Beyond the announcement.

Nvidia is reportedly considering guaranteeing billions in Nscale debt to help finance the construction and electricity infrastructure for an Anthropic data center in West Virginia, according to The Information, in a report later corroborated by Bloomberg. Under the proposed structure, known as a residual value guarantee, Nvidia would temporarily take over the facility if Nscale defaulted, while helping find a buyer and a new tenant. Nscale said in its IPO filing that Anthropic signed a $44.6 billion contract for the first building alone, a 460-megawatt facility; the companies are also discussing a second deal worth roughly $45 billion.

The potential guarantee would deepen Nvidia’s financial ties to Nscale as investors grow concerned about AI industry spending. Nscale had about 25,000 advanced AI chips running in August, compared with nearly 500,000 it plans to deploy, according to The Information. Nvidia also committed $1 billion in convertible financing to Nscale in September, the publication reported.

Sources: TechCrunch

Read the original story ↗

Ollie will shut down its AI assistant on November 6, co-founder and CEO Bill Lennon announced. Focused on privacy and family organization, the service connected users’ email and calendars and offered tools for appointment booking, meal planning and bill payments. Customers will receive instructions to export their data and request refunds in the coming days.

The company had discussed a $7.5 million seed round led by Khosla Ventures with TechCrunch. Lennon said competition from services such as Muse and Instinct made it harder to stand out, grow and raise the capital needed to succeed. Ollie was also one of the first mainstream applications to achieve SOC 2 certification, which is granted after an independent audit.

Sources: TechCrunch

Read the original story ↗

Anthropic may receive a high “junk” credit rating after its initial public offering, according to the Financial Times. Blackstone, an anchor investor in a recent $63 billion deal to finance Anthropic’s chips, reportedly told investors that it now expects a rating in the BB range — the highest tier of speculative grade. The firm had previously expected an investment-grade rating for the Claude maker before 2027.

Blackstone’s forecast is only an estimate. TechCrunch says Anthropic and Blackstone declined to comment. A high junk rating could make it harder and more expensive for Anthropic to raise money for its heavy spending, including computing-capacity deals signed over the past 12 months. Anthropic officially filed to go public earlier in the week, according to TechCrunch.

Sources: TechCrunch

Read the original story ↗

United Airlines CEO Scott Kirby told the Office Hours: Business Edition podcast that 2029 is Boom’s official timeline for its first Overture, but it “is not going to happen” that year and will take longer. He said the company still faces many hurdles to develop the aircraft and obtain certification from the Federal Aviation Administration (FAA); in March, he had put the odds of commercial service at 50-50.

Boom, which completed its Overture factory in North Carolina in 2024, unveiled a Colorado facility to produce gas turbines for AI data centers. Its Sprint Core turbine core is intended to power both those turbines and the supersonic airliner. Ars Technica reports that the aircraft factory has remained mostly dormant after receiving more than $50 million in state taxpayer funding.

Sources: Ars Technica

Read the original story ↗

According to TechCrunch, Alphabet, Amazon, Meta and Microsoft plan to spend more than $740 billion on capital expenditures this year, much of it tied to AI data centers and chips. Investor reaction has worsened as debt rises, free cash flow turns negative and returns remain uncertain: the Nasdaq fell 2.9% on one Friday, while shares of Nvidia, Broadcom and Qualcomm declined.

Alphabet reported negative free cash flow of $9.1 billion in the third quarter and raised capital expenditures to $51.4 billion; Amazon reported negative free cash flow of $42.4 billion over the past 12 months, and Meta reported $2.1 billion in negative free cash flow in the quarter. Microsoft was an exception, with $18.6 billion in positive free cash flow. According to S&P Global Market Intelligence, the five companies collectively raised $326 billion in debt in 2026. The environment could also complicate Anthropic’s IPO, as the company’s growth depends on securing substantial computing capacity.

Sources: TechCrunch

Read the original story ↗

In an interview on the Cheeky Pint podcast, OpenAI CEO Sam Altman said it is “terrible” to keep large amounts of the world’s limited DRAM supply in personal computers that are used only a small portion of the time. He challenged the idea that open-source models and local inference would eliminate the need for data-center infrastructure, saying the cloud allows memory utilization to be maximized in a supply-constrained environment.

The Verge reports that Samsung and SK Hynix have long-term deals with cloud service operators, while projects such as OpenAI’s Stargate have reportedly reserved a significant share of the world’s RAM output for the next few years. The outlet links the billions being spent on those chips to the current higher prices of laptops, phones and other devices.

Sources: The Verge

Read the original story ↗

US tech stocks fell for a second straight day amid concerns about the hundreds of billions of dollars being spent on artificial intelligence. The BBC reported that Intel dropped 28% on Friday, Alphabet fell 1.4% after losing 12% the previous day, and the Nasdaq ended October with its worst monthly decline since March 2025. Reddit fell 29% after reporting weaker-than-expected daily-user figures and warning that Google’s AI-driven search was hurting referral traffic.

Meta, Alphabet, Oracle, Amazon and Microsoft have committed to spending $800bn on AI in 2026. Alphabet expects to invest up to $205bn this year, while Amazon has committed $220bn, including spending on data centres, network equipment and servers. Analysts cited by the BBC raised questions about the returns on those investments and the risk of an AI bubble.

Sources: BBC

Read the original story ↗

OpenAI, Google, Microsoft and Anthropic say they will support a common standard that lets websites use their robots.txt files to state whether they allow their content to be used for AI training. The proposal, called a “policy token,” was put forward by Kevin Kelley, Anthropic’s technical lead for standards, through the Internet Engineering Task Force (IETF). The signal would let publishers apply the same preference across multiple companies instead of configuring each one separately.

The Verge reports that the companies’ AI crawlers will stop accessing sites displaying the signal. Anthropic has already adopted it; OpenAI says it will implement the signal in the coming months, while Google and Microsoft plan to support it next year. Microsoft will also offer separate controls for AI training and grounding, but says websites will still appear in Bing search results.

Sources: The Verge, OpenAI

Read the original story ↗

According to the Financial Times, Blackstone now expects Anthropic to receive a credit rating in the high speculative-grade range, likely BB, after its IPO. The investment firm had previously expected the company to reach investment grade. The change is Blackstone’s forecast, not a downgrade by a rating agency; Anthropic received a BB+ rating from KBRA in June, according to the British newspaper. Ratings below investment grade can make financing more expensive and restrict some institutional investors, according to explanations from the U.S. Securities and Exchange Commission (SEC) and S&P Global.

Anthropic filed its prospectus with the SEC on October 26 to list shares on Nasdaq under the ANT symbol, but had not yet set the number of shares or the price. In a deal announced with Bank of America, Citi and Morgan Stanley, Blackstone is participating in a $63 billion financing package for Broadcom equipment that will be leased to Anthropic.

Sources: Xataka

Read the original story ↗

DoorDash said during its third-quarter earnings call that it had received only “several hundred” orders from third-party AI agents such as Muse, Instinct and Fo. The figure contrasts with the company’s average of 3 million daily deliveries reported in its 2025 report. CEO Tony Xu said DoorDash wants to learn more about this use case and maintain a healthy relationship with agent makers, although customers still predominantly order through the app. He also pointed to a possible opportunity to advertise within these agents.

DoorDash has its own tools, including Ask DoorDash, which accepts text prompts and photos. The company says baskets created with the feature are built five times faster, have nearly 50% higher value and contain about 60% more unique items than the same consumers’ traditional orders. Nearly half of those orders go to local restaurants the customer had never tried before, DoorDash says.

Sources: Ars Technica, TechCrunch

Read the original story ↗

Fujifilm announced an investment of about 10 billion yen (approximately $63 million) to increase its magnetic tape production capacity fivefold compared with fiscal 2025 levels. The project, planned for fiscal years 2026 through 2028 at the Odawara site of the Kanagawa Factory, will fund manufacturing expansion and additional personnel, primarily to produce FUJIFILM LTO Ultrium10. The company says generative AI is causing “exponential” growth in global data needs and driving demand for archiving.

According to Fujifilm, major cloud service and data center operators have expanded their use of magnetic tape because of its relatively low cost and the ability to disconnect cartridges and store them elsewhere, creating a physical barrier against threats such as ransomware. The company says the investment will be its largest to date in dedicated LTO tape facilities.

Sources: Ars Technica

Read the original story ↗

Google’s Project Suncatcher prototype is online and communicating with Earth after entering orbit. Google says its TPUs were booted successfully, allowing Google DeepMind’s Gemma 4 model to respond to a prompt to write a haiku about being in space; the company also tested limericks and telemetry prompts. The satellite is operating nominally, a spokesperson told TechCrunch.

The spacecraft carries four TPUs but can operate only one at a time for about 20 minutes because of its power demands. The test campaign will assess the chips under thermal stress and radiation, which can alter values stored in memory. If the results are positive, Google says it plans to launch two satellites with more advanced chips in 2027 and demonstrate optical links between them.

Sources: TechCrunch, The Verge

Read the original story ↗

The US Department of Justice (DOJ) says it carried out the first broad federal crackdown on “ghost students”: fraudsters who use stolen personal data to apply to universities and obtain student loans. Five people have been arrested, and the DOJ says it seized more than $750,000 in cryptocurrency linked to a network associated with Kenyan nationals. Officials said AI was used to fill out applications, complete coursework and simulate students’ presence in virtual classes.

The Department of Education identified more than 830,000 potentially fraudulent federal student-aid accounts; if all are ghost students, losses would approach $130 million, the DOJ said. Officials linked more than $32 million in actual losses to the five subjects, but said they believe the total is substantially higher. Americans whose identities were used may discover the fraud only years later, when they are pursued for loan repayment and see their credit damaged.

Sources: Ars Technica

Read the original story ↗

According to The Guardian, cited by Tecnoblog, SpaceXSI may be sending real-estate agents to buy homes near the power plant supplying its data centers in Memphis. The facility, in Southaven, uses 69 temporary gas-powered turbines that residents say produce constant noise, sometimes resembling an airplane, as well as pollution. The company may be offering about US$180,000 per home—close to the local average—and making the sale conditional on residents agreeing not to sue over the disruption.

An agent reportedly said the plan involves 418 homes; 18 may already have been negotiated, while five others were bought by MZX, SpaceX’s subsidiary operating the plant. The properties would house employees. The Guardian also reports that a class-action lawsuit is underway and that SpaceXSI has not commented on the alleged acquisition plan. The company had previously announced sound barriers and power adjustments, as well as an agreement to make the plant permanent.

Sources: Tecnoblog

Read the original story ↗

Tyler Shultz, who helped expose Theranos’ inaccurate blood-testing devices, told The San Francisco Standard, in an interview reported by TechCrunch, that the difficulty of conducting due diligence on artificial intelligence products could make deception easier. He said it is hard to distinguish something genuinely new from a ChatGPT-based product and argued that investors should stop handing over large amounts of money after superficial meetings with founders.

The warning comes as attention grows around “You Can See Everything,” a documentary about Elizabeth Holmes, and the parallels Shultz sees between Theranos and the current AI boom. Theranos reached a valuation of about $10 billion even though its technology did not work as promised. The film also shows Holmes and Billy Evans working on Haemanthus, a diagnostics company they call “Theranos 2.0” that, according to TechCrunch, reportedly raised about $20 million.

Sources: TechCrunch

Read the original story ↗

Samsung reported its most profitable quarter, with 107.4 trillion won—about $80 billion—in operating profit. The company said nearly all of the result came from its chip business: the unit posted about $79.4 billion in operating profit, while memory sales reached roughly $104.9 billion, five times higher than a year earlier.

Higher memory costs, driven by demand from AI servers, hurt the device division. Mobile and network operations lost about $450 million, while TVs and appliances lost another $150 million; a year earlier, the combined businesses had earned about $2.6 billion. Samsung says AI demand will continue to exceed supply in 2027 and that next year’s HBM memory is already sold out. For Galaxy, its 2027 goal is to minimize the decline in profit, while the company expects shortages through 2028, compared with SK Hynix’s outlook for shortages through 2030.

Sources: Xataka

Read the original story ↗

Runware told TechCrunch that it plans to manufacture its Sonic Inference Pods on an assembly line, like cars. The company says it will gradually increase production starting next year and reach 30 megawatts of data center capacity per day by 2028, nearly 1 gigawatt per month. Runware’s second factory in Romania is expected to be completed within months and begin operating early next year.

According to Flaviu Radulescu, Runware’s co-founder and CEO, the pods could be brought online within 48 hours after installation and setup, while manufacturing time is expected to fall from three weeks to one day. Customers could provide their own site and GPUs or rent computing from the company. Runware, which has raised more than $60 million, says it will initially fund the expansion with the $50 million Series A raised last December; TechCrunch notes that the plan still faces challenges involving staffing, inventory, logistics and capital.

Sources: TechCrunch

Read the original story ↗

A South Korean government official told AFP that the country plans to open criminal cases against students who use smart glasses to cheat on the national university entrance exam, known as the Suneung. Authorities plan to extend the period during which violators are barred from retaking the test from one year to three, and want the changes to apply to exams next year.

The measure was announced after three students were caught using the devices in exams organized by universities in Seoul. The glasses can take photos and videos and communicate with artificial intelligence assistants, raising concerns about cheating in exams and recording people without their consent.

Sources: g1

Read the original story ↗

Data center expansion has become a major issue in Maine’s Senate race, amid residents’ concerns about utility costs, water use, and climate impacts. A UMass Lowell and YouGov poll found that 76 percent of respondents opposed building a data center in their community, while 58 percent were strongly opposed.

The state Legislature passed a moratorium, but Gov. Janet Mills vetoed it; towns have also adopted temporary suspensions or put expansion to a vote. According to The Verge, the race is projected to draw $81 million in Democratic spending and $114 million in Republican spending. Its outcome could help determine control of the Senate and influence federal policy on data centers and artificial intelligence.

Sources: The Verge

Read the original story ↗

In an amicus brief, the Wikimedia Foundation took neither side in copyright lawsuits involving publishers, OpenAI, and Microsoft. The organization argues that collecting copyrighted works to pretrain large language models (LLMs) may qualify as fair use when the result is new and different from the original sources, but that reproducing material in response to users should require a license when it substitutes for access to the original works.

Wikimedia says a nuance-free ruling could let copyright holders block transformative uses or give AI companies a green light to ignore licensing terms. The foundation also says these systems can help Wikipedia editors fight vandalism, copyright violations, and doxxing, even as services such as ChatGPT draw visitors away and may fail to follow Creative Commons license terms. The organization says an improper ruling could threaten the sustainability of free knowledge and the development of open-source technologies.

Sources: The Verge

Read the original story ↗