Tag: big tech

France Votes To Tax Big Tech By The Gigabyte – And Someone Just Bent The Barrel Back

France’s National Assembly voted on Friday to tax the internet by the gigabyte. An amendment to the 2027 budget would charge every company that sends data to French internet subscribers €0.055 per gigabyte, and its Socialist authors expect it to raise €10 billion a year.

But by Sunday, someone had turned the tax on the party that wrote it.

An internet user put a 44MB PDF from the Socialist Party’s website on a server on his own machine, downloaded it in an endless loop and tallied the tax the traffic would generate: 13,109 downloads, 576 gigabytes and €31.70 in under four hours.

In the simulation, the counter shows the loop running from a local address, 127.0.0.1, so no traffic reached the party’s servers and no internet provider metered a byte. The math matches the amendment’s rate, though: 576.4 gigabytes at €0.055 comes to €31.70.

The potential damage could be huge. VideoLAN, the French nonprofit behind the VLC media player, said the tax would cost it more than €1 million a year – as the amendment doesn’t say how its revenue-based cap would apply to a nonprofit that earns little.

The roll call shows how slim the majority was. Only 35 deputies voted, 11 in favor, 10 against and 14 abstaining, and the amendment was adopted on Oct. 30 on the strength of its 11. National Rally’s 10 deputies abstained, and had they voted no, the amendment would have failed 20 to 11. The yes votes came from five Socialists, two members of La France Insoumise, two Greens and two from the Democratic and Republican Left. All 10 votes against came from the center and the right: four from Ensemble pour la République, five from Droite Républicaine and one from Les Démocrates. Four of La France Insoumise’s deputies also abstained.

The text was filed Oct. 9 by Socialist Laurent Lhardit and dozens of his colleagues, former President François Hollande among them. It would make the “content provider” that sent the data, not the internet provider, liable for the tax, which it calls a tax on digital data producers. Internet providers would be required to meter every sender’s traffic into French homes and businesses under the supervision of the telecom regulator Arcep and the tax administration, and report the totals to the tax authority each quarter. The amendment bars collecting anything about what the data contains or where it goes. When a sender uses a content delivery network, the volume counts as everything leaving that network. The details of the metering would be left to a government decree.

Keep reading

‘He can pry it from my cold, dead hands’: Sam Altman says it’s ‘quite terrible’ that you’re using RAM in your PC when there’s not enough for AI data centers

The increasing demands from data centers powering AI engines has plunged us into a global RAM shortage crisis, and the CEO of one of the companies responsible thinks the memory that is available should be in data centers rather than individual PCs.

Offering up an opinion that’s unlikely to add to his popularity among the general public, OpenAI boss Sam Altman told the Cheeky Pint podcast (via IGN) that “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.”

Individual PCs rarely make full use of their RAM, was Altman’s larger point, whereas data centers often do. “You want maximum utilization in the cloud if you’re trying to deal with this very supply constrained environment,” the OpenAI CEO continued. “And somehow many people’s intuition is the opposite of that.”

The PC owners Altman was referring to have had to deal with rising prices, job insecurity, and data center pollution thanks to the rise of AI chatbots like OpenAI’s ChatGPT, so the opinion he’s shared is unlikely to be supported by the masses.

As you would expect, consumers aren’t particularly pleased with what Altman has had to say. A Reddit thread discussing the comments is packed with pushback against AI, data centers, and the RAM shortage.

Keep reading

Google Plunges After Ugly Top And Bottom Line Miss, As AI Cannibalizes Search And Ads

Today, ahead of the doubleheader of Mag 7 reports when both Microsoft and Google parent Alphabet report, we wrote in our preview that Microsoft has become the consensus long, while Alphabet “still feels like the consensus funding short across megacaps.” And both have to convince a market that, as Goldman’s Rich Privorotsky put it this morning, now has “eye watering expectations”:

“These really matter. Eye watering expectations… buyside probably closer to Azure growth around high 40s%, Google Cloud around mid 90s%. The question isn’t whether those are good numbers, but whether they can surpass already high expectations… Azure probably needs a guidance raise to be properly supportive. The nuance is capacity… if they miss elevated expectations simply because they cannot bring enough compute online. That’s disappointing for near term revenue but potentially even more bullish for the scarcity premium and future spending. Watch customer concentration, how broad the demand really is and, crucially, the relationship between revenue growth, capex and margins. The market wants evidence that the incremental economics work.“

In other words, a cloud beat is just the price of admission, and the debate tonight is what all that cloud costs, and who pays for it.

Within the Mag 7, Microsoft and Alphabet sit at opposite ends. Goldman’s derivatives desk has Microsoft at 8/10 on its positioning score (the same score Meta carried into its print last week, and we all know how that went). Alphabet, meanwhile, keeps getting sold to pay for everything else. JPMorgan’s TMT trader Brian Heavey wrote this morning that “we continue to see GOOG being used as a source of funds (note there was another negative 3P check ahead of earnings – all pointing to a Search miss).” The stock dropped 10% from its October 13 high into Monday.

There are some signs of a thaw. Goldman’s US desk saw “a buyer of 50k GOOGL 15Jan 360 calls” on Tuesday, and Privorotsky notes that “people have barely begun to re-risk. What we’ve really seen so far is the abatement of supply, rather than meaningful fresh demand.”

Away from Goldman, and focusing on Alphabet, the whole debate fits in one line from JPMorgan’s Mark Schilsky, whose Tech Sketch “Earnings Preview Bonanza” (available to pro subs) frames it as “Street (Consensus) Fighter II: Search vs. Cloud”:

“On the one hand, Google Cloud Revenues are likely to crush this quarter, thanks to TPU sales. On the other hand, Search Revenue growth is expected to miss the Street because the alt data (which is always right!) suggests so… GOOGL is also on a path to where Google Cloud Revenues could be equal to Search Revenues by late 2028. That’s massive! Who thought that was possible 1 year ago? And now we’re debating whether or not Search is going to miss by like -50 bps this quarter. Who cares?!?”

Keep reading

As the AI Bubble Threatens to Burst, We Must Prepare to Fight a Big Tech Bailout

AI data centers are unpopular for good reason: they drive up utility costs for communities, crowd out everything from schools to family farms, and they’re bad for our health. But these data centers are also a financial crisis in slow motion.

The AI ecosystem is built on mountains of debt taken on not only by tech firms like Meta (which borrowed a record-breaking $25 billion in late 2025 to fund the building of data centers), but also by the companies that lease data center space to the Big Tech firms. Many of these firms have spent far more than they have come close to earning in profits: One estimate puts tech industry spending to build data centers at $1.9 trillion, but less than half of that money has been made in revenue.

The financial markets are taking notice.

There’s been significant turmoil in the tech sector over the past month — and at one critical company in particular: CoreWeave.

CoreWeave used to be a Bitcoin mining firm called Atlantic Crypto, but it started to pivot to AI in 2019. CoreWeave leases space to Big Tech firms at data centers they build and maintain. CoreWeave has paid for most of its data centers by borrowing: it currently owes more than $51 billion to investors, including two of the most infamous financial firms: Magnetar and Blackstone, as documented by journalist Elizabeth Lopatto at The Verge.

Keep reading

Apple Supports Digital ID Push With Facial Age Estimation

Apple is increasing its support of the digital ID age verification agenda and has released iOS 27.2, bringing a new way for Apple to verify your age: by scanning your face with an “age-estimation model” that will be fed by a selfie.

This is one of the ways in which Apple will check the age of its users, and it applies, as the company’s support page states, “Where required by law.”

The legal requirements that make this necessary in the US apply in Texas, and elsewhere in the world in Australia, Brazil, Singapore, South Korea, and the UK. The iOS update that introduces the new method of age verification was released on October 26.

Apple will ask for age verification when you create an Apple Account, “update the software on your device, try to change certain safety settings, or try to download or purchase apps with an age rating of 18+.”

In Australia, Brazil, Singapore, and the UK, the selfie option is available, while in other places, Apple will ask for a credit card, a scan of an ID, or look at information associated with an Apple Account.

Apple’s explains that the selfie option works by letting the user “Capture all four angles of your face by moving your head slowly right, left, up, and down.” To increase the likelihood of the scan being successful, the company advises users to “Remove anything that might cover your face, such as masks, hats, or sunglasses” and to “Make sure only your face is in the frame.”

The models that will be used to estimate age from these selfies will run “both on-device and on Private Cloud Compute,” Apple says, referring to a service it provides that “extends the security and privacy of your Apple device into the cloud.” The results will then be “combined on your device.”

Apple promises that the selfie “isn’t accessible to anyone, not even Apple,” and that “Once your age is confirmed, your selfie is deleted and isn’t retained in any form.”

Apple has not revealed the accuracy of the age-estimation model, what happens when it misidentifies an adult as a minor, or what record of the result is kept by the company. Those who are not verified as at least 18 by the model are advised to either try another selfie or use another method.

The other methods Apple will use to determine age are: looking at information associated with the user’s account, such as whether they have a credit card on file or how long they have had their Apple Account; and scanning an ID. The image of the ID is “kept for a short period of time to complete this process, after which it’s deleted,” Apple says, and adds that “Debit cards and gift cards aren’t supported.”

Keep reading

Big Tech Pushes Back On Paying For A Grid That Was Crumbling Long Before AI

When Big Tech signed President Trump’s Ratepayer Protection Pledge in March, promising to pay for all the new grid infrastructure upgrades required to service their data centers, we cautioned that “voluntary pledges from trillion-dollar tech giants have a habit of evolving.“

Seven months later, the hyperscalers are pushing back on how much of the grid’s bill is theirs to pay.

The past week brought a Democratic Senate report and a North Carolina utility settlement, and FERC is expected to rule on whether new data centers must bring their own power.

But before anyone grabs a pitchfork, ask the question none of them answers: how much of this bill is really AI’s, and how much is the cost of decades of grid neglect?

On Friday, Senators Elizabeth Warren, Chris Van Hollen and Richard Blumenthal released Power and Profits, the result of a nearly yearlong investigation into seven data center operators including Amazon, Google, Meta and Microsoft. The headline finding is as you would expect: the companies’ promises to cover their full costs apply to infrastructure serving only their own facilities, not the full cost of shared grid upgrades their arrival triggers.

The companies broadly opposed the “but-for” standard consumer advocates are pushing, under which a data center pays for any shared upgrade that “would not have been built but for the data center.” One hyperscaler told Senate staff that such upgrades often provide a “system benefit which should be shared among all ratepayers.”

They also balked at minimum bills, which advocates want set at 80% to 90% of contracted demand whether the power gets used or not. One Big Tech firm called a 90% floor “insanely high.” Another called 85% to 90% “unreasonably high,” since facilities often don’t use all of their projected demand.

Two days earlier, Duke Energy announced a settlement, signed by Amazon, Google, Meta, Microsoft and North Carolina’s utility consumer advocate, that it says ensures data centers “will pay upfront for all costs to connect to the grid.“

The settlement itself is more nuanced. Data centers prepay the full cost of equipment that serves only them, such as a dedicated substation. For shared transmission upgrades, they post a letter of credit covering the estimated cost, collateral that starts shrinking once the facility consistently draws 75% of its contracted demand.

Whether data centers get directly charged for any slice of those shared costs is punted to a Duke study due a year after approval, and would apply only to future contracts.

Keep reading

New Mexico Found a Way to Fine Platforms for Speech It Didn’t Delete

A civil jury has found Meta liable for violating New Mexico’s Unfair Practices Act, and now the judge is being asked to order the company to pay between $35 billion and $40 billion.

At a hearing on October 1 in Santa Fe, special prosecutor Randi McGinn asked Judge Francis Mathew of the First Judicial District Court to order Meta to pay this amount, while a lawyer for the social media giant said such a penalty would be “an astronomical” one, violating “a host of constitutional provisions.”

Meta is being punished for making false statements about user privacy, and also for not removing “hate speech” and “misinformation” fast enough.

The verdict was delivered on September 25, finding Meta had made 26 false or misleading statements, resulting in nearly 43.9 million violations of the state’s Unfair Practices Act.

This would translate to a maximum of $5,000 per violation, or $219 billion in all – more than the $206 billion the tobacco industry had to pay in a settlement in 1998. The state asked for about 20% of this amount.

The number of violations was arrived at by counting the number of people in New Mexico, and the number of Facebook users in the state, and then multiplying these numbers by the number of times Meta made false statements in news outlets and on its own platforms.

The state had to show only that harm was likely, not that it actually occurred.

Meta is being punished for lying about how it handled user data, and the Cambridge Analytica scandal that saw a third party harvest data from about 87 million profiles. The company is also being punished for not removing “hate speech” and “misinformation” fast enough – and the number of violations in this category is about a quarter of the total.

The New Mexico case is not the only one where Meta is being forced to pay up – in March, a Santa Fe jury ordered it to pay $375 million in a case involving child safety on Facebook and Instagram, and in August, the judge in that case added another $567 million for a youth mental health fund.

In August, 47 states reached a settlement with Meta, along with other social media companies, that will see them pay $17 billion. Judge Mathew mentioned this during the October 1 hearing, saying that due process suggests he should opt for a much lower amount than what the state of New Mexico is asking for.

During the hearing, Meta’s lawyer Matt Nicholson described the case as “about 26 statements over 11 years” and said that the state “made absolutely no showing that anyone in the state of New Mexico ever saw them, relied on them or were harmed by them in any way.”

McGinn countered that “the one thing you didn’t hear from counsel is an apology.”

Keep reading

Anthropic President Daniela Amodei Created A Stuffed-Animal Advisory Council While At OpenAI

Before Anthropic existed, Daniela Amodei reportedly developed an unconventional method for sorting through difficult decisions at work: she and her husband created an imaginary group of advisers using plush toys, according to the NY Post.

The couple gave each toy a distinct temperament and point of view, then used those characters to consider how different kinds of managers might respond to the same problem, according to an account in The Atlantic drawn from journalist Kevin Roose’s forthcoming book, The AGI Chronicles.

The Post writes that one character, Jinji, was a laid-back cat whose instinct was to lighten Amodei’s workload and skip unnecessary meetings. A pink bear called Maura represented the opposite approach — organized, demanding and focused on details. Another, Beary Bonds, was a panda meant to bring compassion and generosity into the discussion.

Despite the playful setup, Amodei apparently found practical value in the exercise. She told Roose in 2023 that it was “very useful in analyzing different personality types.”

The practice was largely unknown inside OpenAI, where Amodei had been working since 2018. It also came during an increasingly difficult period for her and her brother Dario Amodei at the company.

Daniela had moved into management of technical teams and developed disagreements with several senior OpenAI figures, including Greg Brockman and Mira Murati. Her relationship with Brockman had previously been much warmer — he had even spoken at her wedding to Holden Karnofsky in 2017 — but their working relationship deteriorated over time.

Dario, meanwhile, had his own concerns about OpenAI’s leadership and the direction of its AI development. He became particularly wary of Sam Altman and Brockman and began organizing with researchers who wanted greater emphasis placed on the risks posed by increasingly powerful AI systems.

That faction communicated privately and held meetings outside the company’s usual structure. Some colleagues reportedly referred to it as “Dario’s clique.”

Eventually, the internal disagreements turned into plans for an entirely new company. In 2020, Dario began assembling a team willing to leave OpenAI and pursue advanced AI research under a different safety philosophy.

Daniela became a central figure in that departure, including negotiating with OpenAI as the group prepared to leave. Those involved reportedly referred to the split privately as “the divorce.”

Keep reading

OpenAI Shops $30 Billion Round To UAE Funds, BlackRock As Altman’s ‘Bad Things’ Remark Draws Bipartisan Fire

With its planned IPO on ice for the moment, OpenAI is pitching a $30 billion funding round to a group of United Arab Emirates sovereign funds led by Abu Dhabi’s MGX, along with BlackRock, at a $1.4 trillion pre-money valuation, Bloomberg reported Monday. There is no lead investor, and the price was set by OpenAI rather than negotiated. Meanwhile CEO Sam Altman has sparked a firestorm in DC with comments that the world should accept “some bad things happening” for the benefits of AI.

The OpenAI round

The UAE funds are considering an investment up to $10 billion between them at that $1.4 trillion valuation – which is is 64% above the $852 billion post-money valuation of the $122 billion round in March. The figure would also place OpenAI’s valuation above Anthropic, which was at $965 billion in May.

OpenAI’s annualized revenue has reportedly passed $40 billion, up 70% since July. That makes the ask roughly 35x run-rate for a company that Fortune says booked $6.7 billion of revenue and an operating loss in Q2, and that the FT reported spent $34 billion last year. One FT source said OpenAI “needs capital.” OpenAI says the March round left it with plenty. OpenAI filed confidentially for an IPO on June 8. Then, on Sept. 12, Altman told Fortune it wouldn’t be 2026 – “an ill-advised moment to go public,” given what’s going on with safety (and then just recently said ‘screw it’ – AI is worth the danger). That said, in April the WSJ reported that OpenAI had missed revenue and user targets and that CFO Sarah Friar was worried the company might not be able to pay for future compute contracts if revenue failed to catch up.

Then there’s the backstop. As we reported last November, Friar suggested the federal government could “backstop” OpenAI’s data-center financing. In June, as we detailed, Altman began floating a plan to hand small OpenAI equity stakes to ordinary Americans, which we read as a backdoor backstop. We asked at the time whether the bailout would come before the IPO or after. Nobody mentioned the third option: a $30 billion bridge round, priced by the issuer, in between.

DeepSeek

Overnight, Bloomberg also reported that DeepSeek is close to locking in at least 80 billion yuan ($12 billion) of new funding, with signed term sheets that could take the total to 100 billion yuan – twice the 50 billion it originally set out to raise. Tencent and CATL are writing the biggest checks. This is the same round DeepSeek paused in late July, as we noted, after transcripts leaked of founder Liang Wenfeng saying that Huawei was giving DeepSeek about 16,000 Ascend 950s while the internet giants got hundreds of thousands, and that DeepSeek could get hold of some processors he called “noncompliant.” The round restarted in August at a valuation of about $74 billion and has now blown through its target. On annualized revenue reported at $400-500 million in July, that valuation is well over 100x sales. OpenAI at 35x looks cheap next to it.

Keep reading

Google Accused of Illegally Killing a Massive Forest for Its New Data Center Project

As the AI race reaches white-hot temperatures, big tech companies are caught up in a heated race to build as many pollution-belching and power-sucking data centers as they can.

And while the Trump administration is largely clearing their way by easing environmental rules for the polluting facilities, other governments are less inclined to give companies a pass.

As Finnish newspaper Kaleva reported last month, Google is under fire for clearing over 300 hectares — the equivalent of almost 500 football fields — for two new data centers in the northern part of the European nation without concluding the necessary environmental assessments.

Despite the assessments still being in progress, the tech giant’s contractors started clearing trees in the area months ago, raising the alarm bells at the Finnish Nature Conservation Association.

“Either this is gross incompetence from Google’s Finnish project company, Tuike Finland Oy, or gross disregard for Finnish legislation, or at worst both,” said the association’s chair, Hanna Halmeenpää, in a statement, as translated by Wood Central.

Satellite images show that four separate sites were bulldozed to make way for two new data centers.

Google maintains that its felling of the trees met the requirements of the country’s Forestry Act and that it had completed extensive environmental evaluations.

Nonetheless, the Finnish Nature Conservation Association has since asked Finnish regulators to suspend activities until the conclusion of the ongoing environmental assessments.

Halmeenpää said she was worried that local bird species may have been impacted by the area being razed. The forest also serves as a critical ecological connection, allowing animals to spread from one habitat to another.

The incident highlights the explosive growth in new AI data center projects across the globe, a major infrastructure undertaking that could have a significant impact on the local environment.

Meanwhile, Google is hoping to appease the Finnish government by having picked it for its largest single investment in the continent, to the tune of $15 billion.

Keep reading