FILE PHOTO: Representations of cryptocurrencies are seen in front of displayed Binance logo in this illustration taken November 10, 2022. (credit: REUTERS/DADO RUVIC/ILLUSTRATION/FILE PHOTO)

REUTERS/DADO RUVIC/ILLUSTRATION/FILE PHOTO

North Korea rewards its top hackers, tasked with heists like the Binance crypto hack that netted a cool $6.3 billion, with perks such as a spacious apartment in the capital of Pyongyang and huge hard currency awards, despite grueling hours of work.

Leader Kim Jong Un has assembled an army of world-class hackers using such incentives in a program more than a decade old that picks 200 of the smartest middle school students each year to study computer engineering, defectors and analysts say.

A fraction eventually become hackers in an investment that has paid off dramatically, as the group has evolved from the early days when it ran illegal gambling websites to pulling off the world’s biggest cyber heist last month.

“They are actually one of Kim Jong Un’s secret weapons,” said Kim Heung-kwang, a North Korean defector who trained hackers on behalf of the state before fleeing to South Korea in 2004.

“There is no other country grooming hackers like this, systematically searching for the smartest children, solely to turn them into operatives,” he added. “Whether they want to be a hacker or not does not matter.”

FILE PHOTO: Representations of cryptocurrencies are seen in front of displayed Binance logo in this illustration taken November 10, 2022. (credit: REUTERS/DADO RUVIC/ILLUSTRATION/FILE PHOTO)

What happened in the Binance hack?

The FBI and Binance have said North Korea was responsible for the Binance hack, but Pyongyang has denied that. Blockchain investigation firms have attributed it to a North Korean group known as “TraderTraitor.”

The denial is part of a pattern in which the North dismisses such blame over the years as fabricated political provocations.

North Korea set up a cyber unit in 1998 but elevated its status after Kim came to power and declared in 2013 that cyber warfare was, along with nuclear weapons and missiles, an “all-purpose sword” crucial to the country’s strike capability.

Candidate selection begins in middle school as early as 11 or 12 years of age, with authorities picking about 150 to 200 children a year to study computer science, said Kim Heung-kwang, the former trainer, and two defectors who worked in IT.

At 17, the top performers are sent to North Korea’s best universities and undergo further screening before a chosen few finally enter the military’s cyber unit as hackers, they added.

They are trained in areas such as cryptocurrency protocols and reverse-engineering foreign hardware devices, in addition to more conventional topics such as advanced mathematics and foreign languages, one of them said.

Those who do not become hackers are trained to develop software or apps for Western firms via remote jobs.

“While South Korea’s best students may pursue medicine or other prestigious careers, North Korea channels top talent into nuclear and cyber programs and essentially guarantees their future,” said Lim Jongin, professor emeritus at the school of cybersecurity at Korea University.

Other privileges on tap for the hackers are travel abroad to allied countries, such as China, where they can more freely access the internet.

Successful exploits can earn as much as 30% of the heist’s profits in US dollars for the hackers’ military unit, and they can even receive a flat in the capital where their parents can also stay, Kim said.

“There are accounts that successful cyber personnel are given apartments in Pyongyang,” Lim said, adding that the flats could range to as much as 132 sq m (1,420 sq ft).

Such performers are also eligible for membership of the ruling party and high-status marriages into elite families, said one of the defectors.

But it is a small group. The North’s cyberwarfare operations employ about 8,400 people, South Korea’s National Intelligence Service estimated in 2024.

North Korean state media do not publicly celebrate such hackers’ achievements, but they regularly highlight students who show talent for programming.

In 2015, North Korea opened the Mirae Scientists Street in the capital, in the construction of which Kim Jong Un had been closely involved, state media said.

Hackers’ formal designation in the North was “scientist” or “researcher,” said one of the defectors.

Other hacking groups associated with North Korea include the Lazarus Group, also known as APT38, which the US accused in 2022 of having stolen hundreds of millions of dollars worth of cryptocurrency from the Ronin blockchain network. The FBI has said TraderTraitor is affiliated with Lazarus.

State backing makes the North’s hackers more dangerous and harder to repel, as they face no consequences for hack after hack, enabling them to perfect their strategies, analysts said.

“In the world of cybercrime, often, if you make a mistake … that is all law enforcement needs to get you,” said Alec Zebrick, senior manager for Asia-Pacific global services at Chainalysis.

But North Korea’s hackers “are doing this for the regime, so they’re going to continue, even after successes,” he added.

“Even after failures, they’ll continue, because it’s very hard for the international community to take action against them.”

North Korean hackers use AI to fake their identities

Other groups that have engaged with North Korean IT workers who earn funds for the Kim regime by landing remote technology jobs at Western companies describe them as highly organized, polite in their interactions, and avid users of AI.

“There clearly was a structure you would see in military. Strict times, cleaning boots, very orderly,” said YouTuber Deeveeaar, who has claimed to have hacked into cameras watching North Korean IT workers at their jobs.

“To say they ‘use a lot of AI’ is an understatement. AI is the very backbone of their operation,” he said, citing the use of paid ChatGPT accounts, Claude Code and Cursor, among many AI tools.

“They write code with it, alter their resumes and help make new ones. And even forge documents and identities.”

Binance said the hackers used a bogus job interview and software hidden in an advertisement to compromise staffers’ computers before using an internal AI tool to seize control of systems.

In 2025, a sanctions monitoring team drawn from 11 United Nations members said TraderTraitor had obtained both ChatGPT and DeepSeek accounts.

Yet the lifestyle of a hacker or elite technology worker for North Korea is grueling, particularly when given a mission or sent abroad, say former IT workers and human rights groups.

Living in cramped spaces, often working stretches longer than 10 hours a day, constant surveillance and strict targets are among the downsides, said former IT workers interviewed for a report last year by advocacy group PSCORE.

“There is always pressure to keep earning money,” said the group’s executive director, Nam Bada. “The stress is extreme, and because of the psychological strain people develop health problems they didn’t have before.”

IT workers must meet a monthly target in dollars, or face humiliation and other punishments, he said.

Still, the hackers’ success is likely to embolden the North Korean leader as they get better and grow more ambitious, analysts say.

“We see more and more them leaning towards what we call ‘big game hunting’,” said Zebrick.

“So they’re targeting some of these larger organizations, potentially have larger vaults of funds, that they could potentially get. These are harder targets,” he added.

“However, as we can see with the Binance case, the payout certainly can make it worth their time.”

This post was originally published on here. 

A smartphone beside a laptop on a desk.

Photo: alleksana / Pexels

Three men have been charged in an alleged scheme that used artificial intelligence to impersonate corporate executives and lawyers, tricking companies into sending at least approximately $129 million to accounts controlled by the group, according to an indictment unsealed Thursday in federal court in Manhattan.

Prosecutors accused Jacob Soall, Sharon Hazan and Guy Sheetrit of participating in a scheme that ran from at least July 2024 through February 2026. The defendants and their associates allegedly targeted thousands of companies worldwide, including employees authorized to approve large wire transfers.

A familiar voice and a confidential deal

The indictment describes messages that appeared to come from a company’s chief executive, sometimes delivered through an electronic-signature service. Employees were told they were helping with a confidential transaction and were asked to sign a nondisclosure agreement.

The group allegedly used AI to mimic executives’ voices and, at times, their images during calls. Employees were then directed to people posing as lawyers at major law firms, who requested transfers for transactions that prosecutors say did not exist.

The alleged losses included about $11.6 million at a British financial-technology company, $4.8 million at a Missouri hospital system and $2.7 million at a Rhode Island bank. The indictment does not identify those companies by name. Prosecutors also describe an attempted $2.85 million transfer at a New York investment bank that its back office stopped.

Payment controls face an AI test

For companies, the alleged scheme shows why a familiar face or voice is not enough to authenticate a payment request. The FBI advises companies to confirm transfer requests, verify changes in payment destinations and consider two-step verification for wire payments. It also warns about requests for secrecy or pressure to act quickly.

The indictment alleges that proceeds moved through bank accounts in several countries and were converted into cryptocurrency to conceal their source and ownership.

Soall was arrested in Spain in June, Hazan in Spain in August and Sheetrit in Britain in September, according to ABC News’s report on the prosecution. Extradition requests are pending. The charges include wire-fraud conspiracy, wire fraud, aggravated identity theft and money laundering. The defendants are presumed innocent unless proven guilty.

JBizNews Desk | Business & Technology

© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.

Photo: NASA/Kim Shiflett

The Federal Aviation Administration proposed changes Thursday to commercial space-launch and reentry safety rules and outlined a broader licensing overhaul for 2027, a package that could affect how operators such as Elon Musk’s SpaceX and Jeff Bezos’ Blue Origin obtain launch approvals.

The FAA’s announcement separates three substantive proposed rules from one final rule making minor technical corrections.

Abort thresholds and reentry debris calculations

One proposal would define when a flight abort is not mandatory and allow operators to toggle abort capability during flight. Another would clarify the requirements and thresholds for when an abort must occur. An abort ends a flight or mission early in response to a problem.

The third proposal would let operators account for a vehicle breaking apart when evaluating whether random reentry could produce a high-consequence event with multiple casualties.

A Reuters report carried by StreetInsider described the changes as a potential boost for SpaceX and Blue Origin.

A broader licensing overhaul is planned for 2027

The FAA said it would begin a broader overhaul of its licensing framework in 2027. It said the work would support an anticipated pace of 1,000 launches a year by 2030, and separately said it expects 10,000 FAA-licensed launches and reentries annually by 2035.

The agency also announced an expansion of its Office of Commercial Space Transportation, including an associate administrator and a team of aerospace engineers to conduct safety analysis, evaluations and mission authorizations. It said applications for the jobs must be received by December 4.

The FAA says the package will reduce regulatory burdens without compromising its safety standards.

What it means for operators and people near launch sites

For launch operators, the practical issue is whether the proposed standards eventually change application requirements and approval times. For people near launch and reentry paths, the relevant issue is how mandatory-abort rules and debris-risk analysis protect the public.

JBizNews Desk | Washington

© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.

New York Stock Exchange facade

Archive: Ryan Lawler/Wikimedia Commons

Investors considering Anthropic’s planned stock-market debut are pushing for a lower valuation after the technology-stock selloff. The discussions could affect the price and timing of the Claude maker’s offering, but the company has not announced a final IPO price.

A report published Thursday, citing four people familiar with investor discussions, says potential investors doubt Anthropic can start its planned roadshow next week and list in November. Some see the recently discussed $1.5 trillion valuation as the high end of expectations and want a more attractive price before committing. Anthropic declined to comment.

Company filing leaves price and share count open

Anthropic announced its public registration filing on October 26, saying it intended to list Class A common stock on the Nasdaq Global Select Market under the symbol ANT. The company said the number of shares and the price range had not been determined.

Its November 2 amended registration statement still leaves the offering price range and share count blank.

Why investors are cautious

The roadshow was initially expected the week of November 9, with pricing the following week. The people familiar with the discussions said the company may need to offer a bigger discount because of its capital needs, quarterly losses and large commitments to partners building AI infrastructure.

Two of the people said investors are waiting for Anthropic’s October revenue figure, left blank in the updated filing, to judge whether growth has picked up after a slowdown. The report also says continued weakness in AI-related stocks could lead the company to wait for steadier markets, even if it and investors agree on a price.

JBizNews Desk | New York

© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.

Photo: Joyce Boghosian / White House (file photo)

Three Republican House committee chairs have asked Nvidia Chief Executive Jensen Huang to give a joint, closed-door briefing on the company’s safeguards against foreign espionage and his interactions with Megaspeed executive Alice Huang, whom the lawmakers describe as a publicly reported Chinese intelligence asset.

The November 4 letter, released by the House Foreign Affairs Committee on November 5, requests an appearance on November 18. It asks Huang to accept that date or propose another congressional session day by November 11.

A request for a briefing, not a subpoena

The signatories are Foreign Affairs Committee Chairman Brian Mast, Select Committee on China Chairman John Moolenaar and House Intelligence Committee Chairman Rick Crawford. They wrote that reports about access to sensitive U.S. technology raised national-security concerns and that their committees wanted to determine the extent of any risk.

Nvidia says customer meetings do not share sensitive information

Nvidia did not confirm whether Huang would attend when asked by CNBC. Instead, a spokesperson said regional cloud customers receive ordinary business information, “not sensitive NVIDIA trade secrets” and “zero information relevant to national security.”

“Meetings with customers are part of a CEO’s job; Jensen talks with thousands of customer executives every year,” the spokesperson told CNBC. The company said it was confident the U.S. government would notify it if officials believed someone posed a threat to American industry.

Reported intelligence ties remain an attributed allegation

The letter describes Alice Huang as a Megaspeed executive and a publicly reported intelligence asset. CNBC said the request followed New York Times reporting that U.S. intelligence had identified her as passing information to Beijing and that she had developed a close relationship with Jensen Huang.

Nvidia’s account is that its chief executive’s interactions were ordinary customer meetings. CNBC reported that Megaspeed did not immediately respond to its request for comment.

The immediate development is congressional scrutiny and a requested briefing.

JBizNews Desk | Technology & Global

© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.

OpenAI has debuted a new AI model geared towards helping cybersecurity professionals.

The company says that the new model, called GPT-6 Cyber, outperforms its general-purpose GPT-6 Astra model on most cyber tasks. Because of the risk that the model could be used to conduct cyberattacks if it fell into the wrong hands, OpenAI is current making the model available only to a small handful approved security customers. 

Fortune first reported that OpenAI was planning to release a specialized cyber model based on GPT-6 in September.

As AI models have become better at finding software flaws and turning them into working attacks, businesses have come under increasing pressure to strengthen their cyber defenses. Gartner forecast in September that worldwide AI cybersecurity spending would reach $51.3 billion this year, nearly double last year’s $25.9 billion.

OpenAI is increasingly leaning into this market, seeing it as an important path to growing its enterprise sales. This is OpenAI’s fourth cybersecurity-focused model this year, following versions released in April, June and August. The company also hired a chief revenue officer, Dali Rajic, who comes from a cybersecurity software background.

Google and Anthropic are also competing for security customers. Anthropic’s Glasswing program gives vetted users access to capabilities that its public models might block. Google, which previously emphasized a lightweight, low-cost model for repeatedly scanning large codebases, has also begun offering vetted defenders access to its frontier Gemini 4 Argon model without cyber guardrails through its Fairwind Program.

OpenAI’s GPT-6 Cyber focuses on advanced vulnerability research and exploit development. To test a server’s defenses, researchers may try breaking into it to see what an attacker could do with a flaw. The same methods can be used in malicious attacks, so general-purpose AI models impose restrictions that can also block legitimate security work. OpenAI has said it trains its Cyber models both to improve specialized cybersecurity skills and reduce refusals for certain tasks that can serve attackers and defenders.

OpenAI tested GPT-6 Cyber on Firefox, Mozilla’s web browser, and the Linux kernel, the core software underlying many servers. Firefox must handle potentially malicious webpages without exposing users’ data, while the Linux kernel controls access to a machine’s resources. Both are open-source projects whose code has been inspected, tested, and hardened by developers and security researchers for years. And yet AI models are still able to find novel vulnerabilities.

OpenAI says researchers used GPT-6 Cyber to find 81 previously unknown, high-severity Firefox vulnerabilities and develop working exploits for 948 Linux kernel vulnerabilities, including 588 previously unknown flaws. The Linux exploits could let an attacker with ordinary local access gain root privileges, the highest level of control over the system. Researchers developed working exploits for more than five times as many Linux vulnerabilities as in a July study using GPT-5.6 Sol, which was considered a state-of-the-art model for cyber tasks when it was released that month.

OpenAI currently offers two levels of access to trusted cybersecurity professionals through its Daybreak program: Daybreak Red supports advanced work such as simulating attacks and building tools that allow for red-teaming, while Daybreak Blue focuses on defensive tasks such as reviewing code, analyzing malware and investigating incidents.

GPT-6 Cyber will initially be available to a select group of approved Daybreak Red customers only, with access expanding to all Red customers over the coming weeks, the company said. Separately, OpenAI said it is expanding early testing of Astra for defensive security work to more organizations and individual security researchers. through Daybreak Blue. 

This story was originally featured on Fortune.com

This post was originally published here. 

Archive Starlink terminal in Ukraine, November 24, 2022

A Starlink terminal in Ukraine, November 24, 2022. Photo: Odapresste/Wikimedia Commons, CC BY-SA 4.0, cropped. Archive Starlink equipment, not Starshield hardware or a current outage. The cropped image remains licensed CC BY-SA 4.0.

Ukraine’s contract with the U.S. government for access to the Starshield satellite network expires around the middle of 2027, and Kyiv wants talks on a new agreement to begin now, a senior Ukrainian diplomat said Thursday, according to Reuters.

Denys Sienik, the charge d’affaires at Ukraine’s embassy in Washington, told the Defense Writers Group that negotiating a new contract takes time. He did not provide an exact expiration month.

“We want to have a clarity, basically, if we can proceed and what the terms of reference will be,” Sienik said, according to Reuters.

The contract he described is Ukraine’s arrangement with the U.S. government for access to Starshield, not an announced direct contract between Kyiv and Elon Musk’s SpaceX. The reports do not establish a signed renewal, a decision to cancel the service or an imminent loss of access.

Starshield is a government service, not commercial Starlink

SpaceX provides Starshield to the U.S. government. The company’s own description distinguishes it from Starlink: Starlink is designed for consumer and commercial use, while Starshield uses related technology for government and national-security customers.

That distinction matters to who negotiates Ukraine’s access and sets its terms. Sienik is asking for clarity from Washington on the future arrangement, not announcing a new consumer Starlink subscription or a replacement provider.

Kyiv says October’s restrictions still lack an explanation

Sienik said Ukraine had not received an explanation from Washington or SpaceX for speed restrictions imposed last month. Reuters reported that the limitations hampered Ukrainian forces’ ability to intercept Russian attack drones and forced the cancellation of some operations.

“Nobody provided answers,” Sienik said of Kyiv’s requests, which he said included approaches to the Pentagon, White House and SpaceX. That is his account of the responses Ukraine received, not an independent finding about who ordered the restrictions.

In a separate Breaking Defense report from Thursday’s event, Sienik said the service was “working now” and the earlier problem had been resolved. He said Ukraine wanted to avoid a repeat and was also exploring backup capabilities, including domestic and European efforts.

The new disclosure concerns the contract’s future, rather than a new outage. It follows earlier comments by President Volodymyr Zelensky about responsibility for the October slowdown, which did not establish who made the decision.

For Ukraine, the unresolved issue is how to secure reliable wartime connectivity beyond the existing contract’s expiry. Thursday’s reports supply an approximate deadline and Kyiv’s request for talks, but no agreed funding, final terms or replacement agreement.

JBizNews Desk | Global & Technology

© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.

Empty office workspace

Salon is ceasing publication after 31 years, the digital news and culture outlet announced Thursday, saying its publishing model was no longer sustainable.

“Today, after 31 years, we are saddened to announce that Salon is ceasing publication,” the company said in a farewell statement published November 5.

The announcement ends the publishing run of an outlet that launched in November 1995, during the early development of web-based journalism. Salon’s coverage spanned politics, culture, criticism and essays, helping establish a place for magazine-style reporting and commentary online.

Company says the model is no longer sustainable

“The business realities of independent digital journalism have changed profoundly over the years,” Salon said. “After careful consideration, we have reached the difficult decision that our publishing model is no longer sustainable.”

The statement did not provide revenue, debt or cash figures, identify a buyer, or announce a bankruptcy filing. It also did not specify how many employees would be affected or set out severance arrangements.

Salon did not attribute its decision to a single technology company, a search-engine change or artificial intelligence. Its announcement described a broad change in the economics of independent digital journalism, without providing a detailed financial explanation.

Editorial archive expected to remain online for now

The Verge reported Thursday that Salon chief content officer Erin Keane said the editorial archive would “stay live for now.” That distinguishes the end of new publication from an immediate deletion of the site’s existing articles, though it is not a promise of permanent access.

The Verge also reported that Salon general manager Amanda Wolfe referred its request for comment to the company’s farewell announcement.

Salon thanked its journalists, editors, photographers, producers, designers, business staff, contributors and readers. It described its history as a collective effort across several chapters of the company.

The closure is a change in the digital-media business, not an announced sale or restructuring plan. The company has confirmed that it is ceasing publication, while details about its staff and the longer-term future of its archive remain limited.

JBizNews Desk | Business & Technology

© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.

Server equipment in a data center

Server equipment in a data center. Photo: Panumas Nikhomkhai/Pexels. Generic image, not Google or Wyoming.

Google announced a long-term $17 billion investment Thursday to build its first data center in Wyoming, a project in Laramie County that pairs a major computing campus with natural-gas power and planned carbon capture.

The company outlined the investment in a Nov. 5 announcement by Steve Walter, its vice president of global infrastructure and energy. Reuters also reported the plan, placing it against growing political resistance to data-center expansion after the U.S. midterm elections.

Google named Tallgrass Energy, Black Hills Energy, American Electric Power and Related Digital as partners. The $17 billion is a long-term investment commitment, not a statement that the full amount has already been spent or that the campus is operating.

How the power plan would work

The project is contracted with AEP to feature a planned 702-megawatt fuel-cell development using American-made Bloom Energy equipment, Google said. Fuel cells turn fuel into electricity through an electrochemical reaction. The fuel in this project is natural gas.

Google says carbon capture and storage would collect a large majority of the fuel-cell development’s carbon emissions and store them underground at Tallgrass’s nearby Eastern Wyoming Sequestration Hub. That is a company claim about a planned system, not a claim that the project produces no emissions or runs entirely on renewable energy.

Google also plans to source power from gas generators operated by Tallgrass and Black Hills Energy. A separate engineering and design study will evaluate adding carbon capture to Tallgrass’s combined-cycle turbines. A study is not a completed retrofit, and Google did not present it as one.

Power bills and water use

For local residents, the practical question is who pays for the electricity and infrastructure a large computing campus needs. Google says it will pay for all the power it uses and cover infrastructure needs directly driven by its growth. That is the company’s commitment; it is not a guarantee about every household’s future utility bill.

The campus will be air-cooled to limit strain on local water resources, Google said. Air cooling reduces the need for water in cooling, but the announcement does not describe the whole project as water-free.

Local benefits are still expectations

Google expects the project to support jobs and bring property-tax revenue to Laramie County and School District 1. Those are projected benefits, not a count of jobs already created or tax money already collected. The company also described grants and partnerships for education, skilled-trades training and local entrepreneurs.

Google and its partners plan a community council to address the effects of data-center growth around Cheyenne. For families and businesses, the issues to watch are whether the promised investment turns into local work, how the power commitments are carried out, and whether the planned emissions controls perform as described.

The announcement is a significant infrastructure commitment, but it does not settle the wider debate over the cost, resource use and pace of the AI computing buildout.

JBizNews Desk | Technology

© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.

Wheeled bins in an alley in Saltaire, Shipley, England. Illustrative archive image, not Curbie.

Illustrative photo of wheeled bins, not Curbie. “Wheelie Bins” by David Merrett, via Wikimedia Commons, CC BY 2.0; cropped.

Revere Robotics is developing a small autonomous robot called Curbie to handle a familiar household task: moving a wheeled trash bin to the curb for pickup and returning it afterward.

The company describes Curbie as a robot designed to couple to standard two-wheel residential roll-out carts. Rather than replace the bin with a proprietary smart container, the design uses the existing cart’s wheels and axle to move it between its storage spot and the curb.

A design in development, not a product launch

The distinction is important for homeowners interested in buying one. Curbie is in active development. The company materials reviewed do not announce a firm price or commercial release date, and its descriptions of field testing and early pilot work are company claims, not independent confirmation of reliable performance.

In a company post introducing the project, Boston-based Revere Robotics says the visuals are concept renders of a prototype under development, not final product footage.

How the proposed curb run would work

The intended sequence is straightforward: couple to the bin, take it to the curb on collection day, release it for pickup and bring it home after collection. The company says the design does not require a special bin or a redesign of municipal collection systems.

That is a proposed approach, not a guarantee of compatibility with every household cart, driveway or collection service. In a follow-up engineering discussion, the company describes challenges including dependable coupling across worn or bent axles and outdoor charging. Those details show why testing matters before the idea becomes an everyday household appliance.

What homeowners can take from it

If validated and brought to market, the design could reduce the work of hauling bins to and from the curb. For now, it is an engineering project to watch, not an off-the-shelf option that homeowners can assume is available to order.

Claims about payload, speed, weather performance and autonomous operation should be understood as development specifications or goals unless backed by completed testing. The reviewed materials do not establish that Curbie works without supervision in every setting.

JBizNews Desk | Boston

© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.

Generic desk scene with a laptop and plant, not a suspected worker or an affected company

Generic desk scene used for illustration; not a suspected worker or an affected company. Photo: Marcin Skalij / Unsplash.

NEW YORK, Nov. 5, 2026 – CrowdStrike says it is finding one or two suspected North Korean IT-worker cases every day, a warning that remote-hiring fraud can become a threat to company systems and sensitive information, Fox News reported Thursday.

Adam Meyers, the cybersecurity company’s head of counter adversary operations, told Fox News Digital that the workers apply broadly, from consumer businesses to defense contractors. In some cases, he said, CrowdStrike has seen access handed to malware teams when a target presents intelligence value.

Meyers did not say whether those teams successfully collected sensitive information after the handoffs. The daily figure describes CrowdStrike’s own suspected-case detections, not a government count of confirmed infiltrations or a nationwide total.

A hiring problem with a security consequence

The concern extends beyond paying a worker who has misrepresented an identity. Once someone is hired, ordinary employee permissions can provide access to corporate systems and data. Meyers said companies should treat identity verification as a security responsibility shared by recruiters and security teams, rather than simply an administrative step in hiring.

In an Oct. 13 public alert, the FBI said North Korean IT workers use stolen identities and increasingly rely on U.S.-based facilitators. Those facilitators may receive and host employer-issued computers, allowing people overseas to work through a U.S. physical location. The bureau said the schemes generate revenue for North Korea’s government and create national-security and intellectual-property risks.

The FBI also warned that workers target jobs with cleared defense contractors and other government contractors, and seek people with security clearances or specialized qualifications. That does not mean every suspected worker has reached sensitive systems, or that a job offer amounts to a successful intrusion.

Earlier federal actions show the breadth

The Justice Department’s Oct. 13 announcement described a charged scheme affecting more than 70 U.S. employers and generating approximately $3 million for the workers. The employers included businesses in health care, communications, cyber support, retail and manufacturing.

The coordinated announcement included charges against 14 people, two arrests and the seizure of approximately 187 computer devices. The Texas component, called Operation Lonestar Roundup, involved searches and other law-enforcement action at more than 40 residences known or suspected to be hosting computers for North Korean IT workers.

Those Texas actions took place Sept. 21-25 and were announced Oct. 13. They were not raids carried out Thursday. The approximately 187 devices included laptops and keyboard-video-mouse equipment, rather than 187 laptops alone. The charges are allegations, and defendants are presumed innocent unless proven guilty.

The employer count in that prosecution and CrowdStrike’s daily detections measure different things. They should not be added together or treated as estimates of the total number of affected American companies.

What employers can do without shutting out real applicants

The FBI recommends layered identity checks and access controls throughout employment, not reliance on a single document, interview or warning sign. It says human resources, IT, security and management teams should combine what they know when reviewing possible identity fraud.

For employers, that means checking that the person doing the work is the person hired, reviewing access in relation to the job, and taking multiple inconsistencies seriously enough to investigate. A review should continue after onboarding rather than end when a new hire receives a computer.

There is also a risk in overreacting. The FBI says aggressive filters can harm legitimate remote employees, exclude qualified applicants and penalize victims whose identities were stolen. It cautions against relying only on lists of names or other isolated indicators. Suspicion calls for substantive review, not an automatic conclusion about an applicant’s nationality or guilt.

Individuals should reject paid requests to lend personal information, account access or a home location to someone else’s employment arrangement, the FBI says. Businesses and people approached about a suspected scheme can contact their local FBI field office and submit a report through IC3.

Thursday’s development is CrowdStrike’s account of continuing suspected cases and possible access handoffs. It is not an announcement of new sanctions, a new federal operation or proof that every suspected case resulted in stolen information.

JBizNews Desk | New York

© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.

Generic circuit-board hardware, not an identified SanDisk or Micron product.

File photo: Albert Stoynov / Unsplash

SanDisk and Micron shares suffered double-digit losses Thursday as the selloff in artificial-intelligence stocks spread through the memory and storage companies that supply the technology behind data centers and consumer devices. Corning, Lumentum and CoreWeave also fell sharply as investors sold across the AI-infrastructure supply chain.

SanDisk’s CNBC quote showed $837.67 at 1:29 p.m. Eastern on November 5, down $199.73, or 19.25%. Micron’s quote showed $683.20 at 1:10 p.m. Eastern, down $112.03, or 14.09%.

The two quotes were recorded at different times, and prices can change sharply before the closing bell.

Memory stocks caught in the AI retreat

The declines were part of a much wider retreat across the technology sector. Reuters reported that SanDisk and Micron were among more than a dozen large technology stocks falling by double-digit percentages. Its report said the semiconductor index had fallen as much as 11% during the session.

Reuters linked the broader retreat to concerns about the pace and financing of the AI data-center buildout, alongside post-election restrictions that could limit construction. Strategists interviewed in that report offered differing assessments of the policy impact and the selling.

CNBC’s midday roundup also described steep losses across semiconductor and AI-infrastructure companies. Its earlier snapshot put SanDisk down about 20% and Micron down almost 13%, illustrating how the figures shifted during the session.

What the companies make

Micron makes DRAM, NAND and NOR memory and storage products. Its business includes cloud and data-center memory, making it part of the supply chain for AI and other computing-intensive applications.

SanDisk makes storage products based on NAND flash technology, including solid-state drives, removable memory cards and USB drives. Its products reach data-center, edge-computing and consumer markets.

Fiber, photonics and rented computing power fall too

In CNBC’s midday roundup, first published at 1:06 p.m. Eastern, Corning was down 13%, Lumentum 14% and CoreWeave 12%.

Corning supplies optical fiber, cable and connections used to move data through AI networks. Its exposure is to the physical links that let computing equipment communicate, rather than memory chips themselves.

Lumentum makes optical and photonic technology, including lasers, modules and optical subsystems used in data-center connections. CoreWeave provides AI cloud computing, including GPU computing, networking and storage used to train and run AI models.

Together, the declines show selling reaching beyond chipmakers into the equipment and computing services around them.

What the drop means outside Wall Street

The immediate impact is on the value of shares held directly by investors or through funds that own these companies.

JBizNews Desk | New York

© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.

Generic circuit-board illustration, not the task force or any specific technology

Generic circuit-board illustration, not the task force or any specific technology. Illustration: Mirella Callage / Unsplash.

WASHINGTON, Nov. 5, 2026 – The U.S. Justice Department on Thursday announced a new Super Intelligence Task Force to address unlawful uses of advanced artificial intelligence by foreign governments and state-backed actors, transnational criminal organizations, and individuals or companies engaged in fraud and other illegal conduct.

Associate Attorney General Stanley Woodward and Associate Deputy Attorney General Stephen Weiss will lead the group, known as DOJ-SITF, according to the department’s announcement. The release uses “Super Intelligence,” or SI, for the advanced AI technologies covered by its initiative.

A new Justice Department enforcement group

The task force is a new DOJ body, separate from the White House SI Force chaired by Jay Clayton, which was announced in October. Attorney General Todd Blanche said the department’s group would support its partnership with Clayton as the government seeks to prevent misuse of critical technologies at home and abroad.

Blanche said DOJ would continue identifying and prosecuting people whose use of the technology violates the law, including conduct that targets American taxpayers, critical U.S. infrastructure or national security. The announcement describes an enforcement and coordination initiative, rather than a new criminal statute.

Woodward said the effort would seek to promote innovation while preventing misuse. Weiss said the department would use available law enforcement tools to deter improper use, strengthen regular reporting on risks and responses, and ensure companies maintain strong internal controls so the technology behaves as intended.

Why companies and consumers should pay attention

The stated scope reaches beyond national-security threats to individual and corporate fraud. For businesses developing or using advanced AI, the announcement signals DOJ attention to internal controls and how risks and responses are reported. For consumers, the department identifies fraud and malicious targeting of taxpayers among the harms it intends to address.

Those priorities are not a detailed compliance rule. The release does not announce a reporting deadline, new funding, staffing numbers or criminal charges. It also does not explain how any regular reporting process would work.

The announcement links to a memorandum. JBizNews has not reviewed that document; this report is based on the department’s public release.

JBizNews Desk | Washington

© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.

Hands using a map and compass, a generic trip-planning illustration.

File photo: Daniil Silantev / Unsplash

NEEDHAM, Mass. – Tripadvisor announced Thursday morning that its chief executive and chief financial officer will leave the company, as the travel platform reported an 11% drop in third-quarter revenue and a large noncash charge against the value of its hotel-related business.

Shares were down 22% in midday trading, according to CNBC’s November 5 report. That is an intraday percentage change, not a closing result or a quoted share price.

Tripadvisor provides travel reviews, guidance and booking services and owns Viator, the marketplace for tours, activities and attractions. The announcements describe a leadership transition and pressure on revenue and profitability.

Goldberg stays through February as board searches

The board and Matt Goldberg agreed that he will depart as president, chief executive officer and a director, according to the company’s management announcement. Goldberg will remain CEO through February 16, 2027, to support the transition. The board is working with an executive search firm.

Michael Noonan, the chief financial officer, plans to leave for another opportunity and will continue in his role through December 31, 2026. Kristen Dalton, currently chief of strategy and operations, will become interim CFO on January 1, 2027. Dalton joined Tripadvisor in 2019 and has held several roles at the company.

Revenue falls, with the sharpest decline in hotels

For the quarter ended September 30, revenue from continuing operations was $436.4 million, down 11% from a year earlier, Tripadvisor said in its earnings release. Experiences revenue fell 3% to $284.8 million, while Hotels and Other revenue declined 23% to $151.6 million.

The loss from continuing operations was $316.9 million, or $2.70 per diluted share. That is distinct from the consolidated net loss, including discontinued operations, of $313.1 million.

The continuing-operations loss was driven primarily by a $369.5 million noncash goodwill impairment in Hotels and Other. Goodwill is an accounting asset, and this charge reduces its recorded value rather than representing a cash payment of that amount during the quarter.

Tripadvisor attributed the impairment to a sustained decline in financial performance and lower forecast profitability, largely because of worsening traffic pressures caused by changes in the search landscape.

On an adjusted basis, the company earned 37 cents per diluted share from continuing operations. Adjusted EBITDA, a measure of earnings before interest, taxes, depreciation and amortization with additional company adjustments, fell 34% to $71.1 million. These adjusted measures exclude items included in the reported accounting loss and should not be confused with it.

Restructuring aims for $40 million in annual savings

Tripadvisor began a restructuring on November 4 to simplify Hotels and Other, including reducing roles in non-core revenue lines, while directing investment and certain product, engineering and data-analysis roles toward Experiences.

The company expects approximately $40 million in gross annualized savings and $15 million to $20 million in pretax restructuring and related costs, mainly severance and benefits, primarily in the fourth quarter. Those are company estimates, not savings already delivered.

CNBC separately reported that Tripadvisor forecast a fourth-quarter revenue decline in the low teens compared with the third quarter. That is a sequential comparison, unlike the year-over-year third-quarter revenue declines reported above.

JBizNews Desk | Needham, Massachusetts

© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.

Archive photo of New Glenn launching from Cape Canaveral on January 16, 2025

New Glenn launches from Cape Canaveral on January 16, 2025; archive photo, not a current launch. Photo: U.S. Space Force / Senior Airman Samuel Becker (public domain).

Blue Origin has pushed the target for its New Glenn rocket’s return to flight to January 2027 as it rebuilds a Florida launchpad damaged in a May explosion, according to reports Thursday.

Reuters reported the delay, citing two people familiar with the matter and an email Chief Executive Dave Limp sent to employees Wednesday. CNBC also reported the new January target, saying Blue Origin had previously aimed to resume flights by the end of 2026.

January is a target rather than a firm launch date. Blue Origin did not immediately respond to Reuters’ request for comment, and the company has not confirmed the revised schedule in the reports reviewed by JBizNews.

What New Glenn does

New Glenn is the heavy-lift orbital rocket built by Jeff Bezos’ space company, Blue Origin. Its role is to carry payloads into orbit, including satellites, making its flight schedule important to customers planning space missions.

The roughly 322-foot rocket had launched three times since its 2025 debut, according to Reuters. It has been grounded since an explosion on Blue Origin’s only launchpad in May, days before a planned mission to launch Amazon satellites.

Launchpad repairs push back the return

The May blast badly damaged the Florida launchpad. Reuters reported that reconstruction delays led to the revised January goal described in Limp’s employee email.

The explosion is background to Thursday’s schedule news, not a new incident. The reported change moves the expected return from late 2026 into the following year while work on the launch site continues.

The reports do not establish a confirmed launch date for Amazon’s satellites. A target for New Glenn’s return to flight should not be treated as a promise that a particular customer mission will launch in January.

What remains uncertain

The next launch depends on the recovery work and the company’s readiness to resume operations. The reports reviewed here do not provide a firm day in January or establish that repairs to the damaged pad are complete.

For customers, the immediate development is a later planning target for New Glenn’s next flight. Blue Origin’s response and a confirmed launch schedule remain outstanding in the reporting.

JBizNews Desk | New York

© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.

Microsoft sign behind a reception desk

Photo: Alex Carmona / Unsplash (archive photo)

Microsoft declined an option to expand its Texas computing agreement with AI cloud provider Nscale, according to a report first published by the Financial Times and subsequently corroborated by an unnamed source speaking to Reuters. The decision was made several months ago, not as a new reversal Thursday, and the existing Texas contract remains in place, Reuters reported.

The option would have more than doubled the size of the existing agreement to rent computing power at a Texas data center, according to Reuters, carried by Investing.com. The FT put the potential extension’s value at tens of billions of dollars. That is the value of a possible expansion, not a disclosed cancellation of already contracted revenue.

An option is different from the existing contract

In this case, the option was a choice to take additional computing capacity beyond the existing agreement. Declining it does not, by itself, terminate the underlying contract. Reuters’ source said Microsoft’s decision preceded Nscale’s September IPO filing and did not affect its existing Texas contract.

The same source said the decision has no impact on Nscale’s approximately $103 billion total contract value disclosed in the filing. Optional capacity should therefore not be subtracted from that figure as though it were a signed order that has now disappeared. Nscale was already discussing the remaining Texas capacity with several major customers, the person told Reuters.

What Nscale supplies to Microsoft

Nscale is an AI cloud provider that supplies computing capacity for artificial-intelligence workloads. Under an expanded agreement struck with Microsoft last year, it planned to deploy roughly 200,000 Nvidia AI chips across data centers in Europe and the United States, Reuters reported.

Its IPO filing said that statements of work signed with Microsoft between September 2025 and April 2026 were worth up to about $43.8 billion through December 2033, excluding optional extensions, according to Reuters. That distinction separates the disclosed Microsoft work from the extra capacity Microsoft could have chosen to add.

The Texas expansion decision

The reporting rests on the FT’s initial account and Reuters’ unnamed person familiar with the matter; it is not a new announcement from either company. Microsoft and Nscale had not responded to Reuters’ requests for comment outside business hours.

This is separate from Microsoft’s previously reported withdrawal from an Nscale project in Essex, England. The Texas story concerns an unexercised expansion option, not a claim that the Essex withdrawal canceled the Texas agreement.

For investors following AI infrastructure, the distinction matters: a potential expansion can signal future demand without being part of firm contracted revenue. Thursday’s report narrows one route to additional Texas business, while leaving the existing agreement and disclosed total contract value unaffected, according to Reuters’ source.

JBizNews Desk | New York

© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.

Zipline drone launching at its California test facility in March2018. Archive illustration.

File photo: Roksenhorn / Wikimedia Commons / CC BY-SA 4.0 (cropped)

SOUTH SAN FRANCISCO, Calif. – Zipline said in a Thursday funding announcement that it raised $1.5 billion to expand its U.S. drone-delivery network, while Walmart agreed to add 300 delivery sites under a new multi-year arrangement. The financing values Zipline at $21.5 billion after the investment, according to the company.

The round was led by Paradigm, WCM Investment Management and Layer Global. Zipline said it will use the money to expand U.S. manufacturing facilities and hire more workers as it builds a national network for deliveries by automated aircraft.

The $21.5 billion figure is a post-money private-company valuation: it includes the new investment.

Walmart plans 300 additional sites

Walmart and Zipline’s agreement covers 300 additional sites over multiple years, expanding the companies’ existing relationship.

Walmart is the retail partner in that expansion; Zipline supplies the drone-delivery network. For shoppers, the idea is to have orders travel by air rather than entirely by road.

Delivery speed is a company-reported measure

Zipline said its median order in October arrived 19 minutes after checkout. The company also said it had made more than one million deliveries in the Dallas-Fort Worth area this year.

Participants in the financing also included Valor Equity Partners, Fidelity Management & Research Company, Uber, Tiger Global, J.P. Morgan, ARK Invest and Salesforce Ventures, Zipline said.

The funding and Walmart agreement address different parts of the expansion: capital for manufacturing and hiring, and a retail partnership for additional delivery locations. The next practical test is turning that planned network into operating sites that customers can actually use.

JBizNews Desk | South San Francisco, California

© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.

The south entrance of Expedia Group headquarters in Seattle

Photo: T014y / Wikimedia Commons (CC BY-SA 4.0)

Expedia shares were down about 12% Thursday in the regular-session snapshot reported by CNBC at 10:34 a.m. Eastern, a day after the travel company issued a fourth-quarter revenue forecast whose midpoint fell short of Wall Street’s estimate. In a Thursday interview, CEO Ariane Gorin said AI agents were helping with trip ideas and planning, but Expedia had not yet seen them change how customers actually book.

The stock decline is CNBC’s Thursday story snapshot, not a closing result or a fresh price quote. It follows the roughly 6% after-hours decline reported Wednesday in our original earnings coverage; the two moves describe different trading sessions and should not be added together.

AI helps planning, but bookings have not shifted yet

Gorin, Expedia Group’s chief executive, told CNBC that artificial intelligence was “a great way for people to do inspiration, to do trip planning.” She added: “We haven’t seen it change the way they’re actually doing booking.” That describes the company’s observation so far, not proof that AI agents cannot complete travel bookings or will never win market share.

Expedia is working with AI agents on visibility and integrations, Gorin said, including understanding how they obtain information and ensuring its services appear in their results. She acknowledged that agents could develop travel programs of their own and eventually take market share.

Her argument for established travel providers centers on trust at checkout and support afterward: accurate amenities, correct prices and someone available to help when a trip goes wrong. Those are Gorin’s reasons for believing travelers will keep booking with Expedia, rather than an independent finding that agents cannot provide those services.

Why the fourth-quarter forecast disappointed

The fourth-quarter revenue guidance, issued Wednesday, was $3.69 billion to $3.75 billion. Its $3.72 billion midpoint was below the $3.74 billion FactSet analyst estimate cited by CNBC.

Gorin attributed the moderation to geopolitical events in the Middle East and Europe, the midterm elections and a fall sale moved from the fourth quarter into the third quarter. She did not attribute the softer forecast to AI taking Expedia bookings. Moving a promotion between quarters can also change the timing of reported demand, which is different from showing that customers have abandoned the platform.

Wednesday’s results showed travel growth

Expedia’s earnings release showed third-quarter revenue rose 12% to $4.938 billion and gross bookings increased 11% to $34.008 billion. Travelers booked 115.7 million room nights, up 7%, while gross bookings from business partners rose 19%. These are the results announced Wednesday, not new Thursday figures.

The company raised its full-year revenue forecast to $16.37 billion to $16.43 billion and full-year gross bookings guidance to $129.5 billion to $130.8 billion. Adjusted earnings were $9.39 a share, and adjusted EBITDA rose 17%. Expedia also reported repurchasing about 2 million shares for $576 million in the quarter and declared a 48-cent quarterly dividend payable December 10.

For investors, Thursday’s interview separates two questions: the near-term revenue outlook and the longer-term effect of AI on travel distribution. A strong past quarter can coexist with a forecast that disappoints, while the CEO’s current booking observations leave open how customer habits may change.

JBizNews Desk | Seattle

© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.

Apple Park in Cupertino, California, photographed from above in September 2022. Archive photo.

Photo: InvadingInvader / Wikimedia Commons / CC BY-SA 4.0. Apple Park, September 2022. Resized; adaptation under the same license.

CUPERTINO, Calif. – Apple is seeking about $7 billion in a U.S.-dollar bond offering, Bloomberg News reported Thursday, citing people familiar with the matter. The proposed borrowing would span seven series of debt maturing from 2028 through 2036.

The amount is not final. Bloomberg reported that the sale’s size could change with investor demand, according to Reuters’ account of the report. Reuters said it could not independently verify the report and that Apple had not immediately responded to a request for comment.

A seven-part offering, with terms still open

Apple’s preliminary prospectus supplement, dated November 5, describes two floating-rate notes due in 2028 and 2029, plus five fixed-rate notes due in 2028, 2029, 2031, 2033 and 2036. It leaves the principal amounts and interest-rate terms blank and says its information is incomplete and may change.

The distinction matters: the filing establishes the structure of the proposed offering, but it does not confirm the roughly $7 billion figure reported by Bloomberg or show a completed sale.

Goldman Sachs, BofA Securities, Deutsche Bank Securities and J.P. Morgan are the joint book-running managers listed in the document. Reuters reported that Goldman declined to comment on the offering’s size, while the other banks did not immediately respond to its requests.

Proceeds for general corporate purposes

Apple says it intends to use the net proceeds for general corporate purposes. The filing lists stock repurchases and dividends, working capital, capital expenditures, acquisitions and repayment of debt among the possible uses. It also says money not immediately needed could be held in short-term investments.

The document does not earmark the borrowing for an artificial-intelligence project. Although large technology companies have been raising debt amid an expansion in AI spending, that broader industry backdrop is not the same as a designated use of Apple’s proceeds.

The proposed notes would be senior unsecured debt, ranking equally with Apple’s other senior unsecured and unsubordinated obligations. The filing says the company had $78.1 billion of unsecured senior notes and $2 billion of unsecured short-term promissory notes outstanding as of September 26.

Apple last sold bonds in May 2025, raising $4.5 billion in a four-part transaction that attracted $10 billion in orders, Reuters reported. For the new offering, the final amount, pricing and allocation across maturities remain the details to watch.

JBizNews Desk | Cupertino, California

© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.

Photo: Nvidia CEO Jensen Huang presents the RTX Blackwell chip at CES 2025 in Las Vegas / Pronoia, Wikimedia Commons / CC0

By: JBizNews Desk 🌎

Moonshot AI, one of China’s leading artificial intelligence labs, has sought to rent access to Nvidia’s most advanced chips inside European data centers, a way around American rules that block the chips from being sold to China, CNBC reported, citing four people familiar with the matter.

The talks, held this year, were about renting computing power from so-called neoclouds, companies that run large numbers of Nvidia graphics chips and rent out capacity. Three of the people said Moonshot wanted clusters built on Nvidia’s Blackwell systems, the chipmaker’s newest and most powerful line. U.S. export controls effectively ban those chips from being shipped to China, but the law does not give regulators broad power to stop a Chinese company from using them remotely through data centers overseas, a gap often described as a loophole.

Two of the people said Moonshot was interested in clusters for inference, the work of running finished models for users. One said there was also interest in clusters for training new models. Talks covered arrangements for clusters of 8,000 or more Blackwell chips, one person said, describing Moonshot as “very aggressive” in its hunt for computing power. None of the people said a deal had been completed.

One person said it was proposed that Moonshot pay 30% to 50% more upfront than a non-Chinese customer, because cloud providers and their lenders want protection in case Washington later bans remote-access arrangements.

Some of the talks took place after July, when the White House said Moonshot had accessed Blackwell chips through a facility in Thailand, one person said. Moonshot did not comment at the time.

Moonshot, founded in 2023 and based in Beijing, builds the Kimi family of AI models. It said a new model released in July closed the gap with leading systems from American labs OpenAI and Anthropic, and it briefly paused new subscriptions, saying demand had stretched its computing capacity. Chinese AI models have surged in global adoption this year, helped by lower prices.

Lawmakers in Washington have proposed closing the loophole with the Remote Access Security Act, which would extend export controls to cloud-based access to advanced hardware. It passed the House of Representatives in January but has not passed the Senate. The Financial Times reported this week that Nvidia has told cloud providers renting access to its chips that they must sell at least 51% of services to non-Chinese customers. Nvidia said in August that Washington had licensed some sales of its older H200 chips to China, but that Beijing restricted those sales as it pushes domestic chipmakers.

“America cannot win the AI race unless America’s AI infrastructure wins worldwide,” an Nvidia spokesperson said when asked about the reports.

Moonshot, Nvidia, the White House, the U.S. Department of Commerce, China’s embassy in the U.K. and China’s Ministry of Foreign Affairs did not respond to CNBC’s requests for comment.

JBizNews Desk | BEIJING

© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.

Photo: Google servers inside a data center / Wesley Nitsckie, Wikimedia Commons / CC BY-SA 2.0

By: JBizNews Desk 🌎

Google has agreed to backstop Anthropic’s 20-year, $19 billion lease of bitcoin miner TeraWulf’s 401-megawatt Kentucky data center site, and could end up owning a stake in the project company if Anthropic defaults, CoinDesk reported, citing a securities filing.

Under the terms, signed November 4, Google may take over the lease if Anthropic defaults or becomes insolvent, or under certain conditions pay a termination fee in exchange for proportional ownership of Raylan Data, the subsidiary that owns the project – not shares in TeraWulf itself, CoinDesk reported.

The takeover requirement generally applies during the first six years of the lease, with an exception for TPU pods using less than half the reserved power, according to the report. Google’s obligations end three months after Anthropic or its investment group reaches an investment-grade credit rating.

Computing capacity from the site is expected to be delivered from late 2027 through the second quarter of 2028, CoinDesk reported.

The arrangement is a guarantee, not an acquisition: Google is not buying TeraWulf, and any ownership would be conditional on an Anthropic default. TeraWulf said in July it was raising $3.5 billion for the Anthropic-leased data center, JBizNews reported.

JBizNews Desk | New York

© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.

Photo: A data center cooling tower / Rsparks3, Wikimedia Commons / CC0

By: JBizNews Desk 🌎

Artificial-intelligence data centers billed as using “closed-loop” cooling can still draw millions of gallons of water a day, because shedding heat to the outside air trades water against electricity, especially in hot weather, CNBC reported.

Closed loops use little water inside the building. The trade-off comes at the cooling equipment outside, which can use more water on hot days or more electricity to stay dry.

Meta’s planned 1-gigawatt data center in Lebanon, Indiana, could have a water allocation of up to 8 million gallons a day by 2031, according to its law firm, CNBC reported. Meta says that figure is a peak for the hottest days of the year, reached only a few days annually, not an average.

Elon Musk’s xAI Colossus 1 site near Memphis has an allocation of up to 1.28 million gallons a day from Memphis Light, Gas and Water and has not reached it, CNBC reported. Musk’s proposed $80 million water recycling plant for the site was put on hold in April.

For residents weighing a proposed AI campus, the report said, the key numbers are both the peak water allocation and the electricity the site will need.

JBizNews Desk | New York

© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.

Photo: The Binance logo / Binance, Wikimedia Commons / Public domain

By: JBizNews Desk 🌎

Binance said it revoked all API keys linked to the trading platform SignalPlus as a precaution after a third party reported a potential security issue, CoinDesk reported.

Binance said its own systems are unaffected and that customer funds are safe.

The exchange told users to review their positions and open orders directly and to wait before reconnecting their accounts, CoinDesk reported.

JBizNews Desk | New York

© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.

Photo: A Bezeq office in Tiberias / Geagea, Wikimedia Commons / CC BY 3.0

By: JBizNews Desk 🌎

Bezeq, Israel’s largest telecom company, is joining the country’s data-center building boom. Its board approved a non-binding memorandum of understanding with real-estate firm Adgar to build a server farm of at least 16 megawatts in Petah Tikva’s Kiryat Arye industrial zone, Globes reported.

The two companies entered a 90-day exclusivity period to reach a final agreement. Under the plan, they would split revenue, expenses and investment equally, with each side putting in NIS 200 million to 250 million of its own money, NIS 400 million to 500 million combined. Most of the spending falls in 2027 and 2028, with operations targeted for 2028.

Adgar, controlled by the Schneidman family, brings the land, the building, access roads and physical infrastructure, and already holds a binding commitment for a 16-megawatt electricity connection. Building permits are expected in the coming weeks but have not yet been granted. Bezeq would buy the systems equipment, lead the technological build-out, and run operations and marketing.

Bezeq forecasts annual revenue of NIS 50 million to 80 million and EBITDA of NIS 25 million to 40 million as occupancy ramps up, figures that are projections, not guarantees. Both companies’ shares rose after the announcement, Globes reported, without giving percentages.

The move would give Bezeq a new growth engine beyond traditional phone and internet service, tapping rising demand for data centers and AI infrastructure. Adgar had previously planned a 38-story office tower on the site before shifting to data centers amid the real-estate slowdown.

JBizNews Desk | Tel Aviv

© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.

Photo: An Apple iPad Pro 11-inch / P1898, Wikimedia Commons / CC BY 4.0

By: JBizNews Desk 🌎

Fewer people are buying tablets. Global tablet shipments fell 23% from a year earlier to 32 million units in the third quarter, the lowest level in nearly three years, research firm Omdia reported. Shipments are devices sent to stores and partners, not sales to shoppers.

Apple kept its lead even though its shipments fell 28%, and its market share slipped to 36% from 38.2% a year earlier. Samsung shipped 5.4 million tablets, down 22%, for a 16.9% share. Lenovo was the only top-five vendor to grow, with shipments up 12% to 4.1 million units, lifting its share to 12.7% from 8.7%. Xiaomi shipped 2.3 million tablets, down 11%, and Huawei fell 43% to 1.8 million, the steepest drop among the top five.

Omdia blamed tighter supply of parts and weak demand, saying recent launches have done little to get shoppers excited because new models are not different enough from the old ones. The firm’s analysts forecast that a meaningful recovery is unlikely before 2028.

In plain terms: people are holding on to their old tablets longer, and with parts scarce, the companies making them are focusing on other products first.

JBizNews Desk | Tech

© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.

Intel CEO Lip-Bu Tan (right) with Commerce Secretary Howard Lutnick in 2025

Photo: Intel CEO Lip-Bu Tan (right) with Commerce Secretary Howard Lutnick in 2025 / U.S. Department of Commerce, Wikimedia Commons / public domain

Intel CEO Lip-Bu Tan bought about $9.7 million of the chipmaker’s stock this week, a show of confidence after the shares tumbled on a weak forecast.

Tan bought 150,000 shares on November 2 through a family trust at a weighted average price of $64.73, according to a Form 4 signed November 4, CTech reported. The trust now holds 1,464,669 shares.

Intel shares closed Wednesday at $58.27, down 3.77%, IBTimes reported. The stock’s slide followed a weak fourth-quarter forecast; the reports do not link Wednesday’s decline to Tan’s purchase.

It is Tan’s second big buy in months. Since August he has purchased a combined 255,263 shares for about $19.7 million, a stake now worth about $14.9 million, per CTech — a paper loss on the earlier buys.

For readers: when a CEO spends his own money buying company stock, investors often read it as a bet that the market is being too pessimistic. It is a signal, not a guarantee.

JBizNews Desk | New York

© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.

Tux, the Linux mascot, at Linux Day 2022 in Milan

Photo: Tux, the Linux mascot, at Linux Day 2022 in Milan / Kowalski7cc, Wikimedia Commons / CC0

OpenAI says one of its AI models found and validated hundreds of security holes in the Linux operating system, a sign that artificial intelligence is getting much better at a job that used to take human researchers weeks per bug.

In a report published November 4, the company said an automated pipeline scanned the Linux kernel, reproduced 1,724 vulnerabilities in a controlled environment, and generated 1,169 working exploits for what are known as local privilege escalation bugs — mapping to 948 distinct vulnerabilities, for which it developed candidate patches — flaws that let someone who already has basic access to a computer gain full control of it.

Two caveats matter. First, these are local attacks: they require the attacker to already have access to the machine, not remote takeover of Linux systems. Second, the 948 findings are candidates OpenAI reported with proposed patches for maintainers to review — not 948 fixes already shipped. Human researchers reviewed the outputs before anything was reported.

The run cost about $1.2 million in computing, or roughly $1,279 per distinct vulnerability. OpenAI noted that applying historical bug-bounty rates to the findings would imply a hypothetical value of $18.9 million to $66.3 million, though that figure is theoretical, not money earned.

For readers, the takeaway is about direction, not alarm: AI can now comb vast amounts of code and flag weaknesses at machine speed. Security teams hope that means holes get found and patched by defenders first — the reason OpenAI says it is working with the Linux community on reporting and fixes.

JBizNews Desk | New York

© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.

Beacon Security founders

Photo: Beacon Security founders / Beacon Security

Palo Alto Networks is in talks to acquire Beacon Security, a two-year-old Israeli cybersecurity startup, Calcalist reported.

The talks are separate from Palo Alto’s reported negotiations to buy another Israeli startup, Clutch Security, for about $500 million.

Beacon, founded in 2024, builds a platform that lets security teams deploy AI agents to detect, investigate and respond to cyber threats. It aggregates data from hundreds of sources across an organization, normalizes and enriches it, and gives AI agents a unified view of the company’s security environment. The company also offers a library of ready-to-use security agents.

Beacon raised $13 million in a July seed round led by Notable Capital. The company says its annual recurring revenue grew more than 300% in the first half of 2026 — a company claim, with revenue and customer numbers not disclosed.

The founders are CEO Gal Tal-Hochberg, Chief Product Officer Or Mattatia and Chief Technology Officer Ido Israeli, all veterans of the IDF’s intelligence units 8200 and 81. Tal-Hochberg previously co-founded HiredScore, an AI recruitment company.

JBizNews Desk | Tel Aviv

© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.

A Polish army K9 Thunder self-propelled howitzer

Photo: A Polish army K9 Thunder self-propelled howitzer, built by South Korea’s Hanwha / Staff Sgt. Matthew Foster, U.S. Army via Wikimedia Commons / public domain

South Korea’s arms makers rode a two-year boom as global rearmament sent buyers their way. Now the rally has reversed, and investors are asking whether the drone era has changed the game.

The combined market value of the country’s top four defense companies — Hanwha Aerospace, LIG D&A, Hyundai Rotem and Korea Aerospace Industries — stood at about 62 trillion won at the end of October, down from a 52-week peak of about 170 trillion won, broadcaster SBS reported, per CNBC. That snapshot predates this week’s trading.

The war in Ukraine has shown how cheap drones can threaten expensive tanks and artillery, and Iranian drone strikes have drained costly air-defense missile stocks. “Modern warfare is being fought with less traditional weaponry, which Koreans are positioned for,” Peter Kim of KB Securities told CNBC. Analysts are not saying tanks and artillery are obsolete, but procurement priorities are shifting toward drones, counter-drone systems and air defense.

The sell-off is not all about new warfighting technology. Investors are weighing disappointing earnings, a stronger won, delayed export orders, high valuations and money rotating into chip stocks. Hanwha Aerospace’s third-quarter operating profit fell more than 46% from a year earlier on delivery-schedule changes and one-off costs, and its shares are down more than 28% this year. Hyundai Rotem’s profit fell more than 11.1%, with shares down almost 60%. Korea Aerospace Industries was the bright spot: sales up about 52%, operating profit up 32%, limiting its share decline to just over 8%.

For defense investors and suppliers, the takeaway: bigger military budgets don’t guarantee fast orders or rising share prices.

JBizNews Desk | Global

© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.

NextDC’s M3 data center in Melbourne at dusk.

Photo: NextDC’s M3 data center in Melbourne / Courtesy of NextDC

NextDC shares closed 50.84% higher in Australian trading Thursday after a vehicle jointly owned by Goodman Group and Digital Realty sought to buy up to 9.9% of the data-center operator through an aftermarket bookbuild, CNBC reported.

Ackland Investments No.1 Trust, owned equally by Goodman and Digital Realty, sought shares at A$9.50 apiece after Wednesday’s market close. The targeted holding is a minority stake, not a takeover of NextDC.

NextDC says outcome is not known

In its November 5 statement, NextDC said it did not know the outcome of the bookbuild and had no takeover proposal before its board. The company described the move as opportunistic.

NextDC also noted that Goodman and Digital Realty had told investors through their broker that they did not intend to submit a control proposal or takeover bid at this time, while reserving the right to do so later.

The A$9.50 bookbuild price was a 45.5% premium to Wednesday’s close. Data Center Dynamics put the proposed stake purchase at about A$715 million in its report on the offer. Neither the targeted percentage nor that reported value should be read as confirmation of a completed acquisition.

Power delays set the backdrop

The approach follows NextDC’s October 21 disclosure of a power delay at its M3 Melbourne facility and revised fiscal 2027 guidance. That earlier setback had sharply reduced the company’s share price.

The rally ran against the tide: U.S. data-center stocks sold off overnight, with CoreWeave down about 9% and Iren down roughly 11% on Wednesday, after the Democratic midterm sweep raised expectations of tighter oversight of data-center development, CNBC reported.

NextDC appointed Barrenjoey and UBS as joint lead financial advisers, Cadence Advisory as an independent financial adviser, and Mallesons as legal adviser.

JBizNews Desk | Sydney

© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.

Nintendo Switch 2 console in its dock

Photo: Nintendo Switch 2 console in its dock / Crisco 1492, Wikimedia Commons / CC BY-SA 4.0

Sales of Nintendo’s flagship Switch 2 console fell 41% in the July-to-September quarter after the company raised prices in the face of soaring memory-chip costs, even as profits climbed.

Nintendo sold 2.67 million Switch 2 units in the three months ended Sept. 30, down 41% from a year earlier and below the 3.82 million sold in the prior quarter, CNBC reported. During the quarter, Nintendo raised the console’s price in the U.S., Canada and Japan, after an earlier increase in Japan.

The hardware slump did not translate into a profit drop. For the six months through September, Nintendo sold 6.49 million hardware units, with revenue of 966.0 billion yen, down 12.1%, while operating profit rose 49.1% to 216.5 billion yen and net profit climbed 11% to 220.9 billion yen, according to the company’s results. Nintendo left its annual forecast unchanged.

The numbers tell a two-sided story: pricier hardware is selling fewer units, but the company is making more money on what it sells. Memory costs have been rising across the electronics industry as data centers soak up chip supply, pressuring console makers’ margins.

JBizNews Desk | New York

© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.

Legrand headquarters in Limoges, France

Photo: Legrand headquarters in Limoges, France / Babsy, Wikimedia Commons / CC BY-SA 3.0

Legrand, the French maker of electrical and data-center equipment, raised its full-year sales target for the second time this year after strong demand from data-center builders drove results above expectations in the first nine months of 2026.

Sales rose 22% excluding currency effects over the nine months, the company said, powered by organic growth in data centers of close to 40%. Adjusted operating margin came in at 20.8% of sales, and net profit rose 19.6%. Data centers now account for about a third of the company’s business.

Legrand now targets 2026 sales growth of 20% to 22% excluding currency, including organic growth of 11% to 13%, up from a previous target of 8% to 10%. The company has made seven acquisitions this year, all in energy and digital-transition markets.

Chief Executive Benoît Coquart told Reuters the company could benefit if criticism of energy-hungry data centers pushes the industry toward more efficient designs. Democrats’ midterm sweep, including in Texas, has raised fears of a backlash against data centers, which local communities criticize for heavy energy and water use. “This should lead to solutions that are more energy-efficient,” said Coquart, whose view is his own, adding that 40% of Legrand’s data-center revenue comes from products that help save energy.

Coquart said he does not expect a market decline, saying customers plan to keep investing: “We are more likely to increase than to reduce our investments,” he quoted them as saying.

JBizNews Desk | New York

© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.

Photo: Seoul’s Yeouido financial district at night / Ox1997cow, Wikimedia Commons / CC BY-SA 4.0

South Korea’s stock market closed out its worst day of the year Thursday as the chip selloff deepened through the session, while investors kept rotating money into China.

The Kospi closed down 5.28% at 5,383.80, a 300-point drop, with Samsung Electronics falling about 7% and SK Hynix about 9% at the close, Chosun Biz and Aju News reported. The slide triggered a sell-side sidecar, a brief trading curb, after Kospi 200 futures fell more than 5%, Yonhap reported.

In Tokyo, chip-testing equipment maker Advantest plunged more than 13% and Tokyo Electron lost about 9% intraday, Calcalist reported in its running market updates.

The mirror image: Hong Kong’s Hang Seng was up about 3.5% in afternoon trading at 24,755, extending the rally that earlier sent the Hang Seng Tech Index up as much as 5.9%, with e-commerce giants Tencent, Alibaba and Meituan leading, Bloomberg reported.

Why the election is splitting Asia’s markets

The Democratic sweep of Congress has investors weighing whether the new House and Senate will tighten regulation of AI data centers, hitting the chipmakers that supply them. “The reality is markets hate uncertainty, and now with the election out of the way, we are seeing buyers step in across the board,” said Ryan Detrick, a strategist at Carson Group. Others, like KCM Trade’s Tim Waterer, see the rotation into China continuing.

Oil keeps falling; U.S. futures slip

U.S. oil fell another 1.6% to $86.67 a barrel and Brent traded at $93.61 after Wednesday’s 8% plunge, Calcalist reported, while Nasdaq futures slipped 1%. U.S. government bonds rose for a second day, pushing the 10-year Treasury yield to 4.84%, its lowest since Sept. 10, and gold rose 2.1% to $4,144 an ounce, according to Bloomberg.

For American households, cheaper oil can mean lower gas prices, and lower bond yields can eventually mean lower borrowing costs. Investors now turn to Friday’s U.S. jobs report for the next read on the economy.

JBizNews Desk | Global

© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.

Palo Alto Networks headquarters in Santa Clara, California

Photo: Palo Alto Networks headquarters in Santa Clara, California / Namaste jinx, Wikimedia Commons / CC BY-SA 4.0

Palo Alto Networks is in talks to buy Israeli cybersecurity startup Clutch Security for about $500 million, Calcalist reported, citing The Information, which cited people familiar with the talks.

The talks are not signed and neither company has made an official announcement, according to the report.

Clutch, founded in late 2023, builds a platform that secures what the industry calls non-human identities: the API keys, service accounts, tokens, passwords and permissions that let software systems and AI agents talk to each other. As companies roll out more AI agents, the number of those digital identities and permissions keeps growing, and they have become a significant target for attackers.

The platform maps every digital identity across a company’s environment, flags excess or exposed permissions, manages their life cycle and detects suspicious activity. Clutch has also been expanding into securing AI agents specifically.

The company was founded by CEO Ofir Har-Chen, CTO Sagi Haas and R&D VP Tal Kimchi. It has raised about $28.5 million to date, per Calcalist: an $8.5 million seed round in 2024 led by Lightspeed Venture Partners with Merlin Ventures and private investors including Shlomo Kramer, Exabeam co-founder Nir Polak and HashiCorp co-founder Armon Dadgar, followed by a $20 million Series A in January 2025 led by SignalFire. Clutch operates out of Tel Aviv and the United States.

If completed at the reported price, the deal would mark a significant exit for the young company and its backers, and another Israeli acquisition for Palo Alto Networks, the Santa Clara-based cyber giant.

JBizNews Desk | Tel Aviv

© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.

Richard Branson speaking at the 2015 Our Ocean conference

Photo: Richard Branson / U.S. Department of State / Public domain

Some billionaire investors saw a threat to the artificial intelligence boom in Tuesday’s Democratic election victories. Stock investors delivered a different verdict Wednesday: they bought shares.

The split matters to anyone with technology stocks in a retirement account, a business counting on AI investment or a community weighing a new data center. A slower building boom could hurt some operators while reducing the spending pressure on the biggest technology companies.

That tension ran through the reactions of major donors and investors after the midterms, CNBC reported Wednesday evening. The network described both a dispute over billionaire political giving and warnings about what new political resistance could mean for AI infrastructure.

The AI worry: fewer data centers

Pershing Square founder Bill Ackman warned on CNBC before polls closed Tuesday that slowing AI development could put America’s technological edge at risk.

During an election-night livestream of the All-In podcast, investor Chamath Palihapitiya linked the future of the AI trade to data-center construction. Altimeter Capital CEO Brad Gerstner warned on the same livestream that the Democratic wave could keep at least 10 to 15 gigawatts of expected AI data-center capacity from coming online in 2027, CNBC reported.

Those are investors’ warnings, not confirmed project cancellations. CNBC also cited analyst Dan Ives’ estimate that 10% of planned U.S. data centers would no longer proceed. That figure is a forecast, not a measured total of abandoned projects.

Texas was one focus of the concern. Democratic Gov.-elect Gina Hinojosa, who defeated Republican incumbent Greg Abbott, has proposed a moratorium on new data-center construction until stronger safeguards are in place, according to CNBC. A proposal is not an enacted statewide ban.

AMD Chief Executive Lisa Su told CNBC that technology companies need to show local communities the benefits of AI’s expansion. Her response put the practical challenge in focus: winning public support for the facilities behind the technology.

Wall Street bought anyway

The Dow rose 1,569 points, or 3%, Wednesday, while the S&P 500 gained 1.6% and the Nasdaq Composite rose 1.4%, CNBC reported. Big Tech companies added more than $750 billion in market value at one point during the session, the network said. That was an intraday figure, not a closing-market-value claim.

CNBC described a possible upside to slower construction for large technology companies: a lighter spending burden and greater value for existing data centers. Smaller operators that depend heavily on borrowing faced a different market reaction, with their shares falling, according to the report.

The rally does not settle whether particular AI projects will be delayed, or whether Wednesday’s gains will last. It shows that concern about the building pipeline did not produce a single verdict across the stock market.

Branson challenges the money race

Virgin Group founder Richard Branson criticized the billions spent trying to influence the election and called for campaign-finance reform, CNBC reported. LinkedIn co-founder Reid Hoffman also called for getting money out of politics while defending his own Democratic donations.

CNBC’s analysis of campaign-finance records put Elon Musk’s giving at $474 million nationwide, including $224 million in Texas, where his preferred Republicans lost the Senate and governor’s races. Those totals are attributed to the network’s analysis, rather than an independent JBizNews tally.

The donors’ argument and the market rally point to different questions after the vote: how much political money can achieve, and which businesses gain or lose if the AI construction boom slows. Neither answer can be read from one day’s share prices alone.

JBizNews Desk | Business

© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.

Samsung DDR4 memory module photographed by D-Kuru

Photo: D-Kuru / Wikimedia Commons / CC BY-SA 4.0

Samsung’s chip rivals in China are competing for more than customers. They are recruiting experienced engineers who know how to make memory chips, offering pay, stock and help for their families as they try to close the technology gap.

The growing talent fight is putting more pressure on established chipmakers to keep key staff and protect their manufacturing knowledge, CNBC reported late Wednesday.

For chipmakers, the stakes go beyond a hiring bill: retaining experienced engineers and protecting know-how matter more as Chinese competitors grow stronger. That does not mean every employee who changes jobs takes trade secrets with them.

Pay packages reach beyond salary

Chinese semiconductor companies have recruited experienced overseas engineers with high salaries, stock options, housing and family support, Counterpoint Research senior analyst Jeongku Choi told CNBC.

Choi said their expertise has helped Chinese chipmakers work through technical bottlenecks as U.S. export restrictions limit access to advanced chipmaking equipment. Chinese manufacturers once dismissed as unlikely to catch up are becoming meaningful competitors, in her assessment.

Counterpoint’s Memory Tracker for the second quarter of 2026 projected that ChangXin Memory Technologies, or CXMT, would reach 10% of global DRAM revenue. That is an attributed projection, not confirmation that the company has reached that share. DRAM is a type of working memory used in electronic devices.

Keeping engineers, protecting knowledge

Established chipmakers are strengthening legal protections and internal security while changing compensation packages to keep key engineers, Choi told CNBC.

A former Samsung employee told the network that the company had used two-year non-compete agreements and monitored where some departing employees went. CNBC said the person asked not to be identified because the information was private.

The legal stakes are already visible. CNBC reported that a former Samsung engineer who later joined CXMT received a five-year prison sentence on appeal on October 22 for leaking Samsung technology, reduced from seven years at the first trial. The reported ruling was last month, not a new decision Wednesday night.

A job change is not proof of a leak

Lawyers interviewed by CNBC described a difficult boundary between an engineer’s general skills and a company’s protected information. Knowledge built through experience is not automatically a trade secret, and proving that a rival used a proprietary manufacturing method can be difficult.

Jamie J. Yang, a partner at Taiwan’s Innovatus Law, told the network that semiconductor production is so complex that one engineer is unlikely to know enough to reproduce an entire production line at another company.

For established manufacturers, the competitive challenge is therefore both practical and legal: make important employees want to stay, define which information is confidential and limit access to it. The report does not establish how many engineers have moved or what the competition will do to chip prices.

Samsung and CXMT did not respond to CNBC’s requests for comment, the network said.

JBizNews Desk | Global

© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.

Aerial view of Apple Park in Cupertino in June 2024.

Photo: Nils Huenerfuerst / CC0

Apple cut daily artificial-intelligence spending allowances for some engineers to $200 from $300 last month and is considering further reductions, Bloomberg reported Wednesday, citing people familiar with the changes.

Some engineering teams could see their caps fall to $100, according to Mark Gurman’s report. Apple has not publicly confirmed the changes and did not immediately respond to a Reuters request for comment.

For employees using AI to write code and handle other work, the reported changes would put tighter limits on usage spending, with larger allowances requiring approval.

Limits cover usage of outside AI models

The allowances cover tokens, the units of data processed by AI models such as OpenAI’s ChatGPT, Anthropic’s Claude and Google’s Gemini. Coding work can consume large numbers of tokens, making engineers among the heaviest potential users.

A Reuters account of Bloomberg’s reporting said some non-engineering groups already have $100 daily limits that could be reduced to as little as $25. Employees who need more access would be able to seek exemptions from senior executives.

Cheaper models and on-device processing

MacRumors, also citing Bloomberg, said some teams were being asked to use cheaper models or move work to on-device AI.

Some employees told Bloomberg that few workers were using the full $300 daily allowance. If so, reducing the cap would not necessarily produce savings equal to the difference between the old and new limits.

The reported limits concern how much employees can spend on AI use, not a disclosed change to Apple’s consumer AI products. The company has not announced a total savings target tied to these allowances.

The unanswered business question is how the lower budgets would affect actual usage and productivity. A cap is a ceiling, not evidence of what each employee was spending, and the possible additional cuts remain unconfirmed.

JBizNews Desk | Cupertino

© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.

Canary Wharf financial district in London at night

Photo: Seb Tyler / Wikimedia Commons / CC BY 3.0

Europe’s biggest fintech just got a lot bigger – and it is coming for American banking customers next.

Revolut confirmed Wednesday that investors have completed a share sale valuing the London-based company at $115 billion, a 53% jump from the $75 billion price tag it carried a year ago and more than double its $45 billion valuation in 2024, TechCrunch reported.

The sale was a so-called secondary transaction: existing investors and employees sold their shares – at about $2,017 each – to buyers led by Glade Brook Capital Partners, Coatue, Durable Capital Partners, T. Rowe Price and Baillie Gifford. No new money went into the company, and Revolut did not say how much stock changed hands, Tech.eu reported.

The new number cements Revolut’s status as Europe’s most valuable private tech company. The firm, which says it has more than 80 million retail customers worldwide, has recently secured banking licenses in the U.K. and Europe and is preparing to launch a U.S. bank in 2027.

For everyday users, the valuation race matters because it determines how fast Revolut can roll out full banking services – from checking accounts to loans – in new countries, including the United States, where it would compete directly with the big consumer banks.

JBizNews Desk | London

© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.

Archive photograph of the BuzzFeed entrance in New York, November 19, 2020.

AP Photo/Ted Shaffrey, File

BuzzFeed brought in barely more than half as much money this summer as it did a year earlier, and its losses grew.

The digital publisher reported third-quarter revenue of $26.0 million, down 44% from $46.3 million a year earlier. Its net loss widened to $18.7 million from $7.4 million, according to its earnings release.

For people who work in digital media or buy ads there, the numbers show the pressure on a familiar business model: turning readers’ attention into advertising dollars and shopping commissions. BuzzFeed’s biggest percentage decline came from commerce and other revenue, the business that includes commissions when readers buy through its shopping coverage.

The shopping business took the hardest hit

Commerce and other revenue fell 71% to $4.8 million from $17.0 million. Advertising, still the largest revenue source, fell 31% to $15.4 million. Content revenue, which includes work such as sponsored material and licensed programming, dropped 20% to $5.8 million.

The weakness ran across all three reported revenue lines. It was not simply one bad advertising quarter offset by growth elsewhere.

BuzzFeed also reported a $4.8 million adjusted EBITDA loss, compared with positive adjusted EBITDA of about $753,000 a year earlier. That is a company-defined, non-GAAP measure that leaves out several expenses; it is separate from the $18.7 million net loss.

July’s cuts, not a new layoff announcement

Chairman and CEO Byron Allen described the period as a transitional quarter. He said the restructuring announced in July reduced the company’s workforce and dedicated contractors by approximately 35%.

Those cuts were already announced, not a new round disclosed with these results. Allen said the company expects them to produce approximately $30 million in annualized savings. That figure is an expectation for a yearly cost reduction, not savings already booked in this quarter.

BuzzFeed deferred its full-year financial outlook while it integrates the Allen Family Digital transaction from May and the July restructuring. That leaves investors without updated full-year financial guidance in this release.

Allen’s next bet is free streaming

For the fourth quarter, Allen said the company plans to take content across digital platforms and connected television as it works toward becoming a global free-streaming platform. The release sets out that ambition, but does not establish that the strategy has turned the business around.

BuzzFeed pointed to a shift in how people reach its own properties: direct visits and internal web and app referrals accounted for 67% of U.S. traffic, up from 65% in the previous quarter. The company said that reduces its dependence on outside platforms and their changing algorithms.

Keeping readers coming back is one part of the challenge. The quarter’s financial results show the other: earning enough from those visits to cover the cost of making the content.

JBizNews Desk | New York

© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.

Dan Margalit and Tara Brendle at a mathematics lecture.

Courtesy of Dan Margalit

Artificial intelligence can accelerate mathematical work without replacing the human ideas and checking behind it, Vanderbilt mathematics chair Dan Margalit argues in an assessment of AI-generated proofs.

In an interview published by Axios on November 4, Margalit described AI as a tireless finisher of work humans have already brought close to a solution, rather than an independent source of every important mathematical insight.

His assessment is an expert’s view of how mathematical work is changing, not a claim that every AI-generated proof is correct or that every reported breakthrough has been independently confirmed.

Human ideas still supply the starting point

Margalit said the problems AI solved were those for which human research had already supplied useful paths. He contrasted those successes with problems where the existing literature offered less direction.

The distinction is between extending known ideas and choosing an original, worthwhile question. Margalit sees a continuing role for mathematicians in framing problems and deciding which lines of inquiry deserve attention.

A proof still needs checking

Margalit also pointed to mistakes found as mathematicians reviewed AI-produced solutions. According to his account to Axios, some papers faced retraction after reviewers identified logic errors.

That makes verification part of the work, not a step that disappears once a system produces a polished answer. A plausible-looking proof can still contain a flawed inference.

He compared the systems’ search through combinations of known theorems with the way computers explore possibilities in chess and Go. The scale of the computation can be impressive without resolving the separate question of mathematical understanding.

A tool for harder questions

Margalit offered software for calculating integrals as a precedent: automation took over work that once filled books, allowing researchers to spend time on harder concepts instead.

Under that view, AI could change what mathematicians spend their days doing while leaving the search for understanding at the center of the discipline.

“Mathematics is the human endeavor of understanding,” he told Axios. His argument is that an answer and an explanation are not interchangeable, and that human judgment remains important even when machines become faster at producing candidate solutions.

Vanderbilt appointed Margalit department chair in 2023. His research focuses on low-dimensional topology and geometric group theory.

JBizNews Desk | Nashville

© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.

A Ukrainian Support Forces soldier installing a Starlink terminal in July 2022.

Photo: Support Forces of Ukraine Command / mil.gov.ua / CC BY 4.0

Ukrainian President Volodymyr Zelensky says he doubts Elon Musk acted alone in last month’s slowdown of the Starshield satellite service used by Ukraine’s military, while acknowledging that he does not know the rules governing the system or which U.S. officials, if any, were involved.

His comments, highlighted by The Hill on Wednesday, came in a PBS News interview published November 2. They are his assessment, not evidence establishing that the U.S. government ordered the restrictions.

“I’m not sure that he could do it personally alone, without any representatives from American government,” Zelensky told PBS, referring to Starshield’s military role. He added that he did not know who on the American side had made or allowed the decision.

For Ukraine’s military, the issue is the reliability of a communications service used in wartime operations. Zelensky said the system was working again; his comments concerned responsibility for the earlier disruption, not an announcement of a new outage.

Suspicion, not a confirmed account

The full PBS transcript records an important qualification. Asked whether Musk could act alone, Zelensky replied: “No, no, he could act alone. I don’t know … I don’t know the rules.” Asked whether he believed Musk had done so, he expressed doubt because Starshield is a government military service.

Zelensky also said he remained grateful for Starlink and had wanted to speak with Musk after the disruption. Starlink is SpaceX’s commercial satellite internet service; Starshield is its military-oriented system supplied under a Pentagon contract.

Administration’s earlier denials

The Hill reported that Defense Secretary Pete Hegseth and Secretary of State Marco Rubio previously said the White House had nothing to do with the slowdown. PBS’s interviewer also noted President Donald Trump’s denial that he had ordered it.

The Hill quoted Rubio’s October explanation that use and speed limits were governed by a private commercial contract. Musk had attributed Starshield restrictions to the U.S. military and president, but also said Ukraine had violated Starlink’s terms by using it for offensive drone strikes, according to The Hill.

Those differing accounts leave responsibility unresolved. Zelensky’s interview did not identify a U.S. official, document or order proving government participation.

JBizNews Desk | Global

© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.

Rows of server racks inside a data center

Photo: Ontariophotos / Wikimedia Commons / CC BY-SA 4.0

The AI data-center boom hit a wall at the ballot box Tuesday, as voters across the country approved bans and restrictions on the facilities and swept in candidates who promised to slow the industry’s buildout.

Voters in 10 states decided 41 local ballot measures on data-center development and sided against the industry in nearly every contest with results available, according to a CNBC analysis of Ballotpedia data; The Washington Post counted at least 38 jurisdictions approving bans or restrictions. The lone exception was Harwood, North Dakota, where voters rejected an added layer of review for industrial projects. In Wells, Maine, more than 80% of voters approved a retroactive 180-day moratorium, and several Ohio cities banned data centers above 25 megawatts outright.

The backlash reached the top of the ticket. In Texas, Democrat Gina Hinojosa won the governor’s race – the state’s first Democratic governor since 1990 – after campaigning on a moratorium on data-center approvals until lawmakers pass “real protections so Texans can have a seat at the table,” the Texas Tribune reported. Texas hosts roughly a quarter of planned US data-center capacity, and Jefferies estimates her proposed moratorium could put 13.3 gigawatts of announced projects at risk. She takes office Jan. 19. Democrats who campaigned on curbing the industry also won governors’ races in Ohio and Georgia, and in Ohio, Democrat Sherrod Brown ousted Republican Sen. Jon Husted after casting him as “the face of data centers.”

Voters told pollsters the issue moved them. Seventy-six percent of Election Day voters opposed data centers in their communities, according to NBC News exit polling cited by The Verge, and 56% called the facilities a crucial or major factor in their vote, a CNBC voter poll found. All told, nearly two-thirds of data-center capacity planned for 2028 could face delays, Axios reported, though most analysts don’t expect Democrats to block construction permanently.

The Hill’s energy newsletter framed the night as a “blue wave” crashing into data centers – the rare issue that moved votes up and down the ballot in red and blue communities alike.

JBizNews Desk | Washington

© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.

Designers sketch and review work at a UX design workshop

Photo: Jan Dittrich (WMDE) / Wikimedia Commons / CC BY 4.0

Figma raised its full-year outlook Wednesday after third-quarter revenue jumped 43%, saying customers are flocking to its AI design tools – but the shares fell about 8% in extended trading as of 4:36 p.m. ET, Reuters reported, after gross margin came in just below expectations.

The San Francisco design software company earned an adjusted 10 cents a share on revenue of $391.4 million in the quarter, topping expectations of 5 cents on $375.8 million. Its net loss narrowed to $81.3 million from $1.10 billion a year earlier, and its 12% adjusted operating margin roughly doubled what analysts expected.

Figma now expects full-year revenue of $1.497 billion to $1.499 billion, up from $1.463 billion to $1.467 billion, and an adjusted operating margin of 11%, up from 9%, with adjusted operating income of $158 million to $163 million. Gross margin of 80% in the quarter fell just short of the 81% analysts expected – the miss that pressured the stock. The company sees fourth-quarter revenue growth of about 33%.

The AI bet. Figma started giving customers more AI credits at the same prices during the quarter, deliberately trading near-term revenue for deeper usage of tools like its Figma Agent design assistant. “You’re potentially trading off near-term revenue, investing in your customers, and then allowing yourself to grow durably over the long term,” finance chief Praveer Melwani told CNBC. More than 80% of Figma’s large paid customers used AI credits weekly through October, and credit utilization has risen 40% since the repricing, Melwani said. Figma also more than doubled the share of work powered by its own design model, which serves customers at lower cost than third-party models, the company said.

The stock remains down about 41% this year as investors have worried AI could undercut design software. Rival Canva said Tuesday it is cutting 22% of its staff after growth slowed.

JBizNews Desk | San Francisco

© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.

Emma Walmsley at a GSK event, in a photo published in 2020.

Photo: Stekzar / CC BY-SA 4.0

HP named former GSK chief Emma Walmsley its next chief executive on Wednesday, with the appointment taking effect January 4, 2027.

The board voted unanimously to appoint Walmsley, who will succeed interim CEO Bruce Broussard. Broussard, who has held the interim role since February 2026, will become board chair on the same date, replacing Chip Bergh.

For HP’s employees, customers and business partners, the announcement sets a date for the end of the interim leadership period while keeping Broussard involved in board oversight.

A move from pharmaceuticals to technology

Walmsley led GSK from April 2017 through December 2025. She has served on Microsoft’s board for seven years and previously spent 17 years at L’Oréal, according to HP’s appointment announcement.

She joined GSK in 2010 and became CEO of its Consumer Healthcare business in 2015 before taking the group CEO role.

HP said the selection followed a global search. Bergh cited Walmsley’s leadership experience across industries and her ability to turn customer insights into business growth.

HP links the appointment to its AI plans

Broussard framed the appointment around opportunities as artificial intelligence moves to the edge, meaning processing closer to users and their devices. That is HP’s strategic view, not an announced product launch or a new financial forecast.

Walmsley said she would focus on the AI era and new sources of growth. HP did not announce specific product changes or a new financial target in the appointment release.

Bergh, who has chaired HP’s board since 2017 and served as a director since 2015, will remain chair until Broussard takes over on January 4.

JBizNews Desk | Palo Alto

© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.

The SpaceX Starbase facility in Boca Chica, Texas

The SpaceX Starbase facility in Boca Chica, Texas

Photo: JennyHPhoto / Wikimedia Commons (CC BY-SA 4.0)

Goldman Sachs and Apollo Global Management are talking to investors about buying portions of a roughly $40 billion financing package for SpaceX’s artificial intelligence chips, Bloomberg News reported Wednesday.

The talks are about selling down pieces of the financing to other investors, not a new chip purchase, and no sale has closed.

The package itself was announced October 26 as the largest GPU financing ever: a master finance lease with an Apollo-managed entity for NVIDIA’s next-generation Vera Rubin chips, with Goldman Sachs as sole structuring agent and anchor commitments from BlackRock, Goldman Sachs Asset Management and KKR. SpaceX said the chips will power Grok and its cloud services, with closing expected before year-end and deliveries in the first half of 2027.

JBizNews Desk | New York

© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.

A MercadoLibre electric delivery van on a street in Chile

A MercadoLibre electric delivery van on a street in Chile

Photo: Igallards7 / Wikimedia Commons (CC BY-SA 4.0)

MercadoLibre grew revenue 45% to $10.75 billion in the third quarter, but profit fell short of expectations as the Latin American e-commerce giant spent heavily on free shipping and new warehouses, and the shares dropped after hours.

Adjusted earnings came in at $8.83 a share, missing the $9.34 analysts had projected, while revenue edged past the $10.69 billion expected. Operating income was $676 million, a 6.3% margin, down from 7.0% in the prior quarter, and net income was $448 million, the company reported Wednesday.

Shares fell 7.8% in after-hours trading, Investing.com reported at 4:14 p.m. ET.

Growth is not the problem. Gross merchandise value rose 38% to $23 billion, payment volume on Mercado Pago jumped 53% to $109 billion, fintech users grew 29% to 93 million and advertising revenue climbed 58%. The credit portfolio swelled 62% to nearly $18 billion. The squeeze came from the cost of that growth: expanded free shipping, fulfillment capacity up almost 20% from the prior quarter, and heavier marketing in Brazil.

The company said it gained market share in its three biggest markets, Brazil, Mexico and Argentina. For the millions of sellers and shoppers who treat MercadoLibre as the Amazon of Latin America, the strategy is familiar: sacrifice profit now to lock in customers for the long run. Investors on Wednesday were not in the mood to wait.

JBizNews Desk | Buenos Aires

© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.

Close-up of an Arm-based processor chip package

Close-up of an Arm-based processor chip package

Photo: Caspertheghost / Wikimedia Commons (public domain)

Arm Holdings beat quarterly profit and revenue expectations but its forecast for the current quarter merely matched them, and the shares slipped after hours.

The British chip designer earned an adjusted 52 cents a share on revenue of $1.44 billion in its fiscal second quarter, up 27% from a year earlier and ahead of the 47 cents and $1.38 billion analysts had projected. On a GAAP basis, earnings fell to 12 cents a share from 22 cents.

Shares fell nearly 4% after hours, Investing.com reported in a 4:38 p.m. ET update.

The guide was the letdown. For the current quarter Arm forecast adjusted earnings of 52 to 60 cents a share on revenue of $1.49 billion to $1.59 billion, essentially in line with the 55 cents and $1.53 billion analysts expected. Investors who had priced in a blowout sold the news.

The engine is licensing. License and other revenue jumped 50% to $773 million while royalties rose 8% to $670 million, the company said in its shareholder letter. Royalties from data-center chips doubled, while smartphone royalties grew more slowly as high memory prices squeeze phone makers. Arm says hyperscalers, the giant cloud operators, now account for more than half its royalty base, a company claim that underscores its push beyond phones.

“AI is driving unprecedented demand for compute across the cloud, devices and autonomous systems, with Arm being the compute platform of choice,” CEO Rene Haas and finance chief Jason Child wrote, calling demand for its new Arm AGI data-center chip “extremely strong.”

The report lands in a rough patch for chip stocks. The Philadelphia Semiconductor Index slumped 15.4% in October and Arm’s U.S.-listed shares tumbled nearly 29% last month as investors question whether the AI spending boom will pay off. Arm’s designs sit inside nearly every smartphone on earth, and it collects a small royalty on each chip sold, so its results are a window into both the phone market and the AI buildout.

JBizNews Desk | Cambridge, England

© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.

A Roll-Out Solar Array being deployed outside the International Space Station

A Roll-Out Solar Array being deployed outside the International Space Station

Photo: NASA

Redwire’s order backlog swelled to a record $619.7 million in the third quarter, up nearly 75% from a year earlier, as the space infrastructure company also doubled its gross margin.

Revenue rose 5.4% to $109 million and gross margin hit a record 32.3%, up from 16.3% a year earlier, the company reported Wednesday. Its net loss narrowed by $18 million to $23.2 million.

Redwire won $188.7 million in new contracts during the quarter, pushing its book-to-bill ratio to 1.73, meaning it is booking work much faster than it can bill it. Among the wins: a $20 million DARPA contract to test a scaled version of its SabreSat spacecraft for a flight in very low-Earth orbit, and a follow-on order from Axiom Space for two of its Roll-Out Solar Array wings, the same technology that powers the International Space Station, for Axiom’s second commercial space station module.

“Redwire continued to see durable demand across our space and defense tech portfolio,” said CEO Peter Cannito, though he cautioned that “revenue timing continues to be uneven.”

The company reaffirmed its full-year 2026 revenue forecast of $450 million to $500 million. It has recorded $323.1 million in revenue through the first nine months.

For anyone watching the commercial space race, the takeaway is that the pick-and-shovel makers, the companies selling the solar wings and spacecraft parts rather than launching rockets, are finding plenty of buyers.

JBizNews Desk | Jacksonville

© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.

Photo: Joe Raedle/Getty Images (archive)

Bright MLS has activated an integration with Ocusell that lets eligible brokers and agents enter or edit a property listing once and send it to Bright and other participating multiple listing services, reducing the need to retype the same information across markets.

The launch, announced Tuesday, is the first major product milestone from the joint venture the companies formed in 2025. It connects listing entry and validation through the Ocusell Network while retaining the rules, data structures and access permissions of each participating MLS.

The live integration should not be confused with a completed nationwide rollout of the broader OneList platform. Bright will share wider rollout timelines and early-access details in future communications.

One entry point, multiple destination systems

Brokers and agents can create and change listings in Ocusell, where information is checked against MLS rules before being pushed to Bright and other destination systems. The workflow is intended for active Bright brokerages managing listings in multiple markets.

That addresses an operating problem for real estate teams that belong to more than one MLS: the same property details, photographs and disclosures may otherwise have to be entered separately, increasing staff time and the chance of inconsistent information.

The integration does not mean every MLS has joined the network or that local rules have disappeared. Participation, subscriber entitlements and each destination MLS’s requirements remain part of the process.

Time savings are vendor claims

Ocusell says its infrastructure can cut listing-creation time by up to 75% and prevent more than 85% of compliance errors. Those are vendor claims, not independently verified performance findings. Actual results will depend on the brokerage’s existing workflow and the participating MLSs.

The company also says it is live across 30 MLSs. That network count does not establish that all of the broader OneList features are available to every Bright subscriber today.

The forthcoming interface is intended to add AI-driven compliance and Fair Housing screening, automated photo tagging and organization, and listing-promotion tools. These are planned capabilities rather than a guarantee that automated screening will catch every problem.

A workflow change for multi-market brokerages

The business case is operational: fewer repeated steps between taking a listing and distributing it to the systems a brokerage uses. It is not a promise that homes will sell faster or that agents will generate more transactions.

For MLS operators, the launch also illustrates a shift toward connected listing-entry platforms that sit across local systems while preserving their separate requirements. The value will depend on how widely those connections are adopted and how reliably updates move between them.

Launch details and the companies’ statements were published Wednesday. The available announcement describes the integration as live, while leaving the broader OneList release schedule for later.

JBizNews Desk | New York

© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.

The Snapchat app open on a smartphone

The Snapchat app open on a smartphone

Photo: Santeri Viinamäki / Wikimedia Commons (CC BY-SA 4.0)

Snap swung to a third-quarter profit as revenue grew 14% to $1.722 billion, though the top line came in about $10 million shy of what analysts expected, and the shares slid after the report.

Net income reached $56 million, against a $104 million loss a year earlier, with earnings of 3 cents a share. Adjusted EBITDA more than doubled to $383 million from $182 million, operating cash flow was $210 million and free cash flow was $164 million, the company reported Wednesday.

Shares fell about 11% after hours in Seeking Alpha’s 4:14 p.m. ET report.

The audience keeps growing, slowly. Snapchat averaged 495 million daily users in the quarter, up 4% from a year earlier, though its most lucrative market, North America, held flat at 92 million. Snap also said it will switch to monthly users as its headline audience measure starting in early 2027, arguing that better reflects the size of its community.

CEO Evan Spiegel cast the quarter as proof the business model is maturing. “AI is becoming a significant advantage for our teams and advertising platform,” he said, adding that Snap is investing in its “long-term vision for augmented reality with the launch of SPECS,” its coming smart-glasses line. Those are his claims; the AI investments have yet to show up as faster revenue growth.

For the half a billion people who open Snapchat every day, mostly to message friends, the quarter means the company behind the app is finally making more money than it spends while it bets on glasses as the next big thing.

JBizNews Desk | Santa Monica

© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.