Increased power demand from data centres could drive supply cuts if big winter storms hit the US, grid organisers warn. CC-licensed photo by Daniel Lackey on Flickr.
You can sign up to receive each day’s Start Up post by email. You’ll need to click a confirmation link, so no spam.
A selection of 9 links for you. Powered up. I’m @charlesarthur on Twitter. On Threads: charles_arthur. On Mastodon: https://newsie.social/@charlesarthur. On Bluesky: @charlesarthur.bsky.social. Observations and links welcome.
A massive Cloudflare outage brought down X, ChatGPT, and even Downdetector • The Verge
Emma Roth:
»
Cloudflare, a networking company that provides DDoS protection and internet content delivery services for many companies around the globe, is recovering after a major outage that took down sites across the web. Users weren’t able to access X, ChatGPT, and even the outage-tracking site DownDetector on Tuesday morning, with some sites displaying an error message that said, “Please unblock challenges.cloudflare.com to proceed.”
The disruption, which started at around 6:20am ET (11.20am GMT), is linked to a “configuration file that is automatically generated to manage threat traffic,” Cloudflare spokesperson Jackie Dutton tells The Verge. “The file grew beyond an expected size of entries and triggered a crash in the software system that handles traffic for a number of Cloudflare’s services.” Dutton added that “there is no evidence” of an attack or other malicious activity.
Cloudflare posted an update to its status page at 9:42am ET, saying: “A fix has been implemented and we believe the incident is now resolved. We are continuing to monitor for errors to ensure all services are back to normal.”
Cloudflare CTO Dane Knecht wrote in a post on X that the service “failed” its customers and the broader internet, adding that “a latent bug in a service underpinning our bot mitigation capability started to crash after a routine configuration change we made… This was not an attack.”
«
If it’s not Amazon, it’s Cloudflare. If it’s not them then it’s someone else. We rely on giant companies for all this connectivity all the time. Guesses for what it will be next month?
unique link to this extract
Google’s Sundar Pichai warns of “irrationality” in trillion-dollar AI investment boom • Ars Technica
Benj Edwards:
»
On Tuesday, Alphabet CEO Sundar Pichai warned of “irrationality” in the AI market, telling the BBC in an interview, “I think no company is going to be immune, including us.” His comments arrive as scrutiny over the state of the AI market has reached new heights, with Alphabet shares doubling in value over seven months to reach a $3.5 trillion market capitalization.
Speaking exclusively to the BBC at Google’s California headquarters, Pichai acknowledged that while AI investment growth is at an “extraordinary moment,” the industry can “overshoot” in investment cycles, as we’re seeing now. He drew comparisons to the late 1990s Internet boom, which saw early Internet company valuations surge before collapsing in 2000, leading to bankruptcies and job losses.
“We can look back at the Internet right now. There was clearly a lot of excess investment, but none of us would question whether the Internet was profound,” Pichai said. “I expect AI to be the same. So I think it’s both rational and there are elements of irrationality through a moment like this.”
Over the past year, some analysts and tech industry critics have expressed increasing skepticism about a web of $1.4 trillion in deals surrounding Google competitor OpenAI in particular. The company has committed to spending $1.4 trillion on infrastructure over eight years, while it expects to generate around $13bn in revenue this year. OpenAI CEO Sam Altman told reporters at a private dinner in August that investors are “overexcited” about AI models and that “someone” will lose a “phenomenal amount of money.”
«
Absolutely everyone is calling this a bubble, yet at the same time nobody is ready to call the top.
unique link to this extract
US faces winter blackout risks from data centres’ power needs • Bloomberg via Yahoo Finance
Naureen Malik:
»
Rising electricity demand from data centres is raising the risk of blackouts across a wide swath of the US during extreme conditions this winter, according to the regulatory body overseeing grid stability.
Power consumption has grown 20 gigawatts from the previous winter, the North American Electric Reliability Corp. said Tuesday in its winter assessment. A gigawatt is the typical size of a nuclear power reactor. Supply hasn’t kept up.
As as result, a repeat of severe winter storms in North America that unleash a polar vortex, of which there have been several in recent years, could trigger energy shortfalls across the US from the Northwest to Texas to the Carolinas. All regions have adequate resources in normal conditions.
“Data centres are a main contributor to load growth in those areas where demand has risen substantially since last winter.” Mark Olson, manager of the reliability assessment, said in an emailed statement.
America’s power grid has been facing rising blackout risks for years as ageing infrastructure is increasingly stressed by severe storms and wildfires. Now the data centre boom, driven by the spread of artificial intelligence, is adding to the strain by supercharging US electricity growth after being stagnant for two decades.
Winter is especially risky because solar generation is available for fewer hours and battery operations may be affected. Gas supplies, meantime, could drop off because of freeze-offs or pipeline constraints.
«
You mean there might be a downside to building all these data centres? Hard to credit.
unique link to this extract
Meta scores a win after judge rules the social media giant is not an illegal monopoly • CNN Business
Clare Duffy:
»
Meta scored a major win on Tuesday after a federal judge ruled that it is not a social networking monopoly, shooting down an argument from the US Federal Trade Commission that it should be forced to spin off two of its most popular platforms.
The Federal Trade Commission sued Meta in 2020, accusing it of violating antitrust law by acquiring nascent, would-be rivals Instagram and WhatsApp to avoid having to compete with them. The seven-week trial in the case saw testimony from a series of prominent figures, including Meta CEO Mark Zuckerberg, who argued that the company has plenty of competition from platforms including YouTube and TikTok.
Federal judge James Boasberg agreed with the company’s argument in his Tuesday opinion, saying that TikTok and YouTube prevent Meta from monopolizing the social network market. He also noted that Meta’s apps and the social media landscape have changed since the FTC filed its case, most recently because of AI-generated content, which undermined the agency’s arguments.
“Meta holds no monopoly in the relevant market,” he wrote.
FTC Director of Public Affairs Joe Simonson said in a statement that the agency is reviewing its options following the decision.
“We are deeply disappointed in this decision,” Simonson said. “The deck was always stacked against us with Judge Boasberg, who is currently facing articles of impeachment. We are reviewing all our options.”
«
The FTC is such a strange organisation in this Trump world. Brendan Carr, its chief, is only interested in whatever helps Trump; nothing else, including such trivia as the US Constitution. Meanwhile its main body sues Meta over social media and is disappointed at not winning, to the extent of criticising the judge. (Boasberg opposes Trump, and signed off various subpoenas for probes into Trump behaviour.)
unique link to this extract
iPhone Air isn’t annual, iPhone Air 2 was never coming in 2026 • Apple Insider
Wesley Hilliard:
»
Rumours suggested Apple had given up on iPhone Air due to one not being prepped for 2026, but that was apparently never the case. A future model with an improved chipset and longer battery is in the works.
The iPhone Air may not be Apple’s most popular model, but it doesn’t have to be. The device was a replacement for Apple’s worst-selling iPhone Plus line and serves as a proof of concept more than anything.
According to the [Bloomberg] Power On newsletter [by Mark Gurman], the iPhone Air was never expected to have a second iteration in 2026, so rumours of any delay are inaccurate. A future model is in the works, but the focus is on a better chipset, not necessarily a second camera system.
The new chip would be built on a 2-nanometre process, which would aid in improving battery life. There are no indications of a structural change, and adding a second camera, as rumored, would likely require a redesign.
A previous rumour suggested Apple had decided to skip a 2026 iPhone Air update after poor sales, which seemed sketchy at best. That rumour was updated to suggest that Apple had delayed the product to ensure there was time for a redesign that would add a second camera.
«
Schrödinger’s phone: it’s coming but it isn’t, selling but it isn’t.
unique link to this extract
OpenAI is piloting ChatGPT group chats • The Register
Brandon Vigliarolo:
»
Feel like your team’s group chat is a bit lifeless? Remote coworkers not really collaborating as well as they should be? There’s a new way to stir the pot now that OpenAI has piloted ChatGPT group chats: cram a chatbot into the conversation and let it chime in whenever it thinks it should.
Where there are chatty bots, there’s OpenAI, naturally, which on Thursday announced the pilot of group chats in ChatGPT. The new feature is now available for mobile and web users logged into ChatGPT, be they free, Go, Plus, or Pro users – but don’t go trying to suck your friends and coworkers into a ChatGPT group chat unless you’re in Japan, New Zealand, South Korea, or Taiwan, as those are the only places getting access to the pilot.
OpenAI is pushing the new group chat feature as a way for people to collaborate at work and school, as well as a way for friends to make decisions on things like dinner destinations, vacations, and the like. Search, image and file uploading, image generation, and voice dictation are all included in group chats with ChatGPT, powered by GPT-5.1 Auto, which automatically shunts requests to the best model (i.e., instant, thinking, or a legacy model for free tier users) for the task.
«
Ben Thompson and John Gruber (the former particularly) are excited about this development, and I can see why too: in companies that are using chatbots, collaborative work is commonplace, and what you really want is for the humans to be able to refer to the chatbot in the course of their work; it’s like what Garry Kasparov calls the “centaur” mode (from when chess players use chess computers for help with tactics). Involving the chatbot in group work makes a lot more sense than having individuals working alone with them.
unique link to this extract
Bitcoin slides below $90,000 as traders grow cautious • Reuters
Rae Wee and Elizabeth Howcroft:
»
Bitcoin fell below $90,000 for the first time in seven months on Tuesday in the latest sign that investor appetite for risk is drying up across financial markets.
The risk-sensitive cryptocurrency has lost all this year’s gains and is now nearly 30% below a peak above $126,000 in October. It was down 0.5% at $91,338.47 during European trading hours, after slipping as low as $89,286.75.
About $1.2 trillion has been wiped off the total market value of all cryptocurrencies in the past six weeks, according to market tracker CoinGecko.
Market participants said that a combination of doubts around future US interest rate cuts and the risk-averse mood in broader markets, which have wobbled after a long rally, was dragging down crypto.
“The cascading selloff is amplified by listed companies and institutions exiting their positions after piling in during the rally, compounding contagion risks across the market,” said Joshua Chu, co-chair of the Hong Kong Web3 Association. “When support thins and macro uncertainty rises, confidence can erode with remarkable speed.”
Speculators who had put money into crypto in the hopes of supportive US regulation have started to pull back, causing steady outflows from ETFs and similar instruments in recent weeks, said Joseph Edwards at Enigma Securities.
“The sell pressure here isn’t extraordinary, but it’s coming at a relative weak point on the buy side … a lot of retail buyers were stung during the flash crash last month,” he said, referring to an October crash in which there were $19bn in liquidations across leveraged positions.
Crypto stockpilers such as [Micro] Strategy, miners such Riot Platforms and Mara Holdings, and exchange Coinbase have all slid with the souring mood.
«
I’m not for a moment suggesting that NOW is the moment when bitcoin is DOOOOMED; its value has long since ceased to have any relation to its real-world use – even ransomware can’t be doing enough business to make it worth that much.
But the idea that it’s a canary in an economic coal mine is much more telling.
unique link to this extract
ChatGPT is huge in India. These locally focused startups found a way to compete • Rest of World
Tauseef Ahmad and Sajid Raina:
»
When Amrith Shenava began experimenting with large language models shortly after the launch of ChatGPT, he quickly realized that Tulu — the language he and some 2 million people spoke in the southern Indian state of Karnataka — had virtually no digital data set. He decided to build one.
Shenava, who has a degree in computer science from Kent State University in Ohio, had earlier launched a translation app, and a language learning app for Tulu. To build the data set for the LLM, he had to collect voice and text data from native speakers including teachers, professionals, homemakers, and members of the Tulu diaspora.
“Most AI systems are built in the U.S. They don’t understand Indian languages or contexts,” Shenava, the 27-year-old founder of TuluAI, told Rest of World. “We need our own models that represent us.”
India has more than 1,600 languages and dialects, but most artificial intelligence systems cater to those that are widely spoken. OpenAI’s ChatGPT supports more than a dozen Indian languages including Hindi, Tamil, and Kannada, the dominant language in Karnataka. Google’s Gemini can chat with users in nine Indian languages.
Spurred by their success, and keen to be a part of the rapid global transition to AI, a handful of Indian startups are building AI tools for so-called low-resource languages such as Tulu, Bodo, and Kashmiri, which have a limited online presence and few written records. The startups are having to build data sets nearly from scratch.
TuluAI holds storytelling sessions and workshops in rural areas, in which local residents — particularly women and elders — narrate their stories, or are asked to read texts and simulate everyday conversations. Participants are taught to record and label the data. Each workshop of one to two days produces over 150 hours of labeled voice and text data, Shenava said.
The startup also collects WhatsApp voice notes from anyone who wishes to send one, with annotators checking transcripts and labels for accuracy.
“Major translation tools miss the context that gives meaning to words. The only way to fix that is to use authentic, human-recorded data that reflects real-life language use,” Shenava said.
«
Excellent idea, and it’s great that there are local resources to stop the Great Cultural Flattening that the use of chatbots could cause.
unique link to this extract
MI5 names two people in alert to MPs and peers about Chinese espionage • The Guardian
Dan Sabbagh:
»
MI5 has issued an espionage alert to MPs and peers warning that two people linked to the Chinese intelligence service are actively seeking to recruit parliamentarians.
The two people, who operate as headhunters on the LinkedIn professional networking website aiming to obtain “non-public and insider insights”, MI5 said, are also targeting economists, thinktank staff and civil servants for their access to politicians.
MI5 named them as Amanda Qiu, from BR-YR Executive Search, and Shirly Shen, who is linked to Internship Union, and told MPs and peers they were using LinkedIn to “conduct outreach at scale”.
The spy agency sent its warning to the speaker of the Commons, Lindsay Hoyle, and his Lords equivalent, John McFall, on Tuesday morning, both of whom relayed its contents to the members of their houses with a cover message.
In his email to peers, McFall said the individuals, linked to China’s ministry of state security spy agency (MSS), were “actively reaching out to individuals in our community”.
Their “aim is to collect information and lay the groundwork for long-term relationships, using professional networking sites, recruitment agents and consultants acting on their behalf”, he added.
Qiu’s profile on LinkedIn, written in English, describes her as having been the chief executive of BR-YR Executive Search for more than six years and says she is based in Beijing. Her listed interests include the UK’s Department for Transport and the Tony Blair Institute.
Shen’s profile, also largely in English, describes her as the co-founder of InternshipUnion, based in Hangzhou, eastern China. She says it “has helped hundreds of students come to China do their internship” and she describes herself as “a positive Asia girl” who would “welcome friends all over world join us to get a magic Chinese experience”.
«
Most spying isn’t safe-cracking subterfuge; it’s done in the open, being charming and getting people to trust you (and often to rely on the money you pay them) so they’ll tell you things they shouldn’t.
unique link to this extract
| • Why do social networks drive us a little mad? • Why does angry content seem to dominate what we see? • How much of a role do algorithms play in affecting what we see and do online? • What can we do about it? • Did Facebook have any inkling of what was coming in Myanmar in 2016? Read Social Warming, my latest book, and find answers – and more. |
Errata, corrigenda and ai no corrida: none notified
The Overspill: when there's more that I want to say 