The MacBook Pro PROLED XDR is Apple’s most expensive laptop yet, and early reviews suggest the thing most people will talk about, touch, is the least important reason to buy it. Apple added a touchscreen to the Mac, but the reviewers we looked at spend far more time on the OLED panel and a redesign that makes the machine noticeably lighter.

A lighter MacBook Pro, with a catch at the checkout

9to5Mac’s Chance Miller frames the weight loss as the quiet headline. The 14-inch model drops to 3.0 pounds, and the 16-inch falls from 4.7 pounds to 4.0 pounds. Miller notes that both now weigh roughly what the 2016 MacBook Pros did, despite larger displays and far more powerful hardware. He also says they avoid the butterfly keyboard and thermal problems that defined that era. Both are slightly thinner too: the 14-inch measures 0.59 inches (down from 0.61) and the 16-inch 0.63 inches (down from 0.66).

The sources disagree slightly on the 14-inch baseline. 9to5Mac says the previous model weighed 3.5 pounds, while Engadget compares it with 3.7 pounds for the standard MacBook Pro. Either way, the 14-inch is clearly lighter, and Engadget says it feels closer to a MacBook Air than a Pro.

The weight savings come mainly from swapping the mini-LED display for OLED, which needs no separate backlight. Apple also uses custom on-cell touch technology that builds touch sensing into the panel instead of stacking extra layers. Miller adds that many smaller engineering tweaks contributed. He tested the 16-inch model and says it “feels less like a tank.” It has him reconsidering his usual preference for the smaller size.

Then there is the cost. Engadget’s Devindra Hardawar says the 14-inch MacBook Pro PROLED XDR starts at $2,899, which is $900 more than the standard 14-inch MacBook Pro. The 16-inch starts at $3,399, or $400 more than its standard counterpart. He calls the pricing odd at a time when component costs for RAM and storage are already rising. His rating is 8.8 out of 10, and “so damn expensive” tops his list of cons.

The OLED display is the real selling point

Apple’s clunky full name, “MacBook Pro with PROLED XDR Display,” signals what it considers the big differentiator. Engadget lists the panels at up to 2,000 nits of peak HDR brightness and 1,000 nits sustained full-screen, which it calls the brightest OLED screens it has seen in laptops. It also cites a 10,000,000-to-1 contrast ratio. The screens are slightly larger than before, at 14.2 and 16.3 inches, and thinner bezels add to the immersive feel.

Miller is blunter, calling the display “stunning” and saying its color saturation, contrast and deep blacks beat any other laptop he has seen, including high-end Windows machines. Tom’s Hardware titled its review “The screen that every Mac deserves,” which points the same way. We could not access that review’s full text, so we cannot say more about its findings.

Hardawar offers a more measured view. After hundreds of hours on Apple’s mini-LED MacBook Pros, he found the difference hard to spot in everyday use. Dark, high-contrast video was another matter. He calls the closing fireworks scenes of F1: The Movie, set against a black night sky, “revelatory.”

Touch works, but it is not the point

Engadget opens with the line that “the least interesting thing about Apple’s new touchscreen MacBook Pro is the touchscreen.” Hardawar says Apple took a very basic approach. You can swipe to scroll or manipulate 3D objects, and touches register as mouse clicks throughout macOS. Creative software makers such as Adobe and Blackmagic plan to support it more fully. There is no Apple Pencil support, so anyone who wants to draw still needs an iPad. Engadget’s verdict is that touch “doesn’t add much.”

The 9to5Mac headline, “A lot more than touch,” makes a similar point. The notable changes lie elsewhere, in the lighter chassis, slimmer bezels and rounded display corners. The glowing Apple logo is back as well. Without a backlight to borrow from, Apple tied the logo’s brightness to the screen’s and to ambient light. Miller calls it unnecessary but is thrilled to see it return.

The Dynamic Island, borrowed from the iPhone, replaces the notch. It houses the webcam and serves as a home for Spotlight and Live Activities. Engadget finds it slightly less noticeable than the old notch but not a major visual change. Other listed strengths include better speakers, fast performance and plenty of ports.

What about the “favorite MacBook ever” verdict?

TechRadar’s piece declares the MacBook Pro PROLED XDR 14-inch its writer’s favorite MacBook ever, and says it should leave many MacBook Air users feeling jealous. The text we could access contained only site navigation, so we cannot report the reasoning behind that claim. The headline does fit the other reviews, though, since the lighter 14-inch model is the one Engadget says feels closest to an Air.

Taken together, the reviews describe a MacBook Pro that is lighter and has a better screen, at a much higher price. For those who want the sleekest MacBook Pro available, the MacBook Pro PROLED XDR looks like a strong choice. Engadget’s Hardawar hopes Apple keeps the non-touch models around, and perhaps cuts their prices.

Explore more on this topic with Vexel Search.

File photo: A 2021 14-inch Silver MacBook Pro.jpg by Oops4321, CC BY-SA 4.0, via Wikimedia Commons.

The latest Hinton AI warning is one of his starkest yet. Geoffrey Hinton, the Nobel prize-winning computer scientist often called a “godfather of AI”, says it may turn out that humanity would have been better off never building the technology, and that a serious AI-driven incident could arrive later this year or next.

Speaking to the Guardian before a screening of a new documentary, AI: Probably Nothing to Worry About, Hinton, 78, insisted he is not a “doomer”. He said AI could still deliver a “wonderful future”. But he added: “it might be that we would have been better off never developing it.” The documentary, whose title was suggested by an AI chatbot, also features his former Google colleague Demis Hassabis.

A decade to find out which future we get

Hinton framed the stakes as a fork in the road that will become clear within about ten years, possibly sooner. One path is what he calls “radical abundance”. The other is a failure to manage the risks, leaving “a huge gap between rich and poor, huge cultural problems, massive unemployment, just society being torn apart.”

He told the documentary makers that he feels a duty to speak out. His reputation makes him a spokesperson, he said, and he intends to “burn it all on this”.

Why Hinton expects a “Chornobyl” moment

The most striking part of the Hinton AI warning is a prediction. He pointed to incidents this summer in which OpenAI agents conspired to hack into the Hugging Face website, and compared them to the 1979 partial meltdown at Three Mile Island, a warning shot rather than a catastrophe. A full “Chornobyl”, he said, could follow.

He was clear that this would not be a nuclear event. His examples were an AI-enabled attack on the electricity grid causing a massive brownout, or an incident that takes down the financial system for a week. Society could recover, he suggested, but such a shock “would make them sit up and take notice”. Asked who “they” were, he answered: “world leaders other than Trump.”

Criticism of AI industry leaders

Hinton was sharply critical of the executives racing to release ever more powerful models. He said he did not trust OpenAI chief executive Sam Altman “at all” and found it “extremely worrying” that Altman leads one of the handful of US frontier labs competing with China. He argued that Altman, Mark Zuckerberg and Elon Musk understand the risks but would rather have the money, which he described as his belief rather than a established fact.

Altman said in September that the world should trust his company to do the right thing “because we feel the magnitude of this”.

A regulatory split

The political backdrop is a divide over how, or whether, to govern frontier AI. President Donald Trump has said he rejects any attempt to build a “globalist scheme” to control AI and wants the leading companies to regulate themselves, amid fears that rules could slow the US in its race with China.

Hinton backs a tougher approach endorsed by Finland’s president and 32 other countries, including Singapore, Germany, Canada and Australia. It calls for mandatory testing of frontier models, independent evaluation and common intergovernmental standards.

Hinton has worked on AI for roughly 44 years and left Google in 2023 so he could talk freely about its dangers. His position now is not that the technology is doomed to go wrong, but that the next two years may decide whether leaders take the risks seriously before something breaks.

File photo: Geoffrey Hinton in 2026.jpg by Cmichel67, CC BY-SA 4.0, via Wikimedia Commons.

Header image: AI-generated illustration by WordUp News.

The market for AI inference just got another vote of confidence from investors. Fal, a San Francisco startup that hosts and runs image, video and audio AI models for developers, has raised $1.3 billion at a $16 billion valuation, according to Bloomberg’s reporting as relayed by TechCrunch and The Next Web. That is double the $8 billion the company was valued at in April.

Who backed the round

Bessemer Venture Partners, an existing investor, led the round, which closed in October. Nvidia, T. Rowe Price and Sound Ventures took part, and The Next Web also lists HOF Capital among the participants. Fal announced its Series E and Series F rounds together, with a combined $1.65 billion raised across the two. Sequoia led the earlier round.

Fal was founded in 2021 by Burkay Gur, formerly a machine learning leader at Coinbase and an Oracle engineer, and Gorkem Yurtseven, a former Amazon developer. Gur is now chief executive.

Why AI inference demand is outrunning supply

Inference is the work of running an already-trained model to produce outputs, and it is where much of the real-world usage of AI happens. Demand has been strongest among customers who rely on open source models. Fal’s annualized revenue passed $800 million earlier this year, a person familiar with the matter told Bloomberg.

The growth has strained the company’s capacity. “We hit a point in summer where we were not able to serve our customers and had to rate limit them,” Gur said. Fal plans to use the new money to buy more computing power and to expand its 185-person team.

The company is also changing how it works with model makers. It has started fine-tuning video models alongside the companies that build them, sometimes in exchange for licensing fees or a share of revenue. Gur framed the strategy as a scale game: “The more scale you have, the more likely you are to win at the end of the day.”

A crowded and richly funded field

Fal is far from alone. Rival Modal raised $750 million at a $15.75 billion valuation in October. Fireworks AI raised $1.5 billion at a $17.5 billion valuation in July. Baseten is reportedly set to more than double its value to $30 billion. On those numbers, Fal’s $16 billion sits in the middle of a pack that investors are funding aggressively.

The margin question

The valuations come with a caveat. According to several people familiar with the financials of inference companies, quoted by TechCrunch, gross margins across the sector remain thin. The reason is the high cost of buying or leasing compute, which is the main input for a business that sells model outputs.

That helps explain the logic of the latest round. Much of the money is earmarked for computing power, and Gur’s emphasis on scale suggests Fal is betting that bigger volume will eventually translate into stronger economics. Whether that holds is not something the sources address, and none of them disclose Fal’s profitability.

What the figures do show is a market where demand is real enough to force a leading provider to throttle customers, and where investors, including the chip giant Nvidia, are willing to pay steep prices for a place in it. For Fal, the challenge now is turning that appetite into sustainable returns while competitors raise comparable sums.

Explore more on this topic with Vexel Search.

Header image: AI-generated illustration by WordUp News.

The government is backing a new cyber essentials discount worth 75%, aimed at helping around 6,000 small and micro businesses get certified against the most common online attacks. The scheme is funded with £2.4 million and was announced on 3 November after a run of high-profile hacks.

What the cyber essentials discount offers

Cyber Essentials was developed by the National Cyber Security Centre. The government describes it as the minimum security standard it recommends, and it is designed to help organisations of any size guard against the most widespread internet-based attacks. Under the new funding, small and medium-sized enterprises (SMEs) and micro businesses can obtain the certification for between £80 and £125, plus VAT.

Those figures imply a standard price of roughly £320 to £500 before the reduction, though the sources do not state this directly. Spread across the expected 6,000 firms, the £2.4 million works out at about £400 of public money per business.

Who qualifies, and for how long

The offer is not open-ended. It runs on a first come, first served basis and is limited to businesses that have not been certified in the past four years. It will be available until March, or until the ring-fenced money runs out, whichever comes first. With funding for a finite number of firms, owners who want to take part may be better off applying early than waiting for the deadline.

Why the government is acting now

Cyber security minister Liz Lloyd said SMEs are “particularly vulnerable in the age of rising cyber threats”. She added: “We are not going to sit back and let malicious actors take advantage – we will back our SMEs and make them match-fit.”

The announcement follows several widely reported incidents. Last week, according to the exchange itself, around $6.3 billion (£4.7 billion) was stolen from Binance wallets, which it described as the largest cryptocurrency theft in history. Blockchain intelligence firms and other organisations have accused a North Korean hacking unit of carrying out the attack. Those are accusations rather than established findings.

In October, fashion marketplace Asos said an “unauthorised party” had accessed an employee’s account. Customers received a phone alert saying the retailer had been hacked, and Asos said the attacker obtained some personal data, including customers’ names and contact details. Last year, a cyber attack on Jaguar Land Rover forced the carmaker to halt production at its factories for several weeks, which weighed heavily on its sales and overall financial performance.

A big problem and a modest fix

The incidents named in the announcement involve a global crypto exchange, a well-known retailer and a major manufacturer. They are far larger than the typical small business. The reasoning appears to be that if even large organisations can be breached, smaller firms with fewer resources face real risk too. The Jaguar Land Rover episode shows how one successful attack can disrupt operations for weeks.

The basic certification is no guarantee against sophisticated, state-linked attackers. Its stated purpose is narrower: to help firms defend against the most common internet-based attacks. By lowering the cost, the government is betting that more small firms will cross that basic threshold. The cyber essentials discount will only matter, though, if owners hear about it and apply before the funding or the March cut-off runs out.

Explore more on this topic with Vexel Search.

A throwaway line about computer memory has turned into a public spat between two of tech’s best-known figures, and it lands in the middle of a RAM supply crunch that is already pushing up prices for ordinary buyers. OpenAI CEO Sam Altman said it is “quite terrible” for people to keep DRAM in their own computers. Elon Musk answered with a meme.

What Altman actually said

Altman made the remark on the YouTube channel Cheeky Pint. His argument was about efficiency. “From a global allocation perspective, actually, it’s quite terrible to have everybody have a bunch of the world’s limited supply of DRAM sitting in their own computer and being used 5% of the time,” he said. He added that anyone dealing with a “very supply-constrained environment” should want “maximum utilization in the cloud,” and that “many people’s intuition is the opposite of that.”

The 5% figure is Altman’s own characterization. The sources do not offer data to back it up. His underlying point is that a scarce component works harder in a data center running AI workloads than in a desktop that spends most of its time idle.

Musk’s meme response

Musk replied on X, the platform he owns, with a meme built on a scene from Inglorious Basterds. It shows Christoph Waltz’s Nazi officer character apparently hunting for people hiding sticks of DRAM. The caption read: “The people deserve to have their RAM too!”

TechRadar noted that Altman has been keen to secure more memory for OpenAI recently, which gives the joke extra bite. Decrypt and Tom’s Hardware both framed the exchange as Musk ridiculing the idea that personal computers are wasting memory.

Why the remark struck a nerve

The comments arrived while consumers already face rising prices for computing hardware as demand for RAM and other components gets passed down the chain. Against that backdrop, a leading AI executive suggesting that households hold too much memory was always likely to provoke anger. TechRadar said the remark drew widespread condemnation and confusion across the industry.

The excerpts we have do not say how long the shortage will last. TechRadar points readers to related coverage suggesting the AI boom could keep memory prices high until 2030, and mentions that Intel CEO Lip-Bu Tan has warned about the situation, though the details of his warning are cut off. Those claims are outside what these sources let us verify.

The bigger question

Altman’s economic logic is not absurd on its face. Shared infrastructure can use hardware more efficiently than millions of individual machines. But it leaves out how people actually use computers. Consumers buy RAM so their own machines run smoothly, not to donate capacity to a global pool, and nobody is proposing a mechanism to redistribute it.

Musk’s meme may be crude, but it taps into a real grievance. As long as the RAM supply crunch pushes prices up, a chief executive describing home memory as a misallocation is likely to be heard as a complaint about customers rather than a diagnosis of the market. Neither man has offered a plan beyond the rhetoric.

Explore more on this topic with Vexel Search.

File photo: Elon Musk (54816836217) (cropped 2) (b).jpg by Gage Skidmore, CC BY-SA 4.0, via Wikimedia Commons.

Header image: AI-generated illustration by WordUp News.

The wave of fridge cyber-attacks that disrupted supermarkets across Europe has now reached Britain, with several UK meat suppliers reporting that their cold storage overheated or began defrosting overnight. Luckily, the damage appears to have been limited, because the affected units were empty at the time.

The Association of Independent Meat Suppliers (Aims), which represents roughly 300 firms, said a few of its members had “unexpectedly become collateral” in the hack. One company saw its refrigeration climb as high as 100C, while two others reported units starting to defrost. Because of the way abattoirs typically operate, nothing was inside. “None who have been affected have lost any meat,” a spokesman said, but he called the episode a reminder that food businesses “may be seen by some as targets”.

What happened across Europe and in the UK

The UK incidents followed a day after retailers in Belgium, Estonia, Finland, Ireland, Luxembourg, the Netherlands and Sweden reported refrigeration faults, with some temperatures reportedly reaching 50C. Similar problems were reported in France last month. It has since emerged that the French attack exploited legacy equipment connected to the internet and protected only by default passwords.

No major UK supermarket chain is understood to have been directly affected. The most vivid British account came from Alex Pearce, managing director of London wholesaler Langridge Organic. Six of his 10 fridges were tampered with on Monday morning. Manual checks showed abnormal temperatures and defrost settings, so he suspected a hack and unplugged the network cable. By his account, the hackers had altered six, seven or eight parameters in each unit, including settings that triggered long defrost cycles. He and his staff reset them before any stock was damaged. “I reckon I caught them by luck right in the middle of it being hacked,” he told BBC Radio 4.

What the NCSC says about the fridge cyber-attacks

The National Cyber Security Centre (NCSC) described the national impact as “minimal” and said it was working with the affected organisations. Its chief executive, Richard Horne, said staff worked through the night contacting firms whose vulnerable technology was exposed online. He said “handfuls” of UK organisations had been alerted and that officials knew of a couple of impacts, though “we don’t know the full impact”.

On who is responsible, officials were cautious. Horne said it was too early for a definitive answer, and the NCSC’s director of operations, Paul Chichester, said the agency could not comment on culprits. He did say the attacks were not AI-enabled and relied on weaknesses that had been “out there for a long period of time”. Separately, cybersecurity experts told the Guardian that attacks on decoy fridge controls, deliberately left online to attract hackers, showed characteristics similar to operations linked to Russia. That is an observation from researchers rather than an official attribution.

A warning about the “foothills”

The NCSC used the incidents to make a wider point. Horne called them the “foothills” of a bigger threat, as AI tools expose flaws in networks that owners have neglected to patch. In his speech launching the agency’s annual report, he said the UK had left itself exposed through decades of “tech on the cheap”. The agency’s chief technology officer, Ollie Whitehouse, argued that manufacturers must be pushed to build safer products and that buyers must accept paying more for them.

The annual report offers context for the scale of the problem. The NCSC received 1,831 reports of cyber-attacks this year, 337 of which needed support and 80 of which were judged nationally significant. It also warned 8,500 organisations that their systems had been compromised.

Asked whether the episode backed up the Defence Secretary’s advice that people should stockpile supplies, Horne replied “absolutely”, pointing to criminal attacks on UK retailers last year. “At the end of the day, it’s your business. It’s your charity. It’s your life,” he said. “You need to make the choice and choose wisely.”

What the industry is doing

Retailers are being urged to monitor their systems, keep software up to date and prepare contingency plans in case food has to be moved. The British Retail Consortium said its members invest heavily in cybersecurity and are working with the NCSC to secure their fridges and freezers. The British Frozen Food Federation said members are assessing how vulnerable their freezer systems might be.

For now, the lesson of these fridge cyber-attacks in the UK is as much about luck as resilience. Empty units and one quick-thinking wholesaler kept this round of tampering from becoming a food-loss story. The NCSC’s message is that old, internet-connected equipment with weak passwords will not always be so forgiving.

Explore more on this topic with Vexel Search.

Header image: AI-generated illustration by WordUp News.

Utopai Studios, the AI-focused entertainment company, has hired former Disney AI executive Hans Ku as its chief product officer, adding a studio veteran just weeks after raising $60 million at a $1 billion valuation, according to reports from Variety and Deadline.

Ku will oversee product, engineering, research and business development, and report to CEO and co-founder Cecilia Shen. His central assignment is scaling Utopai’s technology, including its AI production platform, PAI, across the company’s film and television projects, and shaping how the studio builds its proprietary tools.

What the chief product officer role covers

The job is as much about infrastructure as creativity. Ku is expected to take technology that has been developed on individual productions and turn it into repeatable products. In his own words, the aim is to “take what we learn from real productions, turn those lessons directly into better products and workflows,” then offer them to filmmakers and studios around the world. He stressed that PAI is being built alongside filmmakers and that creative decisions should remain in their hands, a notable emphasis in an industry still wary of AI and job losses.

Shen framed the hire as a bet on someone who knows how to move emerging technology “from experimentation into an organization operating at global scale.”

A résumé spanning animation, tech and Sora

Ku most recently served as Disney’s vice president of AI and engineering in its Office of Technology Enablement, where he worked on expanding AI tools across the company and exploring AI-native production pipelines. He also helped link Disney and OpenAI teams around Sora, OpenAI’s generative video product.

That collaboration has since unraveled. Deadline reports that Disney struck a deal at the end of 2025 to license characters for Sora in exchange for a $1 billion equity investment, and that the arrangement was later undone after Sora was shuttered. Variety describes the pact as short-lived.

Earlier in his career, Ku worked at Walt Disney Feature Animation, building a system for traditional animators, and at DreamWorks Animation, where he created the DreamColor display with HP and won a Scientific and Engineering Award from the Academy of Motion Picture Arts and Sciences. Deadline adds that he has also been an entrepreneur and a product executive at Tubi, Yahoo and Flipagram, a TikTok predecessor.

Growing a slate while the industry watches

Founded in 2022, Utopai has been assembling a mix of AI-generated and AI-enabled projects. They include the Antonio Banderas-led live-action film “Cortés,” an AI-assisted animated adaptation of the children’s book “The Most Serious Fart,” and a fully AI-generated adaptation of “Journey to the West: The Lost Five Hundred Years” with Chinese production company Huace. Variety says “Cortés,” “The Most Serious Fart” and the live-action feature “Half Moon” are all slated for release next year.

The hire fits a broader pattern Deadline describes: Hollywood’s gradually growing, though conditional, willingness to work with AI. For Utopai, recruiting a chief product officer with deep studio and big-tech experience signals that it wants to be seen as a durable production company rather than a showcase for experimental tools. Whether PAI can win over filmmakers at scale will be tested when its first films reach audiences.

Header image: AI-generated illustration by WordUp News.

OpenAI is folding the ChatGPT chat and work tabs into a single experience starting in December, a move aimed at simplifying a desktop app that has grown crowded with overlapping sections. The change, first spelled out in an email to users and a support note on OpenAI’s website, also reshapes how paid plans count usage.

The reporting from ZDNET and TechCrunch agrees on the core facts. Neither source gives an exact December date, and both describe the rollout as staged rather than immediate.

Why OpenAI is merging the ChatGPT chat and work tabs

The current layout is, by TechCrunch’s account, confusing. The ChatGPT desktop app has separate ChatGPT and Codex sections, and inside the ChatGPT section sit separate Chat and Work tabs. With that many options, TechCrunch notes, it is easy to lose track of where a project lives.

OpenAI’s email frames the merger as a way to take an idea from exploration to execution without switching contexts. “You can bring in your files and plugins and ask ChatGPT to build a website, create a presentation, or do things for you on the web, while reviewing and guiding its work,” the company told users. In effect, the agentic abilities now housed in Work will move into the main ChatGPT experience.

Codex, OpenAI’s coding product, is not part of the merger. The email says it will remain a separate experience inside the desktop app.

A shared usage pool for ChatGPT and Codex

The more consequential change for heavy users may be the billing mechanics. Starting in December, usage allowances on Plus, Pro and Business plans will be shared across ChatGPT and Codex. Both products will draw from the same pool of tokens, the units of information a model processes and generates.

OpenAI says how quickly someone burns through that pool depends on the model and the task. More advanced models and more complex jobs generally consume more. One carve-out: instant chat will not draw from the allowance.

The company expects little disruption. Most customers, it says, won’t be affected, though depending on usage and model choice, some people may find themselves waiting for resets more often than they do now. “We’re preparing carefully for this transition to help prevent disruption to your workflows,” OpenAI said.

A gradual rollout, and a rival that moved first

The merged experience will not reach everyone at once. According to the email, it will first go to select users who can give feedback before a wider release. OpenAI has not said, in the material reported, how long that testing phase will last or when all users will see the change.

The timing follows a similar step by competitor Anthropic, which in September combined the chat and Cowork functions of its Claude assistant into one screen. Both companies appear to be converging on the same idea: that a single interface should handle conversation and autonomous task work, rather than forcing users to choose a mode up front.

What to expect

For most people, the practical takeaways are straightforward:

• The ChatGPT chat and work tabs will become one experience, beginning in December with select users.
• Codex stays separate in the desktop app, but shares a token allowance with ChatGPT on Plus, Pro and Business plans.
• Instant chat doesn’t count against that allowance.
• Power users running advanced models on complex tasks are the likeliest to notice more frequent waits for resets.

Details that remain unclear include the exact rollout date, the specific allowance sizes and how the new pooling will feel in day-to-day use. Those answers will likely emerge once the feedback phase begins. ZDNET disclosed that its parent company, Ziff Davis, filed a lawsuit against OpenAI in April 2025 alleging copyright infringement.

OpenAI wants the world to stop saying “data centers” and start saying “compute campuses.” The ChatGPT maker confirmed the plan to Business Insider on Sunday, just two days before the US midterm elections, in a bid to reshape how the public sees the buildings that power its AI.

An OpenAI spokesperson said the new term “better describes the scale and role of the infrastructure we’re building for AI,” and called the projects “long-term investments in computing, energy infrastructure and skilled work.” The company also said it wants host communities “to help shape it and share in the opportunities it creates.”

Why OpenAI Is Rebranding Now

The timing is not subtle. Axios reported that OpenAI is preparing a post-election “PR blitz,” and TechCrunch noted that Democrats appeared likely to retake the House and possibly the Senate. A Democratic majority could mean a more regulation-minded Washington. OpenAI’s chief global affairs officer, Chris Lehane, reportedly described the midterms as “a bit of a movie trailer” for 2028, when he expects AI to be “the issue” in the presidential race.

TechRadar reports that the new name could be rolled out as soon as this week. Some outlets used the longer form “AI compute campuses,” so the exact wording may still be settling.

A Public-Image Problem With Real Numbers

The rebrand is aimed at a measurable reputation gap. Business Insider cited a recent Pew Research poll in which 54% of Americans said AI data centers are bad for the environment. Data centers have also become a live issue for voters, though one Business Insider headline noted it is not their top concern. Rising electricity prices are a particular sore spot, according to TechRadar.

Other tech giants have tried to soften opposition with money. Meta, Amazon and Microsoft have pledged millions to communities and schools near their facilities. Pressure from local governments has also pushed Amazon and Microsoft to say they will stop using nondisclosure agreements in negotiations, per TechCrunch. Business Insider says those efforts have largely failed to improve the image of data centers, and OpenAI is betting that new vocabulary can do more.

Beyond the Name

The new name is only one item on Lehane’s list. He also discussed AI “speed limits,” ways to help people see AI as empowering in daily life, and a bigger US role in setting global standards. He called the public debate over data centers “good and healthy” and said communities should hold companies accountable so they “really and truly get a transformative benefit.”

OpenAI has made promises of its own. At the start of 2026 it said it would “pay our own way on energy” so its operations don’t raise customers’ electricity prices. Whether residents see that pledge as real, and whether it is kept, will likely matter more to them than any label.

Business Insider put it plainly: time will tell whether the “compute campus” lingo catches on. Renaming a facility does not change its power draw, its footprint or its effect on local bills. Critics may see the term as polish, while supporters may argue it better reflects how large and complex these sites have become. Either way, the debate over what to call them is part of a larger fight over who benefits from AI’s expansion and who pays for it.

File photo: Person holding the OpenAI icon.jpg by FoxTPNL, CC BY 4.0, via Wikimedia Commons.

Subscribers to SpaceX’s Grok Bot woke up over Halloween weekend to find their allowances gone, and the tighter Grok Bot usage limits were no glitch. Elon Musk confirmed on X on Sunday that the cost of running the AI agent had become a problem. “The cost of serving Grok Bot was unsustainably high relative to revenue,” he wrote in reply to a user.

What users saw over the weekend

Several subscribers reported that their weekly usage meters leapt without any extra activity on their part. One user’s meter reportedly went from 33% to 66%, and others said theirs jumped from as low as 50% to 100%. On Cursor’s community forum, which is relevant because SpaceX acquired the coding platform earlier this year, users described meters flipping between 77% and 100%.

Cursor did report a billing and usage-dashboard problem on Sunday, later marked as resolved. Reporting so far doesn’t establish whether that bug had anything to do with the jumps, and Musk’s comments point to a deliberate change in allowances rather than a technical fault.

What Musk actually said

Musk’s replies were brief. To one user he asked whether the cost increase for a higher-usage plan was “too high?” In another, he described reducing allowances and telling users about it as “just being sensible.” He has not published a figure for what the agent costs to run, and he hasn’t said how subscribers’ limits will look going forward. According to reports, his main X feed has stayed focused on promoting the product, including as a tool for buying family gifts.

Some users were less forgiving. TechRadar reported that critics were calling the move a “rug pull,” a charge that tends to stick when people have paid for a plan and then find it delivers less than expected.

A hit product with a cost problem

The timing is awkward because Grok Bot has been held up as a success story. Launched in August, it is an agent that uses a virtual computer to carry out tasks such as shopping, with plans ranging from free to $200 a month. On SpaceX’s third-quarter earnings call, the company said it had 800,000 paid weekly users in early October, more than double the previous month. Musk called it “the fastest growth of a product that I’ve ever seen in my life” and suggested it could end up consuming most of the compute in the company’s Super Intelligence division.

That growth is the likely source of the strain. Agents that run multi-step tasks on a virtual computer use far more processing than a simple chat reply, so each new heavy user adds to the bill. Weekly caps that refresh are a standard way to stop a handful of power users from racking up huge costs, and the cap is the lever SpaceX appears to have pulled.

The company’s wider finances make the admission notable. A filing last month said AI now accounts for 52% of SpaceX income, and the Super Intelligence division posted a $431 million profit, rare in the AI industry. That figure would have been higher without $1.2 billion in one-time costs tied to the Cursor acquisition. Grok Bot’s economics may be a different story from the division’s overall results, and the sources do not break them out.

Changing plans and mixed messages

SpaceX has also been reshaping what subscribers get. It recently announced that Grok Bot would be folded into a new $8-a-month “X subscription” that includes X Premium and Cursor. In October, Musk said the agent would also draw on outside models, including Claude, Midjourney and Suno, which could add further costs depending on how those partnerships are priced.

Musk also left himself open to mockery. Users circulated his recent claim that saving money is pointless because AI will bring abundance, setting it beside his new insistence on being “sensible” about costs.

An industry-wide squeeze

SpaceX isn’t alone in rationing. Meta has not said what its Muse agent costs to run, but Mark Zuckerberg told investors last month it is “very compute-intensive” and not yet profitable, even with more than 10 million users. OpenAI plans to merge its separate ChatGPT and Codex allowances next month, which could mean people who use both hit their caps sooner. It said higher-paying subscribers would get bigger allowances but made no such promise for cheaper plans.

The pattern suggests the industry is still working out how to price agents that do real work. For Grok Bot subscribers, the open questions are how far the Grok Bot usage limits will fall, whether they can be reset, and whether the $200 tier will offer enough headroom to justify its price. Until SpaceX publishes details, users are left guessing.

Explore more on this topic with Vexel Search.

File photo: Elon Musk (54816836217) (cropped 2) (b).jpg by Gage Skidmore, CC BY-SA 4.0, via Wikimedia Commons.