energy wanted to build a large artificial intelligence computing center in Oton, a town in the central Philippines. The site covers about 10 hectares, roughly 14 soccer fields. At full load it would use about as much electricity as a small city.

    At a public meeting on August 14, residents pushed back. The project’s presenter walked out and said there was no reason to keep talking.

    The lesson: these giant computer buildings need more than electricity and land. They also need neighbors who agree to host them.

    A public consultation on a proposed 50-megawatt AI data center in Oton, Iloilo, ended without resolution on Friday, August 14, 2026. The Manila Times reported that the project’s Norwegian proponent, Daniel Stefan Robertsen, walked out of the meeting amid mounting opposition from residents and declared there was no reason to continue the discussion.

    Robertsen had presented the project under the name Norway Green Energy. The plan was a 10-hectare hyperscale AI data center campus in Barangay Abilay Sur, Oton. A barangay is the smallest unit of local government in the Philippines, roughly equivalent to a village or neighborhood. The project’s status is now uncertain.

    Executive Summary

    The proposal was substantial for a provincial site: 50 megawatts of capacity on a 10-hectare campus, described as hyperscale. That term usually means facilities built to the scale and specifications of the largest cloud and AI operators. The consultation was meant to put the project before the people who would live beside it. It ended with the proponent leaving and no agreed path forward.

    The episode matters beyond Oton because it shows where an AI data center project can stall before any construction starts. Developers typically screen sites for grid power, fiber connectivity, land and water. The Oton consultation shows that local acceptance belongs on the same list. A project that cannot get through a public consultation has an unresolved siting input, however strong its power or land position.

    Consent Works Like Power: Without It, the Site Doesn’t Close

    Data center developers rank candidate sites by the inputs they cannot build without. These include deliverable grid power, a fiber route, buildable land, and water or an alternative cooling design. Each input gates financing. Lenders and prospective tenants want evidence that it is secured before capital is committed. Local consent works the same way, even though it rarely appears on a site-selection scorecard. In the Philippines, a project of this kind generally needs local clearances, starting at the barangay level and running through municipal zoning and business permits. Projects with potential environmental effects may also require clearance under the national environmental impact assessment system. The public consultation is where affected residents’ positions enter that process.

    That is why the Oton meeting matters. A proposal for 50 megawatts on 10 hectares is a large footprint for one barangay. A consultation that ends with the proponent declaring there is no reason to continue leaves the question of local acceptance unresolved. Until that changes, the site lacks one of the inputs every later stage depends on. The developer, any prospective tenant, any lender, and the Oton local government are all left waiting on the same open item.

    The lesson for the wider AI buildout is modest but concrete. When a developer treats consultation as a formality to get through, the siting schedule assumes an outcome the community has not yet given. Iloilo is a single case and does not by itself establish a regional trend. It does show the mechanism clearly.

    Why 50 Megawatts Draws Scrutiny in a Provincial Setting

    Fifty megawatts is a meaningful load anywhere. Run at full capacity for a year, it comes to about 438 gigawatt-hours of electricity. AI data centers also draw power differently from homes or shops. They run close to flat around the clock, so the demand does not ease overnight. In a metropolitan market, that load is absorbed into a large, dense grid. At a provincial site, the same load is a bigger share of the local system, and residents can reasonably ask how it will be supplied.

    Communities facing AI data center proposals usually ask a consistent set of questions. Where will the power come from? Will it affect local reliability or rates? How will the servers be cooled, and does that use water? What will the site sound and look like? How many local jobs will it create? The name Norway Green Energy suggests a renewable-energy angle. However, no generation plan was set out publicly. A consultation is where a developer can answer those questions in detail, and it is the setting in which concerns can turn into conditions rather than into outright opposition.

    What a Walkout Does to the Project’s Options

    Leaving a consultation does not by itself end a project. It does narrow the options. A developer in this position can seek a new consultation with a revised proposal, change the design to address objections, or look for a different site. Each option costs time. The last one discards whatever work went into securing land and local relationships in Abilay Sur.

    The claim worth examining is the implied one: that the discussion had nothing left to offer. From a siting perspective, the opposite is usually true. A consultation that surfaces strong objections tells the developer exactly which conditions it would have to meet, whether on power sourcing, water use, noise or community benefits. Stopping the conversation leaves those conditions undefined, and that is harder to finance around than a known set of commitments.

    For the Philippines’ broader push to attract AI and cloud infrastructure beyond Metro Manila, the episode is a useful early data point. Provincial sites can offer cheaper and larger land parcels. But the host community is closer to the project there, and its consent is part of the site’s value, not a step to schedule around.

    Background

    The Philippines has become a growing market for data centers, driven by cloud adoption, digital services and, more recently, demand for AI computing capacity. Most existing capacity is concentrated in and around Metro Manila, where grid infrastructure, fiber routes and corporate customers are densest. Developers looking for larger and cheaper sites have begun to consider provincial locations.

    Iloilo province, on the island of Panay in the Western Visayas, is anchored by Iloilo City, one of the region’s main commercial hubs. Oton is an adjacent municipality. The proposed campus in Barangay Abilay Sur would have been one of the larger AI-oriented data center projects proposed outside the capital region. That scale is one reason local consultation carried so much weight.

    Sources

    Source: Norwegian proponent’s walkout halts Iloilo AI data center plans (The Manila Times). A public consultation on a proposed 50-megawatt AI data center in Oton, Iloilo, ended in uncertainty after its proponent walked out amid resident opposition.

  • Hack-Back Memo Puts State Hackers Out of Scope, Despite Attacks on 45 Water Systems

    Hack-Back Memo Puts State Hackers Out of Scope, Despite Attacks on 45 Water Systems

    TL;DR · 30-second read

    The Short Version

    The American government will now let approved private companies break into, and even wreck, computers run by foreign criminal hacking gangs. In March, officials said they would not do this.

    Each company must set aside at least $1 million, which it loses if it breaks the rules. Every attack needs written approval from the Justice Department and Homeland Security.

    The catch: hackers working directly for a foreign government are off-limits. That matters because the order came after suspected Iranian attacks on water suppliers in 45 American towns.

    Tom’s Hardware reported that President Donald Trump signed a presidential memorandum on August 12 creating the first US program that lets vetted private companies conduct offensive cyber operations against foreign cybercrime organizations. These include operations that destroy data and systems. Participating firms must post at least $1 million in escrow, which they forfeit if they break program rules. Every operation needs written approval from Department of Justice and Department of Homeland Security officials.

    A National Coordination Center will run the program. Implementation guidance is due within 60 days. The policy reverses public statements administration officials made in March.

    Executive Summary

    The memorandum authorizes two kinds of activity. “Cyber Surveillance Operations” means getting into foreign systems without permission to gather intelligence while staying undetected. “Cyber Effects Operations” means disrupting or destroying those systems and their data. Both large firms and smaller specialists will be eligible. Participants may also sign commercial deals with other private firms and with state and local agencies to receive threat data and propose operations based on it.

    This matters for two reasons. First, it turns offensive cyber activity, long treated as a government monopoly in the US, into something a licensed private sector can do under supervision. Second, its target definition is narrow. A group qualifies as a target unless “clear intelligence exists” that it is part of a foreign government or wholly run at a government’s direction. So the program is aimed at criminal ransomware crews, not at the state-directed actors behind many attacks on physical infrastructure. For operators of water, power and data center infrastructure, that distinction shapes what the program can and cannot do for them.

    The Target Line Runs Straight Through Critical Infrastructure

    The memo arrived after suspected Iranian cyberattacks on water suppliers in 45 US municipalities. It also followed a CISA alert about Iranian hackers targeting programmable logic controllers (PLCs), the small industrial computers that open valves, run pumps and switch equipment at water and energy companies. Yet state-directed hackers fall outside the program’s own definition of a valid target. Private operators could go after a ransomware crew that encrypts a utility’s billing systems. They could not go after a government-run unit tampering with its treatment-plant controllers.

    The line is drawn deliberately. Ransomware crews that operate with a state’s tolerance but not under its formal control stay within scope, a description that fits much of the Russia-based ransomware ecosystem. Once an actor is clearly an arm of a foreign state, the matter stays with government. For utility and industrial security teams, the practical conclusion is that nothing changes for defending operational technology against nation-state intrusion. Segmentation, monitoring and incident response for control systems remain their responsibility. The new program targets the financially motivated threat, not the geopolitical one.

    There is also an ambiguity problem. Eligibility turns on whether “clear intelligence” ties a group to a government, and in practice attribution is often contested and slow. A group judged criminal today may later prove to have state links. That question will matter a great deal to any firm whose operation lands on the wrong side of the line.

    A $1 Million Bond on a Destructive Capability

    The escrow requirement is the program’s main economic control. A minimum of $1 million, forfeited for rule-breaking, is a meaningful commitment for a small specialist firm. Whether it is proportionate to the harm a destructive operation could cause is a separate question. Wiping servers that turn out to be shared with innocent third parties could produce losses well above that figure. The minimum works as a deterrent and an entry filter. It is not obviously a compensation fund.

    The commercial provisions may matter more for the market’s shape. Participants can contract with other private firms and with state and local agencies, receive threat data and propose operations based on it. That creates a potential new service line for security vendors: selling disruption, not just detection. Two background facts point the same way. Congress earmarked $1 billion for offensive cyber operations in last year’s spending bill. Google said in August last year that it was preparing to take part in disruptive actions against cybercriminals. Large platform companies with deep telemetry into criminal infrastructure look like natural participants. Smaller firms may compete on specialized tasks, as the eligibility rules anticipate.

    Escalation, Legal Exposure and a Classified Annex

    DOJ and DHS officials cannot approve operations likely to cause loss of life or amount to an armed attack under international law. The memo does not prohibit such operations outright, however. Approval authority for them sits in a classified annex. Companies weighing participation, and the public, therefore cannot see the full rulebook for the highest-risk cases.

    Personnel face exposure too. Jake Williams, vice president of research and development at Hunter Strategy, told TechCrunch that Americans involved in these operations “could easily be classified as non-uniformed combatants while traveling overseas.” That is a risk that falls on individual employees, not only on their employers, and firms will need to account for it in staffing, travel policy and insurance.

    The speed of the reversal adds uncertainty. In March, Thomas Lind, then a senior adviser at the Office of the National Cyber Director, said the administration had no plans to authorize private offensive operations: “We’re not interested in fighting pirates with pirates.” National Cyber Director Sean Cairncross said the same week that companies running offensive campaigns were not what the administration meant when it asked industry for help. A policy that changed direction within months could change again, which makes long-term investment in offensive capability a harder bet.

    Collateral Risk Is Written Into the Rules

    The memo requires any company that unintentionally hits a US person or a system on US soil to halt operations and notify the government immediately. That clause concedes that operations aimed abroad can land at home. Criminal groups commonly route activity through rented servers and compromised machines owned by unrelated parties. Hosting providers, cloud platforms and data center operators therefore have a direct interest in how approvals weigh that possibility, and in how quickly they would be told if their infrastructure were affected.

    Defenders more broadly face a quieter risk: retaliation. A criminal group whose systems are destroyed by a US firm may not distinguish between that firm and US targets generally. The memo’s safeguards govern how operations are approved. They do not change who absorbs the response.

    Background

    For decades, US law and policy have treated breaking into someone else’s computer as illegal, even when that someone is an attacker. Offensive cyber operations have been reserved for government agencies such as the military and intelligence community. Proposals to let private victims “hack back,” including the Active Cyber Defense Certainty Act introduced in Congress in 2017, never became law. Critics warned of misattribution, collateral damage and escalation.

    Pressure grew as ransomware crews, many operating from jurisdictions beyond the reach of US law enforcement, hit hospitals, pipelines and municipalities. Some large technology firms have already worked with authorities to take down criminal infrastructure, and Google said last year it was preparing to take part in disruptive actions. The August 12 memorandum is the first formal framework for private firms to go further, under federal supervision.

    Sources

    Source: White House authorizes private companies to launch ‘hack-back’ cyberattacks that destroy data and systems, targeting foreign cybercrime organizations. Tom’s Hardware on the August 12 presidential memorandum creating a supervised private offensive cyber program.

  • One Script, 15+ Government Tenants: Why Shared Hosting Multiplies Espionage Risk

    One Script, 15+ Government Tenants: Why Shared Hosting Multiplies Espionage Risk

    TL;DR · 30-second read

    The Short Version

    Security researchers at Symantec say a small hacking crew based in China spent months reading the email of government officials across Asia and the Middle East. It also ran a cryptocurrency scam business from the same control screen.

    Its most effective move: instead of breaking into government ministries one at a time, it broke into the single company that hosted all of their email. One planted piece of code reached more than 15 government email systems at once.

    The takeaway is simple. When many organizations rely on one shared provider, one break-in can expose all of them.

    Industrial Cyber reported on August 14 that Symantec’s Threat Hunter Team, working with Carbon Black researchers, has exposed Jewelbug, which Symantec describes as a China-based hackers-for-hire group. According to Symantec, the group ran espionage campaigns against governments, militaries and state telecom operators across the Middle East, Southeast Asia and South Asia. It also ran a cryptocurrency fraud operation aimed at Chinese-speaking victims, and both activities were managed from a single browser-based control panel called XG-Web.

    In under three months, Symantec found, the group logged more than one million implant check-ins and stole over 580,000 browser cookies, thousands of credentials and more than 2,300 email bodies. An implant is a piece of malware that reports back to its operators. Its largest operation compromised a shared web-hosting platform run by a Middle Eastern state telecom provider, which placed malicious code in front of more than 15 government webmail tenants at once.

    Executive Summary

    Symantec’s research is unusually detailed. It is built on visibility into the group’s own control panel and victim database, so it describes more than a list of intended targets. It documents actual compromises: harvested mailboxes of senior officials, intercepted internal network traffic, and implants running on Windows machines, Linux servers and network devices. Jewelbug is also tracked under other names, including Earth Alux, REF7707 and CL-STA-0049.

    For infrastructure operators, the most important finding is architectural rather than about any single piece of malware. Jewelbug did not breach each ministry one by one. It gained write access to a centrally hosted webmail system run by a state telecom and planted one script, and that script fired for every ministry employee who logged in. A second finding points in the same direction. The group’s Linux and router implant, ClientKing, is built to reach servers and network equipment, the layer that carries everyone else’s traffic.

    Symantec also reported that some ClientKing builds were configured to route traffic through the internal corporate proxy of a major U.S. aerospace and industrial manufacturer. That extends the group’s footprint beyond Asia and the Middle East into industrial infrastructure.

    One Write Permission, Fifteen Ministries

    The shared-hosting compromise is the clearest demonstration in Symantec’s report of how multi-tenancy changes the economics of an attack. Multi-tenancy means many customers running on one platform. The Middle Eastern government’s ministries did not each run their own mail servers. They were tenants on a web-hosting platform operated by the state telecommunications provider. Once Jewelbug had write access to the central webmail system, it needed only one malicious script. That script ran each time any ministry staff member logged in or checked their email, across more than 15 government tenants.

    The attack chain that followed was conventional, and that is part of the lesson. The injected code captured session cookies and email addresses. Session cookies are the small tokens that keep a user logged in, and stealing one can let an attacker impersonate that user without a password. Windows users were then shown a fake Adobe Flash update prompt, which installed Antino, the group’s backdoor. Antino added a malicious browser extension that gave operators control of the browser and a route into internal government networks. None of these steps required a novel exploit against each ministry, because the shared platform had already delivered every tenant to the attacker.

    Symantec’s own summary makes the point directly: compromising a shared hosting provider and placing a watering hole on every government tenant turned a single intrusion into access across an entire national webmail estate. A watering hole is a trusted site rigged to infect the people who visit it. The implication for telecoms, hosting providers and data center operators that serve government or regulated customers is that the provider’s control plane is part of every tenant’s attack surface. The control plane is the administrative layer with write access to shared code. Tenant isolation at the data layer does little if one set of credentials can change the code all tenants run.

    The Network Layer Is in Scope

    Most espionage reporting centres on endpoints: laptops, email and browsers. Jewelbug’s toolkit goes further. According to Symantec, ClientKing is a Linux and router implant that can reach servers and network devices, and all of the group’s implants feed one victim database. Routers and Linux servers sit below the level where most endpoint detection tools operate. They often run for years between refreshes, and they see traffic from many downstream users at once.

    That matters for the telecom operators Symantec says were targeted. A foothold on network infrastructure can outlast a cleaned-up laptop and can observe traffic rather than just one user’s files. Symantec noted that the group intercepted internal network traffic, and that the common thread across its espionage targets was government communications and the providers that host them. The researchers said that combination would give an intelligence customer broad, durable access to official correspondence.

    The detail about the U.S. aerospace and industrial manufacturer needs careful reading. Symantec said a couple of ClientKing builds were configured to beacon through that company’s internal corporate proxy. That shows the group’s tooling was set up to operate inside or through that environment. The report as described does not establish the extent of any compromise there.

    A Hack-for-Hire Business Model, Visible From the Inside

    Symantec’s central attribution finding is that foreign-government espionage and commodity cryptocurrency fraud ran on the same infrastructure, by the same small team, from one control panel. The researchers call that pairing the signature of a hack-for-hire entity running for-profit crime on the side. They tied the fraud and search-engine-optimisation arm, with high confidence, to the sole legal representative of an SEO company in Changsha, Hunan Province. That attribution rests on identity documents, a business license and a stamped authorization letter. Symantec was explicit that the link between this individual and the espionage operators is not fully established. Its assessment is that the SEO business most likely supplied access, infrastructure and delivery.

    The operational discipline is notable for defenders. The group has built five generations of command-and-control code, the servers and software attackers use to direct their malware. A scheduled job checked its own domains against VirusTotal every 12 hours so operators could rotate away from anything flagged. VirusTotal is a widely used public malware-scanning service. Antino disguised its communications as ordinary Microsoft cloud traffic. Taken together, these choices suggest that blocklists and domain reputation alone will lag this kind of operator. Behavioural monitoring and integrity checks on shared code are better placed to catch it.

    For hosting and telecom operators, the practical questions are about blast radius. Who holds write access to code that every tenant executes? Is that code monitored for unauthorized changes? Are session tokens bound tightly enough that a stolen cookie is of limited use? None of these are exotic controls. The Jewelbug case shows what happens when one of them fails on a platform many customers share.

    Background

    Symantec is one of the longest-established names in enterprise security. Its enterprise business has been part of Broadcom since 2019, and its Threat Hunter Team publishes research on state-linked and criminal hacking groups alongside Carbon Black, another Broadcom-owned security brand. Different vendors often track the same group under different names, which is why Jewelbug also appears in industry reporting as Earth Alux, REF7707 and CL-STA-0049.

    Telecom operators and hosting providers have long been attractive espionage targets because they sit upstream of many customers at once. Governments in many regions consolidate ministry email and web services onto shared platforms, often run by a state telecom, to cut cost and complexity. The Jewelbug case shows the other side of that trade-off: consolidation also concentrates risk. Earlier in 2026, Symantec reported on Seedworm, an Iran-linked group whose campaigns spanned government, airport, manufacturing and financial targets.

    Sources

    Source: Symantec reveals Jewelbug espionage campaign targeting Asian governments, telecoms, critical infrastructure (Industrial Cyber): coverage of Symantec’s findings on a China-based hack-for-hire group running espionage and cryptocurrency fraud from one control panel.

  • Namecheap’s 15-Hour Outage Shows Why Denser AI Halls Can’t Rely on One Chiller Plant

    Namecheap’s 15-Hour Outage Shows Why Denser AI Halls Can’t Rely on One Chiller Plant

    TL;DR · 30-second read

    The Short Version

    A storm damaged the cooling system at a data center in Phoenix. A data center is a warehouse-sized building full of computers that must be kept cold to keep working. Without enough cooling, the web company Namecheap went dark for about 15 hours.

    Websites, email and even Namecheap’s own customer help line went offline. Fixing the cooling took hours. Getting everything else running again took much longer.

    This matters more as computers built for artificial intelligence, which run far hotter, fill new buildings. Backup cooling inside one building may not be enough.

    Namecheap, a leading domain registrar and web host, suffered an outage of roughly 15 hours on August 14 after what it called “a failure of cooling systems” at its data center in Phoenix. TechRadar reported that the company’s hosting plans, its EasyWP managed WordPress service, DNS (the internet’s address book, which points domain names at servers) and email were all offline, along with customer accounts and Namecheap’s own support desk. The facility’s operator, PhoenixNAP, said heavy overnight storms had caused damage that led to “elevated white space temperatures” in the data halls.

    Two of the data center’s four chillers were back online around 12pm EDT. Core databases returned at 5:30pm EDT, Namecheap.com and live chat about 40 minutes later, and full service at 3:50am EDT. Chief executive Hillan Klein apologised, saying “we failed you today,” and committed to publishing a post-mortem.

    Executive Summary

    A storm damaged the cooling plant of a Phoenix data center, and one of the best-known names in domain registration and small-business hosting went down for most of a day. Customers lost websites, email and name resolution. They also lost the usual way to get help, because Namecheap’s help desk and live chat went down with everything else.

    The incident matters beyond Namecheap’s customers because it isolates a failure mode the industry usually discusses in terms of power: the thermal plant. The building had four chillers, the industrial refrigeration units that produce the cold water used to pull heat out of server rooms. That redundancy did not prevent an outage when a single external event, a storm, affected the cooling system. Even after partial cooling returned, restoring the service stack took many more hours.

    That lesson gets sharper as operators pack far hotter AI hardware into new halls. When heat density rises, the time between losing cooling and losing servers shrinks. Resilience that stops at the building’s own chiller plant leaves less margin than it once did.

    Four Chillers, One Storm

    Data center cooling is normally designed with spare capacity. The common “N+1” approach means one more chiller than the load requires, so a single unit can fail or be serviced without consequence. Namecheap’s Phoenix facility had four chillers. Bringing two of them back online was the milestone that began recovery, which indicates the plant had been running well short of what the halls needed. Neither company had said, as of August 15, exactly how many units failed or which part of the cooling system the storm damaged.

    This is the core distinction between component redundancy and site resilience. Spare chillers protect against one machine breaking. They protect far less well against a common-mode event, meaning a single cause that hits several pieces of equipment at once. Storm damage to shared infrastructure is exactly that kind of event. PhoenixNAP’s own explanation traced the problem to weather rather than to an isolated equipment fault, so the redundancy on paper did not match the resilience in practice.

    Cooling Came Back Hours Before the Service Did

    The published timeline is the most instructive part of this incident. Partial chiller capacity returned around noon. Core databases did not come back until 5:30pm, and the public website and live chat returned about 40 minutes after that. Full service was not restored until 3:50am. Most of the outage therefore happened after the thermal problem had begun to ease.

    That gap reflects how hosting platforms behave after a thermal event. When a room overheats, servers throttle, shut themselves down, or are powered off by operators to protect the hardware. Bringing them back is not a single switch. Storage has to be verified, databases checked for consistency, and dependent services restarted in the right order. For anyone writing recovery objectives, the lesson is that restoring the cooling plant starts the recovery clock. It does not stop it.

    Why AI Halls Can’t Rely on One Chiller Plant

    Namecheap’s outage involved conventional web hosting, not AI hardware. The mechanism still applies directly to AI buildout. A room’s temperature rises after cooling fails at a rate set by how much heat the equipment keeps producing. Conventional hosting racks give operators some buffer. AI training and inference racks, built around dense GPU servers, concentrate far more heat in the same floor space, so the window for an orderly shutdown or failover gets shorter.

    Liquid cooling, now standard in many AI designs, moves heat off chips more efficiently than air. It still hands that heat to the facility’s central plant: chillers, dry coolers or cooling towers. A storm that damages that plant affects a liquid-cooled AI hall as surely as it affected Namecheap’s hosting floor, and the higher density leaves less time to react. The practical conclusion for operators and tenants is that resilience needs to extend beyond one plant. That can mean heat-rejection equipment that is physically separated and weather-hardened, or the ability to move workloads to another site. Buyers of AI capacity have reason to ask colocation providers how their cooling redundancy performs against shared causes, not only against a single failed unit.

    When the Help Desk Shares Fate With the Outage

    The outage also took down Namecheap’s support channels, leaving staff unable to answer live chat or email. The company fell back on its status page and on Klein’s posts on X. Namecheap did keep customers informed with detailed status updates throughout. The episode still shows why tools for talking to customers during a crisis are best hosted somewhere other than the infrastructure they report on.

    Klein’s commitment to examine “where our existing safeguards and protocols failed” is the right question. The test will be whether the post-mortem addresses architecture as well as the storm, specifically whether services like DNS and email had anywhere else to run when one building lost its cooling.

    Background

    Namecheap is one of the most widely used domain registrars, the companies that sell and manage website addresses. It also sells web hosting, managed WordPress through its EasyWP service, email and DNS, mostly to individuals and small businesses. Because many customers buy their domain, website and email from Namecheap together, one infrastructure failure can take down a business’s whole online presence at once.

    Data centers spend a large share of their engineering effort on cooling, because nearly all the electricity servers use becomes heat. Chilled-water plants are the workhorse of larger facilities and are normally built with spare units. Heat density per rack is rising sharply as AI hardware spreads, so the reliability of this thermal plant is becoming as central to uptime as power supply and backup generators.

    Sources

    Source: TechRadar: ‘We failed you today’: Namecheap down for several hours after a data center cooling failure, leaving customers furious, covering the roughly 15-hour Namecheap outage caused by a storm-related cooling failure at its Phoenix data center.

  • Upper Merion’s 4.6M sq ft Denial Shows Why Self-Powered AI Campuses Still Need Town Hall

    Upper Merion’s 4.6M sq ft Denial Shows Why Self-Powered AI Campuses Still Need Town Hall

    TL;DR · 30-second read

    The Short Version

    A suburb outside Philadelphia voted down five giant buildings meant to hold the computers that run artificial intelligence. Together they would have covered roughly 80 football fields of floor space.

    The builder promised $55.4 million a year in local taxes, mostly for schools. Neighbors worried about noise, dirty air and home values, and more than 17,000 people signed a petition against it.

    The twist: to avoid raising local electric bills, the site would have made its own power with gas-burning generators. That fix became part of the fight. The builder says an appeal is coming.

    WHYY reported that Upper Merion Township’s Board of Supervisors on Thursday, August 13, 2026, rejected all five applications from developer Brian O’Neill for a hyperscale data center campus totaling more than 4.6 million square feet near King of Prussia, Pennsylvania. The campus would have been built at Discovery Labs, a health care and life sciences office property run by O’Neill’s company, and was designed to generate its own power using natural gas turbines with diesel backup generators.

    The vote followed a July planning commission recommendation against three of the five projects, a developer lawsuit seeking more review time that a Montgomery County judge cleared on the afternoon of the vote, and a resident petition with more than 17,000 signatures. O’Neill told 6ABC an appeal is planned.

    Executive Summary

    A suburban Philadelphia township has turned down one of the largest data center proposals in the region, despite the developer’s estimate of $55.4 million a year in local tax revenue, nearly $38 million of it for the local school district. The board’s stated grounds were procedural and specific: Chair Tina Garzillo said the township never received landscape, traffic impact, emergency management or sanitary sewer plans, and called the projects “at best speculative and conceptual in nature.”

    The case matters beyond Upper Merion because of how the campus was to be powered. Generating electricity on site is increasingly pitched as the responsible way to build AI-scale facilities without burdening the grid or ratepayers. Here, that design put gas turbines and diesel generators in front of a township board and its residents, and air pollution became a central objection. Self-generation did not remove the local approval gate; it added a power plant to what that gate had to judge.

    For developers, investors and utilities trying to forecast where AI capacity will actually get built, the vote is a reminder that a proposed campus is not capacity until it clears local land-use approval, and that the power strategy is now part of that approval.

    Self-Generation Swapped a Grid Question for a Local One

    The campus was pitched as a good grid neighbor: it would generate its own electricity so it would not strain the regional power system or push up household bills. That is an increasingly common design for AI-scale facilities, often called on-site or behind-the-meter power because the electricity is produced on the property rather than drawn from the utility. But in Upper Merion, the power plant came with the building. The proposal included natural gas turbines and diesel backup generators, and air pollution became one of residents’ leading objections, alongside noise and water use.

    That is the mechanism this vote exposes. A data center that buys power from the grid leaves most generation questions to utilities and grid planners. A data center that makes its own power brings combustion equipment, fuel supply and emissions controls onto a site that a township board must approve, in front of neighbors who will live beside it. O’Neill said scrubbers would control emissions and that discussions had been held with Bloom Energy about using fuel cells on site, but no deal had been finalized, which left the board and residents to evaluate the turbine-and-diesel version of the plan.

    Upper Merion is not the only place where on-site power has drawn local enforcement. In Vineland, New Jersey, where a data center under construction plans to use Bloom fuel cells, the city issued a stop-work order on parts of the 300-megawatt facility (a megawatt is a measure of electrical capacity), alleging that a liquefied natural gas tank and fuel cell units were being installed before city approval. Generating power on site may answer concerns about grid strain and electricity bills, but it adds a power plant to the list of things local officials get to review.

    The Board Ruled on Missing Plans, Not Just Public Anger

    The emotional scenes at the meeting can obscure what the board actually said. Garzillo cited the absence of landscape, traffic impact, emergency management and sanitary sewer plans. Those are ordinary engineering submissions for any large industrial project, and their absence gave the board a procedural basis for denial that stands apart from residents’ objections.

    The process had been heading this way for weeks. In July, the township planning commission voted not to recommend three of the five projects. The developer then sued, seeking an extension and arguing there had not been adequate time to address township concerns; a Montgomery County judge vacated the injunction on the afternoon of the vote, clearing the board to reject the extension. Vice chair William Jenaway said the developer had earlier opportunities to respond to review comments and did not do so in a timely fashion. The two sides’ accounts of the timeline differ, and an appeal would likely test them.

    For the industry, the lesson is about entitlement risk, meaning the risk that a site never receives the land-use approvals it needs. As AI demand pushes developers to repurpose existing properties, here a life sciences office campus, into data center sites, applications that arrive without complete plans risk being denied on the paperwork alone, whatever their scale or revenue promise.

    A $55.4 Million Tax Pitch Met a 17,000-Signature Petition

    The developer’s economic case was large on its face: an estimated $55.4 million a year in local tax revenue, with nearly $38 million going to the Upper Merion Area School District. For a suburban township, that is the kind of figure that usually carries weight. It did not carry the vote. Residents weighed it against quality-of-life costs that do not appear in a revenue projection, and one resident nearing retirement said a loss of home equity “would destroy our savings.”

    Both sides’ claims deserve the same scrutiny. The tax figure is the developer’s own estimate, and its value depends on assumptions about how much gets built and when. Residents’ fears about property values and health are sincere, but their magnitude depends on specifics that were never settled, such as how many turbines would run, how often, and with what controls. The water debate is similar: O’Neill said a closed-loop system would reuse cooling water, while hyperscale facilities in general can consume up to 5 million gallons of drinking water a day. Closed-loop designs reduce consumption substantially, but how much make-up water this campus would need was not specified.

    The backdrop is shifting against developers who lead with revenue. A University of Pennsylvania Annenberg Public Policy Center survey found a 12-point jump in widespread resistance to new data center construction over four months. In that climate, detailed answers on emissions, water and noise are likely to matter more to approval than headline tax estimates.

    What Comes Next for AI Capacity Around Philadelphia

    The denial is not necessarily final. The developer can appeal or submit new or revised plans, and has said an appeal is planned. Residents have said they will oppose any appeal or new application. O’Neill is also seeking approval for a data center in nearby Plymouth Township, which becomes the next test of whether a more complete application, and possibly a different power plan such as the fuel cells under discussion, changes the outcome.

    For operators, investors and utilities counting future AI capacity in the region, the practical point is that local land-use approval is a gate that neither scale, tax revenue nor a plan to stay off the grid can bypass. Capacity forecasts that treat proposed campuses as inevitable will miss decisions like this one, which are made parcel by parcel at township meetings.

    Background

    Upper Merion Township sits in Montgomery County, Pennsylvania, and includes much of the King of Prussia area, a major suburban commercial and office hub northwest of Philadelphia. Data centers, the buildings that house the servers behind internet services, have existed since the mid-20th century, but the rise of power-hungry artificial intelligence has pushed demand toward much larger hyperscale campuses that need substantial electricity and cooling.

    That demand has driven developers to look at existing commercial properties, such as office and life sciences campuses, as potential data center sites, and to consider generating their own power rather than waiting on the grid. It has also produced growing local resistance in the Philadelphia region and beyond, where township boards, not state or federal agencies, hold the zoning and land-use approvals that decide whether a campus can be built.

    Sources

    Source: Upper Merion board denies hyperscale data center campus near King of Prussia, WHYY’s report on the township supervisors’ vote rejecting a five-building data center campus.