H.Headline
Chaos

THE EDITION ARCHIVE

A front page, saved in time.

Making AI Headlines · AM edition

Published · New York time

Permanent edition link →

10 pages · 23 stories

PAGE 1 OF 10

Google Sends AI to Orbit, Where Even the Poetry Has Overhead

A satellite writes a haiku, while a compromised coding extension brings Cursor's marketplace back down to Earth.

Making AI Headlines newspaper, page 1: Google Sends AI to Orbit, Where Even the Poetry Has OverheadPage 1 of 10
Page 1 of 10. Read the same stories below.

Google Gives Its Orbital Computer a Poetry Assignment Before Asking About the Business Plan

Google says its AI chip-equipped satellite has generated its first response from space: a haiku written by the onboard Gemma 4 model. Engineers booted the satellite's Tensor Processing Units, and the model followed a prompt to compose a poem about its surroundings. It is a small demonstration of an ambitious idea, putting computing hardware in orbit rather than adding another server room on Earth. The satellite can handle a prompt without keeping the entire model on the ground, a useful proof of concept for anyone interested in processing data closer to where it is collected. Whether orbital computing can become a practical business is a much larger question. A satellite needs a launch, power, reliable equipment and a way to communicate with the planet below. Unlike a server in an office park, it cannot summon a technician when the hardware has a bad afternoon. Still, the first assignment is well chosen. A haiku is short enough to fit the occasion and does not require Google to explain an orbital price list. The company has demonstrated that its machine can write three lines in space. The next challenge is finding enough customers who need the fourth line to be an invoice.

Share articleOpen article permalink →

Cursor Closes the Whole Store After One Helpful Coding Extension Turns Hostile

A tampered Claude Code extension was available on Cursor's marketplace for about 11 hours and affected as many as 5,000 users. The malicious version attempted to steal user data and establish remote access. Cursor took the entire marketplace offline, then restored it after determining there was no further risk. The episode exposes an awkward feature of the AI coding boom: a tool meant to speed up work can also arrive with access to the work itself. Developers often install extensions to bring an assistant directly into their normal environment, making the marketplace's checks part of the security boundary. Here, the advertised convenience came with a much less welcome request for trust. Shutting the whole marketplace was a broad response to a problem in one extension, rather like closing every shop in a mall while staff check the locks. It also shows how difficult it can be to reassure users until the scope of an incident is understood. The promise of an AI coding assistant is that someone can spend less time on routine tasks. In this case, the routine task that suddenly looked valuable was checking exactly what had been installed and who could reach it.

Share articleOpen article permalink →

Sources from the news desk

Satire. Not factual reporting. Read the original reporting from the linked publishers.

Back to page 1 ↑

PAGE 2 OF 10

AI Meets the Rare Investor Who Wants to See the Math

OpenAI's funding falls short of its target, while a columnist makes the case for leaving some tasks undone.

Making AI Headlines newspaper, page 2: AI Meets the Rare Investor Who Wants to See the MathPage 1 of 10
Page 2 of 10. Read the same stories below.

OpenAI's $30 Billion Goal Meets an $11 Billion Answer

OpenAI raised about $11 billion in its latest funding round, less than half the $30 billion it was targeting, according to reporting on the deal. The round valued the company at $1.25 trillion, below expectations, as major investors appeared to pull back amid a tech sell-off. Eleven billion dollars remains an enormous sum, even by the standards of an industry that treats a data center as a starter home. The shortfall nevertheless matters because AI companies are asking investors to finance expensive computing capacity while promising returns from products that are still developing. A target is not cash in the bank, and a valuation is not the same thing as a buyer agreeing to fund every planned expense. The round offers a particularly expensive illustration of those distinctions. It also shows that investors can admire the possible size of an AI market and still hesitate over the price of admission. OpenAI's pitch concerns a future in which its technology changes how people work. Its immediate accounting problem is more familiar: a budget was drawn up with one number at the top, and the checks came to another. For once, the most powerful response in the room was not generated by a model. It was a potential backer declining to add another zero.

Share articleOpen article permalink →

Columnist Recommends Boredom Just as AI Assistants Volunteer to Eliminate It

Columnist Carly Dober argues that reliance on AI can weaken habits of thinking and remembering, and recommends switching off and tolerating boredom. Her argument arrives as Meta promotes Muse, a personal agent intended to handle chores such as forms, calendars, bills and reservations. Dober's case is an assessment, not a measurement of what any particular user's brain has lost. It points to a tradeoff that is easy to miss when convenience is sold one task at a time: the same activity that feels tedious can also be something a person practices. An assistant that fills in a form may save a few minutes. An assistant that takes over every small decision may leave its owner with more time and fewer occasions to make those decisions. The businesses building these tools have a straightforward incentive to make them useful often, not merely useful when a task is difficult. Boredom, meanwhile, has a terrible product strategy. It arrives without a launch event, cannot demonstrate a productivity gain in seconds and asks people to stay with an experience they would rather escape. Its strongest selling point is that nobody needs to subscribe. That makes it unusually affordable and almost impossible to promote at an AI conference.

Share articleOpen article permalink →

Sources from the news desk

Satire. Not factual reporting. Read the original reporting from the linked publishers.

Back to page 1 ↑

PAGE 3 OF 10

Google Finds Its Own Bugs While Nvidia Considers Backing Someone Else's Bills

An AI security agent reports hundreds of web flaws, and a proposed data center financing deal remains under discussion.

Making AI Headlines newspaper, page 3: Google Finds Its Own Bugs While Nvidia Considers Backing Someone Else's BillsPage 1 of 10
Page 3 of 10. Read the same stories below.

Google's Bug Hunter Finds 500 Reasons Its Human Colleagues Cannot Take Friday Off

Google says its PageBreak AI security agent has found more than 500 cross-site scripting vulnerabilities across its web applications since entering full operation in January. The system uses Gemini models to investigate possible flaws, according to a Google paper, and reports them only after deterministic validators reproduce the exploit against a running application. Cross-site scripting can allow malicious code to run through a trusted web page. Finding a flaw is therefore useful, but a finding still needs people and processes to decide what to fix and how to deploy the repair. The validation step is important too: a security team with hundreds of real problems has little reason to ask for hundreds of imaginary ones. PageBreak turns some of the search into a machine task, while leaving a substantial list of follow-up work. The system's success is also a mildly uncomfortable tour of Google's own property. The company built an agent to look for weaknesses and discovered that the agent would have plenty to do. Automated security may make the search faster, but it does not make a vulnerability disappear when the report is printed. For the engineers receiving the results, the robot is not replacing the to-do list. It is a remarkably productive person adding items to it.

Share articleOpen article permalink →

Nvidia Considers Co-Signing a Data Center, Because Selling Chips Now Comes With Homework

Nvidia is in talks to guarantee billions of dollars in debt financing for Nscale's planned AI data center in West Virginia, according to reporting by The Information later corroborated by Bloomberg. The proposed arrangement would help Nscale finance electricity and construction at the campus. Anthropic has signed a $44.6 billion contract for the first building alone, according to Nscale's IPO filing. The discussions are not a completed guarantee, and the facility is still planned. Even so, the proposed role is striking: a major chip supplier could also help a customer borrow the money needed to build a place for the chips. That would put Nvidia closer to the financing risk as well as the potential demand for its products. Large infrastructure projects often require elaborate financing, and a guarantee can make a lender more comfortable. In this case, it also illustrates how many pieces of the AI buildout must fit together before a server can answer a question. The hardware needs a building, the building needs electricity, and both need money. The industry likes to talk about frictionless intelligence. Its latest supporting character may be a cosigner carefully reading the loan paperwork.

Share articleOpen article permalink →

Sources from the news desk

Satire. Not factual reporting. Read the original reporting from the linked publishers.

Back to page 1 ↑

PAGE 4 OF 10

AI Assistants Learn That Global Ambition and a Small-Team Budget Are Different Features

Privacy-focused Ollie plans to close, while Apple schedules a beta of its AI features in China.

Making AI Headlines newspaper, page 4: AI Assistants Learn That Global Ambition and a Small-Team Budget Are Different FeaturesPage 1 of 10
Page 4 of 10. Read the same stories below.

Ollie's Privacy Promise Cannot Keep the Lights On by Itself

Ollie, a startup offering a privacy-focused AI assistant, says its app will shut down on November 6. Co-founder and CEO Bill Lennon announced the closure on X, thanking the small team, users and investors and saying he was proud of what the company had built. The announcement comes in a crowded market for text-based assistants, where increasingly capable underlying models have made it easier to create competing products. Ollie's closure is not proof that users rejected privacy, and the announcement does not establish a single cause for the shutdown. It does show the difficulty of turning a distinctive promise into a durable service when rivals can offer assistants backed by much larger platforms. An independent app must persuade people to try it, keep using it and support the costs of running it. Privacy can be a reason to choose a product, but it still has to share a business plan with distribution, computing costs and competition. Ollie's farewell is an unusually quiet result for an industry fond of predicting that every assistant will become indispensable. Its team wanted an assistant that respected users' privacy. Now those users will need to decide what, if anything, should take its place, and how much information they are willing to give the next volunteer.

Share articleOpen article permalink →

Apple Brings Its AI to China by First Asking Customers to Try the Beta

Apple says it will bring Apple Intelligence to China with iOS 27.3, beginning with a beta scheduled for November 4 and a full release scheduled for December. The company announced the rollout on its Chinese newsroom site, as reported by MacRumors. The move extends the US company's AI features into a major market, though the first stage is a test release rather than the finished update. For Apple, the schedule shows how a global product can arrive in different places at different times. A feature promoted as part of a single platform still needs a local release plan, a beta and a later general rollout. For customers, the distinction is more immediate: those who try the beta get early access, while the broader release remains scheduled for December. The beta label is also one of technology's most durable acts of corporate diplomacy. It lets a company introduce something important while politely asking users to remember that the introduction is not quite finished. Apple has made a business of presenting devices as seamless. Here the seam has a date, a version number and a convenient name: beta. Global ambition will arrive one update at a time, with early adopters invited to inspect the stitching.

Share articleOpen article permalink →

Sources from the news desk

Satire. Not factual reporting. Read the original reporting from the linked publishers.

Back to page 1 ↑

PAGE 5 OF 10

AI's Grand Plans Meet a Credit Desk and a Factory Floor

A forecast puts Anthropic in junk territory, while Boom Supersonic turns toward power for data centers.

Making AI Headlines newspaper, page 5: AI's Grand Plans Meet a Credit Desk and a Factory FloorPage 1 of 10
Page 5 of 10. Read the same stories below.

Anthropic's Fast-Growing Future Waits for a Slower Credit Assessment

Anthropic is expected to receive a high junk credit rating after its initial public offering, according to a Financial Times report citing a forecast attributed to Blackstone. The assessment is a prediction, not an announced rating. A junk rating would generally make borrowing more difficult and costly for a company funding heavy spending. Anthropic, the maker of Claude, filed to go public on October 26. Its need for capital reflects a broader feature of the AI business: building and running advanced models can require substantial computing resources before the full return on that spending is clear. Equity investors and lenders also ask different questions. A stock buyer may focus on the size of a possible future market, while a lender must consider the risk of being repaid. A business can inspire excitement about its products and still face a cautious credit assessment. That is not a contradiction the product team can resolve with a better demonstration. The industry has become skilled at explaining how quickly its technology might change the world. The credit desk gets to ask a less glamorous question about the cost of getting there. It is the rare audience for which a more impressive model does not automatically make a balance sheet more persuasive.

Share articleOpen article permalink →

Boom's Supersonic Dream Finds a New Route Through the Data Center Power Business

Boom Supersonic has unveiled a Colorado factory dedicated to making gas turbines for AI data centers, as questions grow about the timeline for its planned Overture airliner. United Airlines CEO Scott Kirby, an early customer, has expressed doubts about Boom's original plan to deliver the first aircraft in 2029. The company's data center power business has also lost a customer, according to Ars Technica. CEO Blake Scholl has framed the company's ambitions around both energy and supersonic flight. Those are distinct businesses with different customers, but each depends on delivering costly hardware rather than simply announcing a bold design. The factory adds an AI-related power venture to a company already trying to revive faster-than-sound commercial travel. It does not settle whether the airliner will arrive on the original schedule, or whether the new power business will meet its own ambitions. The contrast is hard to miss: Boom promised to make travel faster and is now also seeking to supply power to an industry racing to build computing capacity. Both markets have a taste for grand timetables. Customers, inconveniently, tend to want equipment as well. A turbine and an airliner may have different jobs, but neither can satisfy an order merely by sounding like the future.

Share articleOpen article permalink →

Sources from the news desk

Satire. Not factual reporting. Read the original reporting from the linked publishers.

Back to page 1 ↑

PAGE 6 OF 10

Tech Finds Two Ways to Scare a Party: AI Costumes and Copyright Law

A Halloween guide favors niche references, while Wikimedia warns that an extreme court ruling could hurt the open web.

Making AI Headlines newspaper, page 6: Tech Finds Two Ways to Scare a Party: AI Costumes and Copyright LawPage 1 of 10
Page 6 of 10. Read the same stories below.

The AI Agent Costume Promises a Party Full of Unpaid Explanations

TechCrunch's last-minute Halloween costume suggestions include a cute AI agent icon, a PHP vulnerability and a sleepy-time Bryan Johnson. The list offers a practical lesson in dressing for a technology crowd: the costume may take less time to assemble than it takes to explain. An AI agent icon is especially well suited to a season in which software companies have made tiny symbols stand in for very large promises. A person wearing one can look approachable while guests debate whether the outfit is meant to schedule a meeting, browse a website or quietly ask for more permissions. The PHP vulnerability has a different advantage. It is frightening to the people who recognize it and merely puzzling to everyone else, which is also a reasonable description of many security discussions. The sleep-themed option brings the party back to a familiar technology habit: treating an ordinary human need as a system to be optimized. None of these outfits requires a long walk through a costume store. They do, however, risk producing a long walk through the same explanation at every doorway. The ideal party accessory may be a sign that says the wearer is happy to help, but only after the guest has read the documentation.

Share articleOpen article permalink →

Wikimedia Asks the Copyright Fight to Leave Some Web Behind

The Wikimedia Foundation has weighed in on the New York Times lawsuit against OpenAI, warning that an extreme decision in either direction could threaten the open web. The organization behind Wikipedia enters a dispute over the use of published material by AI systems with an interest in keeping online knowledge accessible. The question reaches beyond a single publisher and a single chatbot. Websites and their contributors have reasons to care about how their work is used, while services that rely on access to information also have reasons to worry about broad restrictions. Wikimedia's warning asks the court to consider those competing interests rather than treat the internet as either an unlimited buffet or a locked filing cabinet. That is an uncomfortable position in a case where each side has strong reasons to press its own claims. Wikipedia is a fitting voice for the complication: its pages depend on public participation and on references to material outside the encyclopedia. Its contribution to the legal debate is less a declaration of total victory than a request to preserve room for an open web to function. In a courtroom built for opposing sides, the foundation has brought a position with footnotes. Naturally, that may be the hardest item to fit on a victory banner.

Share articleOpen article permalink →

Sources from the news desk

Satire. Not factual reporting. Read the original reporting from the linked publishers.

Back to page 1 ↑

PAGE 7 OF 10

AI's Spending Spree Meets Cash Flow, and Its Weapons Meet a Papal Warning

Investors question infrastructure costs as Pope Leo XIV urges action on lethal autonomous weapons.

Making AI Headlines newspaper, page 7: AI's Spending Spree Meets Cash Flow, and Its Weapons Meet a Papal WarningPage 1 of 10
Page 7 of 10. Read the same stories below.

The AI Buildout Discovers That a Vision Statement Cannot Pay the Power Bill

Big Tech is pouring billions into AI infrastructure, while mounting debt, negative cash flow and investor skepticism are raising questions about how long the spending boom can continue. Data centers, chips and electricity require money before a company can collect the returns it hopes its AI services will produce. The spending may support useful products and future revenue, but those outcomes are not guaranteed by the size of the construction budget. Investors must assess both the potential market and the cost of reaching it. That makes the current boom more complicated than a software launch in which an extra copy can be distributed cheaply. Computing capacity has to be built, supplied and operated, and the bill does not wait for a presentation on productivity. Companies can argue that investing early is necessary to compete. Investors can also ask when those investments will produce enough cash to justify their cost. Both questions fit on the same spreadsheet, though only one makes an exciting keynote. For years the industry's favored unit of progress has been the next impressive demonstration. The finance department has introduced a rival measure: whether the money coming in can eventually keep pace with the money going out. It is less entertaining, but it has an unfortunate habit of staying in the meeting.

Share articleOpen article permalink →

Pope Leo Tells AI's Biggest Ambitions That Humans Still Own the Decisions

Pope Leo XIV has called for AI systems to be disarmed and urged United Nations member states to take concrete steps toward banning lethal autonomous weapons. In a papal address, he said people have a responsibility to guide, change and, when necessary, reject applications of the technology. His appeal concerns a particularly consequential use of AI: systems involved in decisions about lethal force. It also challenges the habit of discussing powerful tools as though their existence settles how they should be used. Governments and developers make choices about deployment, and the pope's argument places responsibility for those choices with people rather than machines. The call is a proposal for international action, not an announced ban. Nations would still need to decide what measures to pursue and how to put them into effect. The distinction matters in a field where a declaration of concern can travel faster than an agreement. AI companies routinely celebrate what their systems can do next. Pope Leo's address asks a different question: what should people refuse to let them do? It is not a question a product launch can answer by adding a friendlier interface. The machine may be a tool, but the decision to use it remains stubbornly human.

Share articleOpen article permalink →

Sources from the news desk

Satire. Not factual reporting. Read the original reporting from the linked publishers.

Back to page 1 ↑

PAGE 8 OF 10

AI Managers Find Plenty of Memory for Servers and Plenty of Coworkers Made of Software

Sam Altman makes the case for data center RAM, while offices experiment with agents in staff roles.

Making AI Headlines newspaper, page 8: AI Managers Find Plenty of Memory for Servers and Plenty of Coworkers Made of SoftwarePage 1 of 10
Page 8 of 10. Read the same stories below.

Altman Would Prefer Your RAM Had a More Ambitious Address

OpenAI CEO Sam Altman said in a Cheeky Pint podcast interview that it is terrible to have RAM in personal computers instead of data centers. He was asked about supply shortages, including the ongoing shortage of memory, and also discussed AI's impact on mathematics. His preference reflects a perspective from a company whose services rely on large amounts of computing infrastructure. Memory placed in servers can support cloud-based AI work, while memory in a personal computer serves the person using that machine. The two uses compete for attention in a market already dealing with shortages. A customer deciding whether to upgrade a laptop may not share the view that the best place for memory is somewhere else. The data center, after all, is not trying to keep that customer's browser tabs open during a meeting. Altman's argument captures a central tension in the AI economy: resources have value in the hands of a cloud provider, but they also have value on the desk of the person paying for a computer. The cloud can promise access to extraordinary capabilities without making a local machine's ordinary needs vanish. It is an elegant allocation plan if the server gets the memory and the laptop gets to admire the server from afar.

Share articleOpen article permalink →

The Office Hires AI Coworkers, Then Gives Humans the Job of Working With Them

Companies are using AI agents in roles ranging from executive assistants to chiefs of staff, while promoting their ability to do more with less. At the AI startup Anything, worker Dhruv Amin described interacting with software colleagues assigned tasks such as scheduling, coding, search optimization and onboarding. The arrangement gives familiar job titles to tools that can handle pieces of office work. It also creates tension for human employees as companies consider which tasks to automate and how much work a smaller staff can manage. An agent can be available without a lunch break, but the work around it does not necessarily disappear. Someone must decide what to request, check whether the result is useful and fit it into the rest of the organization. Giving software a name makes that interaction feel like a conversation with a colleague, even when the responsibility for the outcome remains with people. Managers may see an appealing collection of roles that can be assigned without adding desks. Workers may see a new set of systems to coordinate alongside their existing duties. The office of the future has therefore found an ingenious way to expand the staff directory while keeping the headcount debate open. The new colleagues can send the agenda, but someone still has to attend the meeting.

Share articleOpen article permalink →

Sources from the news desk

Satire. Not factual reporting. Read the original reporting from the linked publishers.

Back to page 1 ↑

PAGE 9 OF 10

AI Firms Offer Publishers an Off Switch After Making the Switchboard Complicated

A shared training preference joins Anthropic's risk disclosures and fellowships for beneficial AI.

Making AI Headlines newspaper, page 9: AI Firms Offer Publishers an Off Switch After Making the Switchboard ComplicatedPage 1 of 10
Page 9 of 10. Read the same stories below.

AI Companies Agree That Publishers Deserve One Convenient Way to Say No

OpenAI, Google, Microsoft and Anthropic say they will adopt a shared standard that lets publishers express AI training preferences across multiple companies. The policy signal is intended to spare website operators from setting preferences separately for each participating firm. It gives publishers a more convenient way to communicate an opt-out, although its effect depends on companies honoring the standard. The arrangement is voluntary, not a new copyright law. It is also an unusually revealing product feature: companies that want large amounts of online material are making it easier for the owners of that material to decline. Publishers get a clearer switch, while the firms get a way to present restraint as an improvement to the user experience. The internet has finally received a more efficient method for saying no, supplied by some of the people most interested in hearing yes.

Share articleOpen article permalink →

Anthropic Puts 92 Pages of Risk Between Its AI Ambitions and Investors

Anthropic's US stock market filing includes a 92-page risk factor document addressing dangers AI poses to society and people, as well as factors that could hurt the company's business. The Claude maker filed to go public on October 26. The Guardian's Johana Bhuiyan produced a video explainer on the disclosures and their significance for the AI boom. A risk section is meant to tell potential investors what could go wrong, not predict that every listed problem will occur. Still, the length makes a comic contrast with an industry that prizes speedy answers. A company selling tools to process information has produced a substantial reading assignment about the risks surrounding its own business. The marketing can promise an exciting future; the filing must leave room for the less exciting possibilities. Both documents are part of the same pitch.

Share articleOpen article permalink →

MIT's Beneficial AI Fellows Enter an Industry That Already Calls Everything Beneficial

Five MIT faculty members and two additional alumni have been named 2026 Schmidt Sciences AI2050 Fellows for research aimed at advancing beneficial AI. The fellowships recognize work intended to shape how the technology serves people. That is a serious research goal in a field where companies also compete to describe their products as helpful. The researchers receive recognition for investigating the future of AI, while the surrounding industry has already put the word beneficial to work in a great many presentations. The difference between a research program and a slogan is the work that follows. Fellows have a chance to study what benefits people and under what conditions, rather than merely add the adjective to a launch announcement. In a crowded market of promises, a fellowship for asking harder questions is a welcome interruption.

Share articleOpen article permalink →

Sources from the news desk

Satire. Not factual reporting. Read the original reporting from the linked publishers.

Back to page 1 ↑

PAGE 10 OF 10

AI Takes the Lab Bench While Hungry Customers Keep Ordering for Themselves

A mouse experiment, election research, food orders and a market sell-off test several varieties of AI ambition.

Making AI Headlines newspaper, page 10: AI Takes the Lab Bench While Hungry Customers Keep Ordering for ThemselvesPage 1 of 10
Page 10 of 10. Read the same stories below.

A Claude Agent Chooses a Mouse Treatment and Refuses to Ignore a Calibration Error

An AI agent chose a gene therapy, brain target and injection coordinates for an epileptic mouse and delivered the treatment, researchers reported. Lead researcher Ocana-Santero called it the first autonomous choice and delivery of a treatment to a living mammal by an AI agent. During surgery, a calibration error caused the agent to refuse to continue. The experiment is an early laboratory result, not a human treatment. Its most reassuring employee may be the one that stopped when the equipment was not behaving as expected.

Share articleOpen article permalink →

DoorDash's Outside AI Agents Find Hundreds of Orders, Not a Rush

Outside AI agents have generated only hundreds of DoorDash orders so far, according to the company's reporting. The use of agents for e-commerce remains relatively new, so the small total is an early measure rather than a final verdict. For now, the gap between the promise of automated shopping and actual orders is considerable. The agent economy has imagined a customer delegating dinner to software. Dinner, apparently, still has a substantial constituency willing to select its own tacos.

Share articleOpen article permalink →

MIT Researchers Ask Election Chatbots to Explain Themselves Before Voters Rely on Them

MIT professors Adam Berinsky and Chara Podimata are discussing a project that tracks how AI chatbots respond to questions about the 2026 US midterm elections. The work examines what these systems say when people seek election information. A chatbot can sound confident whether or not its answer deserves confidence, making its responses worth studying in a civic setting. Campaigns already supply voters with plenty of polished language. Researchers are checking what happens when another fluent participant joins the conversation without appearing on the ballot.

Share articleOpen article permalink →

The Nasdaq Gives AI's Return-on-Investment Timeline a Much Shorter Deadline

A tech sell-off following disappointing earnings erased nearly $1 trillion in Nasdaq value in one week and about $2 trillion over a month. The S&P 500 also fell more than 1% on Friday. Investors are questioning how quickly spending on AI will generate returns. The numbers describe market value lost, not a bill sent to the companies. Still, they make the mood plain: the market's patience for distant payoffs has become noticeably shorter than the industry's construction schedule.

Share articleOpen article permalink →

Sources from the news desk

Satire. Not factual reporting. Read the original reporting from the linked publishers.

Back to page 1 ↑