This week’s reception promises to be just as grand, including a stop inside Zhongnanhai, the rarefied compound where China’s top leadership lives and works. The agenda too will be just as thorny, with Iran being a new source of tension, alongside trade, technology and Taiwan.
Such a distributed computing network makes sense in that “computation for AI inference can and should be distributed at the ‘edge,’ deployed on smaller platforms closer to population centers and users,” said Benjamin Lee, a computer architect and engineer at the University of Pennsylvania, in correspondence with Ars. “The strategy could impose much smaller impacts on the grid because inference requires a few GPUs, unlike training which requires thousands of them working in concert,” he said.
However, AI inference tasks can be as varied as document question-and-answer, software code generation, and multi-turn conversations—each with different computational requirements and performance expectations, Lee cautioned. So it will be important to ensure that individual compute nodes can deliver the performance necessary for each task, along with maintaining network connectivity among the nodes.
Lee also questioned whether it’s necessary to downsize data centers to the “granularity of a few GPUs” in order to reduce their burden on the power grid. He speculated that deploying conventional 20-megawatt data centers instead of 1-gigawatt hyperscale data centers could prove similarly beneficial.
Then there is the issue of security. XFRA nodes spread across suburbia could become more vulnerable to certain data security threats than centralized data centers. “Many side-channel attacks require physical proximity to the machine, which data centers can guard against,” Lee said. “Distributed GPUs in individual homes are much more difficult to protect.”
Thieves may also see XFRA nodes alongside houses as a tempting target, given that the Nvidia GPUs within can each sell for around $10,000. Several comment threads on Reddit have already speculated on that possibility, with some commenters suggesting they would feel tempted to secure such compute resources for themselves as residents. “Of course, there is the risk of losing the actual hardware itself to theft,” Lee said.
Any potential benefits and complications will become more evident during SPAN’s pilot deployment phase. But at a time when Silicon Valley is currently abuzz about orbital data centers and ocean-going AI data centers, data center nodes embedded in suburbia may stand on more solid footing—at least until homeowner associations catch wind of them.
Android Show: I/O Edition is a short, Android-only presentation kicking off at 1 PM ET on May 12th, ahead of Google I/O.
Android Show: I/O Edition is a short, Android-only presentation kicking off at 1 PM ET on May 12th, ahead of Google I/O.
Google I/O is still a week away, but Google’s big announcements kick off today with the “Android Show: I/O Edition,” where it’s expected to announce the major Android ecosystem highlights coming to its annual developer conference.
Android Show: I/O Edition will be streaming live on YouTube and the Android website on Tuesday, May 12th at 10AM PT / 1PM ET. Based on last year’s Android Show, we can expect the highlights to include a look at the next major Android update along with announcements about Gemini features on Android and potentially teases of some more forward-facing projects, like Aluminium OS and Android XR.
The Android Show livestream is definitely less time-consuming than I/O’s main keynote. Last year’s Android Show lasted just over 20 minutes, while the full Google I/O keynote ran for almost two hours.
The main event during Tuesday’s Android Show will likely be a look at Android’s next big update, which could include some design changes — at least, that’s what some Android users suspect. In the 15-second teaser video for this year’s Android show, the Android robot momentarily changes from its usual solid green to a translucent, glassy-look with splashes of color inside. This got some users speculating that Google is giving Android a Liquid Glass look like iOS 26, but Android ecosystem president Sameer Samat replied to a post about this saying, “Not happening!”
While the Android robot in the video does look a bit like Apple’s Liquid Glass, it also resembles early glimpses of the Gemini app redesign that has started popping up on macOS, iOS, and Android. Liquid Glass or not, it’s possible Android 17 will have some visual tweaks that Google will go over during Tuesday’s livestream, along with new features and the update’s release window.
Betas and early developer previews of Android 17 include a few key features to look forward to, like “app bubbles” that allow users to pull up in a small windowed “bubble” they can move around, like a windowed app on PC. Android 17 is also expected to include more location controls, including an option for one-time location sharing with apps and an indicator showing when a “non-system app” is accessing your location.
The Android Show: I/O Edition will likely include some highlights about updates coming to Google Gemini, particularly features baked directly into Android (or Gemini’s Android app).
One of this year’s announcements could be the broader roll-out of the Gemini redesign that users have started seeing on the chatbot’s app. It features a new pill-shaped prompt box with streamlined buttons. The redesign also adds some subtle splashes of color in the background, which look very similar to the “glassy” look of the Android robot in Google’s teaser video for this year’s Android Show.
The Android Show could give us our first official look at Aluminium OS, Google’s PC version of Android. An early glimpse of it was leaked in January in a bug report that included screen recordings showing a UI that blends elements of both Android and ChromeOS.
Even if Google does show off Aluminium OS during Tuesday’s livestream, we might still have to wait a while for a proper release. Court documents from Google’s search antitrust case note that while Android ecosystem Sameer Samat stated that Google hopes to launch it in 2026, a full launch might not happen until 2028, particularly for the “enterprise and education sectors.” It could be that Aluminium OS will only be available in beta or for testing this year, with a full launch coming later.
It’s also worth noting that Aluminium OS may not end up being compatible with all existing Chromebooks. The court documents also mention that Aluminium OS “is not expected to be able to support older hardware” due to “heavier software.”
Last year’s Android Show ended on a teaser for Android XR, with Android ecosystem president Sameer Samat donning a pair of XREAL smart glasses, ahead of some Android XR announcements at Google I/O 2025. It’s possible glasses will make an appearance at this year’s Android Show, too. Although much like last year, Google could also end up saving this for the main I/O keynote.
Swiss sneaker company On saw more strong growth during its first quarter, beating Wall Street’s expectations on the top and bottom lines even as direct-to-consumer revenue fell short of forecasts.
During the quarter ended March 31, On’s direct-to-consumer sales, revenue from its own website and stores, grew 16.4% to 322.3 million Swiss francs (US$414.2 million), falling short of the 326 million francs analysts had expected, according to StreetAccount.
Meanwhile, revenue from its less profitable wholesale channel increased by 13.3% to 509.6 million francs, beating expectations of 499 million francs, according to StreetAccount.
In a news release, the company said “even against an uncertain macroeconomic backdrop,” it decided to raise its profitability outlook and reiterate its 2026 net sales growth forecast.
“That’s not particularly aimed at the consumer …. There’s just a lot of things going on, like, for example, the war now in Iran that probably nobody saw coming,” said co-CEO Caspar Coppetti in an interview with CNBC. “We’re a bit in a bubble, I would say, as a brand, because we cater to an affluent and aspirational consumer. So our customers are not really dependent on the gas price.”
On is now expecting its gross profit margin to reach at least 64.5% in 2026, up from its previous forecast of 63%. The outlook continues to include a 20% tariff on imports from Vietnam into the U.S., even though the duty is no longer in effect after a ruling from the U.S. Supreme Court earlier this year, and excludes any potential tariff refunds that result from the decision.
Coppetti said the company has applied for a refund and is continuing to plan for a 20% tariff on Vietnamese imports because the situation remains fluid and it has concerns more duties could still come. Still, even if tariffs were to ease, it would be “immaterial” to the company’s performance, he said.
On is expecting its adjusted earnings before interest, taxes, depreciation and amortization margin to be between 19.5% and 20% — up from a previous range of between 18.5% and 19%.
Here’s how the premium sportswear company performed during the quarter compared with what Wall Street was anticipating, based on a survey of analysts by LSEG:
The company’s reported net income during the quarter was 103.3 million francs, or 31 cents per share, compared with 56.7 million francs, or 17 cents per share, a year earlier.
Sales rose to 831.9 million francs, up 14.5% from 727 million francs a year earlier.
On posted another significant sales increase as the company looks to win back investors who have soured on the stock as its rapid growth story begins to slow down. Year to date, the stock is down almost 27% as some analysts doubt the company can grow into a true footwear heavyweight that’s as popular in Paris as it is in Ohio.
As On’s profitability has increased, Coppetti said it is reinvesting in the brand and its growth areas, including apparel and new sports like tennis. The strategy has been particularly effective in China, where sales are growing by a high-double-digit percentage and its apparel penetration is as high as 30%, compared with around 6% companywide.
That stands in stark contrast to Nike, which has seen its business struggle in the region as Chinese consumers opt for local brands and move away from legacy incumbents.
“Chinese consumers are becoming more and more savvy, and they’re looking for the special things. So either they go local, or they’re looking for that extra touch,” said Coppetti. “We’re also European. We’re Swiss and so, you know, the high quality, the attention to detail, really resonates.”
Just before the quarter came to an end, On announced co-founders David Allemann and Coppetti would become the company’s co-CEOs, replacing Martin Hoffmann, known as the company’s face on Wall Street.
Hoffmann had been CEO since 2021, but he shared the job with fellow CEO Marc Maurer for most of that time and was only a solo chief executive for less than a year before On shuffled up its C-suite yet again.
In a news release at the time, the company framed Hoffmann’s departure as a “planned hiatus” and an opportunity to “pursue philanthropic interests,” but his decision to step down also came as the company has grown increasingly complex.
Coppetti told CNBC that On has been “founder-led” from the beginning, so there won’t be any major changes now that he and Allemann have taken over.
“Nothing changes on the strategy,” he said. “We remain as committed as ever to executing this premium strategy with a good mix of ambition and Swiss conservatism, so to speak.”
“So long as the visit proceeds smoothly and Trump concludes he was treated respectfully, then the uneasy calm in the bilateral relationship will endure. If, on the other hand, Trump leaves feeling disrespected or trifled with, then he could have a change of heart,” says Ryan Hass, Director of the John L Thornton China Centre at the Brookings Institute.
Both privilege escalation vulnerabilities stem from bugs in the kernel’s handling of page caches stored in memory, allowing untrusted users to modify them. They target caches in networking and memory-fragment handling components. Specifically, CVE-2026-43284 attacks the esp4 and esp6 () processes, and CVE-2026-43500 zeroes in on rxrpc. Last week’s CopyFail exploited faulty page caching in the authencesn AEAD template process, which is used for IPsec extended sequence numbers. A 2022 vulnerability named Dirty Pipe also stemmed from flaws that allow attackers to overwrite page caches.
Researchers from security firm Automox wrote:
Dirty Frag belongs to the same bug family as Dirty Pipe and Copy Fail, but it targets the frag member of the kernel’s struct sk_buff rather than pipe_buffer. The exploit uses splice() to plant a reference to a read-only page-cache page (for example, /etc/passwd or /usr/bin/su) into the frag slot of a sender-side skb. Receiver-side kernel code then performs in-place cryptographic operations on that frag, modifying the page cache in RAM. Every subsequent read of the file sees the corrupted version, even though the attacker only ever had read access.
CVE-2026-43284 is found in the esp_input() process on the IPsec ESP receive path. When an skb object is non-linear but lacks a frag list, the code skips skb_cow_data() and decrypts AEAD in place on the planted frag. From there, an attacker can control the file offset and the 4-byte value of each store.
CVE-2026-43500, meanwhile, resides in rxkad_verify_packet_1(). The process decrypts RxRPC payloads using a single-block process. Splice-pinned pages become both a source and destination. That, paired with the decryption key being freely extracted using the add_key (rxrpc), allows an attacker to rewrite contents in memory.
Either exploit used separately is unreliable. Some Ubuntu configurations use AppArmor to prevent untrusted users from creating namespace contents. That, in turn, neutralizes the ESP technique. Most other distributions by default don’t run rxrpc.ko, which neutralizes the RxRPC arm. When chained together, however, the two exploits allow attackers to obtain root on every major distribution Kim tested. Once the exploits run, attackers can use SSH access, web-shell execution, container escapes, or compromise low-privilege accounts.
“Dirty Frag is notable because it introduces multiple kernel attack paths involving rxrpc and esp/xfrm networking components to improve exploitation reliability,” Microsoft researchers wrote. “Rather than relying on narrow timing windows or unstable corruption conditions often associated with Linux local privilege escalation exploits, Dirty Frag appears designed to increase consistency across vulnerable environments.”
Researchers at Google-owned Wiz said exploits will be less likely to break out of hardened containerized environments such as Kubernets with default security settings in place. “However, the risk remains significant for virtual machines or less restricted environments.”
The best response for anyone using Linux is to install patches immediately. While fixes likely require a reboot, protection from a threat as severe as Dirty Frag outweighs the cost of disruptions. Anyone who can’t install immediately should follow the mitigation steps laid out in the posts linked above. Additional guidance can be found here.
This story originally appeared in The Algorithm, our weekly newsletter on AI. To get stories like this in your inbox first, sign up here.
A few months before he was awarded the Nobel Prize in economics in 2024, Daron Acemoglu published a paper that earned him few fans in Silicon Valley. Contrary to what Big Tech CEOs had been promising—an overhaul of all white-collar work—Acemoglu estimated that AI would give only a small boost to US productivity and would not obviate the need for human work. It’s okay at automating certain tasks, he wrote, but some jobs will be perfectly fine.
Two years later, Acemoglu’s measured take has not caught on. Chatter about an AI jobs apocalypse pops up everywhere from Senator Bernie Sanders’s rallies to conversations I overhear in line at the grocery store. Some previously skeptical economists have gotten more open to the idea that something seismic could be coming with AI. A California gubernatorial candidate said last week that he wants to tax corporate AI use and pay victims of “AI-driven layoffs.”
On the one hand, the data is still on Acemoglu’s side; studies repeatedly find that AI is not affecting employment rates or layoffs. But the technology has advanced quite a bit since his cautious predictions. I spoke with him to understand if any of the latest developments in AI have changed his thesis, and to find out what does worry him these days if not imminent AGI.
One of the biggest technical leaps in AI since Acemoglu’s paper has been agentic AI, or tools that can go beyond chatbots and operate on their own to complete the goal you give them. Because they can work independently rather than just answering questions, companies are increasingly pitching agents as a one-to-many replacement for human workers.
“I think that’s just a losing proposition,” Acemoglu says. He thinks agents are better thought of as tools to augment particular pieces of someone’s work than something malleable enough to handle a person’s whole job.
One reason has to do with all the various tasks that go into a job, something Acemoglu has been researching in his work on AI since 2018. For example, an x-ray technician juggles 30 different tasks, from taking down patient histories to organizing archives of mammogram images. A worker can naturally switch between formats, databases, and working styles to do this, Acemoglu says, but how many individual tools or protocols would an AI require to do the same?
Whether or not agents will supercharge AI’s impact on jobs will come down to whether they can eventually handle the orchestration between tasks that humans do naturally. AI companies are in heated competition to prove that their AI agents can work independently for ever longer periods without making mistakes, sometimes exaggerating the results—but Acemoglu says many jobs will be spared from an AI takeover if agents can’t fluidly switch between tasks.
For years Big Tech has been offering staggering salaries to recruit AI researchers. But I asked Acemoglu about a different hiring spree I’ve noticed: AI companies are all building in-house economics teams.
OpenAI hired Ronnie Chatterji from Duke University in 2024 to be its chief economist and announced last year that Chatterji will work with Jason Furman—Harvard economist and former advisor to Barack Obama—to research AI and jobs. Anthropic has convened a group of 10 leading economists to do similar work. And just last week, Google DeepMind announced it had hired Alex Imas, an economist from the University of Chicago, to be its “director of AGI economics.”
Acemoglu has noticed colleagues getting snatched up for these roles too. “It makes sense,” he says: AI companies are well aware that public skepticism about AI, in large part due to job concerns, is growing. And they have strong incentives to shape the economic narrative around their technology (consider OpenAI’s latest proposal for a new era of industrial policy).
“What I hope we won’t get,” Acemoglu says, “is that they’re interested in economists just to further their viewpoints or further the hype.” That tension hangs over the emerging field of “AI economics”; it’s concerning that some of the most influential research about AI’s impact on work may increasingly come from the companies with the most to gain from favorable conclusions.
I don’t think of AI as hard to use; most of us interact with it via chatbots that use plain language. But Acemoglu says we should consider how it compares with the sort of software that kicked off earlier tech transformations, like PowerPoint for slide decks and Word for documents.
“Anybody could install these on their computer and get them to do the things that they want them to do,” he says. They spread accordingly.
“We have not seen the development of apps based on AI that have the same usability,” he says. Even if anyone can chat with an AI model, it tends to take a while for the average worker to get practical and productive use out of it. That’s part of the reason why AI has not yet shown any seismic impact on the job market or the economy. One of the key signals Acemoglu is watching, then, is the creation of apps that make AI easier to use.
But he acknowledges that for a while, we’re going to see all sorts of conflicting evidence about AI: anecdotes that college grads are finding the job market worse and worse, but no noticeable effect of AI on productivity, for example. “There’s a huge amount of uncertainty,” he says. And that’s the most telling thing about the AI economy right now: the certainty of the rhetoric alongside the uncertainty of everything else.
Microsoft is using dynamic CPU scaling to speed up the Start menu, app opening times, and more.
Microsoft is using dynamic CPU scaling to speed up the Start menu, app opening times, and more.
Microsoft is currently testing a new speed boost feature in Windows 11 that is designed to improve app launch times and make things like the Start menu feel more responsive. The feature, which is reportedly called “Low Latency Profile,” will ramp up CPU frequency in short bursts to improve the speed of menus, flyouts, apps, and more — much like how macOS handles similar tasks.
Windows 11 testers have been trying out the new unannounced feature over the past week, and noticing significant speed improvements launching File Explorer or the Start menu, as well as apps like Outlook, the Microsoft Store, and Paint.
Windows Central reports that this new boost mode can result in up to 40 percent faster app times for Microsoft’s own apps, and up to 70 percent faster for the Start menu and context menus throughout Windows 11. While the early results look promising, some online commentators have criticized Microsoft for using CPU bursts to improve its operating system, drawing the attention of a Microsoft executive.
Scott Hanselman, vice president of technical staff for CoreAI, GitHub, and Windows, defended Microsoft’s speed boost changes to Windows 11 over the weekend, pointing out in a post on X that “your smartphone already does this” and that Microsoft isn’t cheating by boosting CPU clocks temporarily. Microsoft is using a common practice, used by macOS and Linux, to dynamically scale a CPU to prioritize interactive tasks. “Apple does this and y’all love it,” said Hanselman. “Let Windows cook.”
Microsoft’s latest speed boost to Windows 11 is part of sweeping changes coming to the operating system to improve performance, reliability, and user experience. Microsoft has also started removing “unnecessary” Copilot buttons from Windows 11, and made Windows Update a lot less annoying.
Circle Internet Group has raised $222 million in the presale of Arc, the native token of its new blockchain, as the company looks to expand beyond its core business of issuing the USDC stablecoin, CNBC has learned.
Andreessen Horowitz served as the lead investor in the raise with a $75 million investment. Other investors include BlackRock, Apollo Funds, New York Stock Exchange parent Intercontinental Exchange, SBI Group, Janus Henderson Investors, Standard Chartered Ventures, General Catalyst, Marshall Wace, ARK Invest, IDG Capital, Haun Ventures and the crypto exchange and CoinDesk owner Bullish.
The raise gives Arc a fully diluted network valuation of $3 billion.
“[Blockchain] infrastructure is becoming as important as mobile operating systems or cloud platforms,” Circle CEO Jeremy Allaire told CNBC in an exclusive interview. “We want to build an operating system that has many, many stakeholders in it … major companies who are running the infrastructure with us and who ultimately help to govern it.”
“We’re becoming a broader internet platform company,” Allaire added. “We’re entering the operating system business and we’re doing it by building this multi-stakeholder distributed model with a token, with a distributed network. But it is an operating system business. And we’re also getting into the apps business.”
Arc is a public blockchain designed for institutional finance. Allaire emphasized it’s about more than stablecoins and payments — noting it can “run the actual economy.”
“The economy is not just representations of values, it’s every contract that undergirds those financial relationships … the systems of governance that we use to govern all these economic institutions,” Allaire said.
As a 25% stakeholder in Arc’s initial supply of 10 billion tokens, Circle can participate in operating validator infrastructure, generating new fee revenue and earning staking income. The majority of the tokens, 60%, will go to participants who build on, use and contribute to the Arc network. The remaining 15% will be allocated to a long-term reserve.
Investors should follow transactions, asset issuance and success on the network found by the developer community, Allaire said.
He added that the economy is becoming increasingly machine-operated, with AI agents handling more of the operational and contractual work currently managed by humans.
“We’re entering this era where software machines will power the economic system,” he said. “Software will do most of the work — that is what AI agents represent.”
The company also unveiled a set of services and tools designed to help developers build AI agents that can manage transactions, access online services and make payments using USDC.
Circle’s ambitions with Arc reflect the existential shift other crypto firms are facing: They need to evolve beyond the businesses that were built in crypto’s early days around the speculative cycles of cryptocurrencies, and toward more durable businesses with steadier and more diversified revenue.
“While USDC has become the trusted digital dollar for banks, corporations, and financial
institutions seeking the speed of crypto without its volatility, there remains a problem. The internet infrastructure which USDC runs on today wasn’t built with big institutions in mind. It was built for individuals and crypto enthusiasts. That’s where Arc comes in,” a16z crypto wrote in a blog post Monday morning.
If Arc is successful, it could allow Circle to own more of the infrastructure its flagship USDC stablecoin runs on. Today, USDC depends heavily on networks like Ethereum and Solana for settlement and distribution partners like Coinbase.
The initiative is as much about playing defense as it is about growth. While regulation supporting stablecoins legitimizes them – including the GENIUS Act signed into law last year and the STABLE Act, which is set to get an initial vote this week in the Senate Banking Committee – some investors worry banks and fintechs may launch their own competing dollar tokens, removing the need for a third-party issuer.
Circle is the first publicly listed company to conduct a token presale – an early sale of digital tokens before a blockchain project officially launches.
Crypto companies like token sales for their ability to raise large amounts of capital and build an early user community. They are frequently likened to IPOs since both represent public-facing fundraising mechanisms that result in a transferable financial interest.
Also known as “initial coin offerings,” or ICOs, token sales became notorious for their role in fueling the 2017 crypto peak – when the market grew so quickly that projects launched with little oversight, leading to some high-profile failures and scams.
The landscape has shifted significantly since then. Under the Trump administration’s more crypto-friendly regulatory posture, the Securities and Exchange Commission is increasingly focused on establishing frameworks for compliant tokenized securities and on-chain capital formation, creating conditions that could encourage the return of ICO-style fundraising in a more mature and sustainable structure.
“It is a major shift in how stakeholders can participate in the growth of networks,” Allaire said. “Every company in the world, over time, will be tokenized, meaning your shares will be tokens … [and] you will use digital tokens as mechanisms of engagement with your customers and stakeholders.”
Iran’s semi-official Tasnim news agency said Tehran’s proposal, sent via Pakistan, which has served as a mediator, included an immediate end to the war on all fronts, a halt to the US naval blockade of Iranian ports and guarantees of no further attacks on Iran.
