FlashFeed

Full text

The network has become the nervous system of any organization running AI at scale. A single distributed training run can chew through thousands of GPUs for weeks, and one congested uplink can slash throughput by more than 30 percent. Plenty of datacenter networks aren't keeping up now that AI sits at the center of the workload. The traffic itself has changed. China’s daily AI token consumption jumped from roughly 100 billion a day at the start of 2024 to more than 30 trillion by mid-2025. That's a 300-fold rise in 18 months, by official count. Machines do more of the talking now, too: bots and agents make up 51 percent of internet traffic, outnumbering humans for the first time in a decade. These pressures are forcing companies to change the way they think about datacenter network fabrics. What is a network fabric? A fabric is the dense mesh of switches and links that carries traffic between servers (east-west) and in and out of the datacenter (north-south). AI training traffic moves east-west: GPUs in a training cluster swap enormous volumes of data, and a single lossy link can stall the whole job. But AI inference generates new flows north-south and over the WAN. Why does automating it need a model of itself? This is where most automation falls down. You can script a change onto a fabric, but a script has no idea how the parts relate, so it can fix one switch and quietly break its neighbor. The answer is to give the network a model of itself. HPE Networking Apstra Datacenter Director, for instance, keeps a live graph of every device and the links and policies between them. Intent-based networking runs on top: an operator declares the outcome they want, and the system writes the configuration, checks it against the graph before anything ships, then continuously verifies that the running network still matches. A deterministic model can point at the real root cause instead of burying the operator in alarms. How does the fabric catch trouble before users do? Telemetry is only part of the solution. Add in AIOps to optimize the experience of the network operator and the application end user. Instead of asking whether a switch is up, newer systems ask whether users are getting a good experience, then trace a slowdown to the exact port or optic behind it. HPE Mist Networking Datacenter Assurance scores fabric health on that basis. Operators query the Marvis AI Assistant, which has evolved into a reasoning agent, providing a simple way to interact with the network. Predictive models go earlier still: by watching voltage, temperature, laser readings and CRC error counts, they flag a failing optic before it drops a link. The operator hears about it first, not the user. What about security? None of this holds if security is an afterthought. AI pipelines move sensitive data east-west, between servers, where a perimeter firewall never looks. So segmentation moves inside the fabric. Workloads are walled off from each other and every flow is inspected, not just the ones crossing the edge. What does it mean for your team? Get the rebuild right and fewer change windows die on a typo, while root-cause analysis stops being an archaeology dig. Network engineers don’t vanish, but they do evolve from herding switches in the CLI to designing the fabric,handling the exceptions automation throws up, and getting time back to work on strategic initiatives important to the CIO. The datacenter is the substrate the rest of the AI stack runs on; build it to run itself and everything above it gets easier. Sponsored by HPE.

Full text

Qualcomm Dragonfly AI200 AI accelerator rack is the first of multiple releases planned by the chip designer as it aims to score wins in the data center segment

The upcoming Dragonfly AI250 accelerator leverages its proprietary High Bandwidth Compute (HBC) to offer a theoretical 18x the amount of bandwidth of its sibling

Qualcomm's push comes amid an increasingly lucrative datacenter market grappling with memory shortages

It is no secret that the modern AI server ecosystem is dominated by Nvidia in most countries, even as China increasingly leans towards Huawei as its own home-grown provider for similar solutions.

Qualcomm may not be one of the first companies that come to mind when you think about AI data centers or the chips housed inside them, with many investors feeling it has missed the boat altogether in the server segment.

Qualcomm's recent Investor Day 2026 event was a reminder that it is not only still in the game but also has ambitions to carve out a large piece of an ever-increasing pie by taking a different route than most of its HBM-leveraging competitors.

An alternate ecosystem to Nvidia's industry standards?

Much of Qualcomm's Investor Day event focused on its plans to become a sizable player in the AI data center market, which is currently dominated by OEMs deploying a mix of Nvidia and AMD accelerators alongside custom silicon (ASIC) offerings from Google, Meta, Microsoft, and even Amazon's AWS.

It aims to do so by differentiating itself from the competition, relying on its own area of expertise to carve out an edge: efficient Low-Power Double Data Rate (LPDDR) memory stacked in a 3D array above its AI accelerators to drive the next generation of AI inference workloads.

The near-memory compute architecture isn't exactly a new play in a market teeming with similar approaches , but the numbers are hard to argue with when it comes to Qualcomm's offerings.

Qualcomm's upcoming Dragonfly AI200 rack delivers 43 TB of LPDDR5X capacity and 414 TB/s of memory bandwidth per rack, built from accelerator cards each carrying 768 GB of LPDDR5X, which makes it an interesting offering, but much of the focus that hyperscalers will have will be on its Dragonfly AI250 sibling that incorporates High Bandwidth Compute (HBC) under the hood.

While it offers the same memory capacity per rack, its ability to leverage memory at up to 18x its sibling's bandwidth yields a theoretical peak memory bandwidth of up to 7.4 PB/s per rack, a far cry from the AI200's 0.4 PB/s.

The Dragonfly is positioned as an inference-centric accelerator for a reason; however, HBM is still better suited to certain tasks, such as training models rather than inference, making it the memory of choice for Nvidia's Blackwell and upcoming Rubin GPUs, as well as AMD's Instinct offerings.

With that being said, Qualcomm's solution is intriguing, even if the numbers are for specific use cases and its ability to court Hyperscaler giants such as Microsoft and Meta tends to indicate that it has a potential win, at least on paper, as AI datacenters continue to increase focus on inference-centric solutions to deploy their increasingly complex models to wider audiences.

Full text

The Large Hadron Collider (LHC), the world's largest and most powerful atom smasher, has entered a planned four-year shutdown that will upgrade it to its most capable form yet.

The particle accelerator was switched off Monday (June 29) and is scheduled to come back online in 2030 as the High-Luminosity Large Hadron Collider (HiLumi LHC), with improvements that will allow it to smash together roughly 10 times more particles than its original design. That data could help spark new discoveries in fundamental physics and shed light on the nature of dark matter , antimatter and the early universe.

"This is a very important moment," HiLumi LHC project chief Markus Zerlauth told Agence France-Presse . "From Monday, we will be entering a new phase."

Since its first successful proton collision in 2009, the LHC has allowed physicists to test theories about particle physics and the Standard Model of the subatomic world. It was essential to the discovery of the Higgs boson in 2012, which helped explain how tiny fundamental particles acquire mass. The collider covers a 17-mile (27 kilometers) loop at the border between France and Switzerland near Geneva.

The current shutdown is the third long-term, planned pause in the collider's operations. The first, a two-year shutdown beginning in 2013, consolidated connections between superconducting magnets and boosted the energy of the colliding proton beams. A second pause, from 2018 to 2022, involved a series of upgrades, replacements and preventive maintenance.

In the current period of dormancy, dubbed Long Shutdown 3 (LS3), specialists will install upgrades to boost the collider's luminosity by a factor of 10. That will increase the number of particle collisions, roughly tripling the number of times particles will smack into each other compared with the existing setup. Once the final version of the particle accelerator is online, it will run until the end of its operational lifespan in the 2040s and will be earmarked for replacement by a new, higher-energy particle accelerator in the years that follow.

"It really is an opportunity to explore the universe in a way we haven't done before," Mark Thomson , director general of CERN (the European Organization for Nuclear Research), where the LHC is housed, told New Scientist .

Civil engineers work on upgrades to turn the Large Hadron Collider into the High Luminosity Large Hadron Collider, significantly increasing the facility’s rate of particle collisions. (Image credit: Samuel Joseph Hertzog, CERN) That jump means that experiments will produce much more data, which will allow scientists to study known phenomena, such as the Higgs boson, in more detail and increase the odds of observing rare events. For example, the HiLumi LHC is expected to produce about 380 million Higgs bosons over its lifetime of a decade or so, compared with the 55 million it's made to date. The data could help scientists solve problems with the Standard Model, which currently doesn't incorporate dark matter or dark energy , the primary forms of mass and energy in the universe.

Related stories

Large Hadron Collider finds 1st evidence of the heaviest antimatter particle yet

'Beauty' particle discovered at world's largest atom smasher could unlock new physics

'X particle' from the dawn of time detected inside the Large Hadron Collider

"The LS3 represents a huge and complex logistical and engineering undertaking," Jean-Philippe Tock , head of the LS3 coordination team, said in a statement . "In the LHC alone, 1.2 km [0.75 miles] of magnets and components will be removed and replaced with new equipment, and across the whole complex, dozens of projects are planned, involving thousands of engineers, physicists, technicians and support personnel."

While the LHC won't be smashing any particles together during the shutdown period, researchers will continue to analyze data already collected during experiments during the prior operational window.

Though the LHC's primary purpose is fundamental physics research, technologies used to upgrade the collider could find their way into everyday life. For example, some instruments and techniques originally developed at CERN are now being used in medical imaging, sensors and art restoration.

Full text

Meta just released a new ad for its creeper glasses. In the video, Kylie Jenner, the new face of the glasses called Starfire, goes through a day-in-the-life style video from her point of view. Mostly, she’s led around her own house in a haze by various vendors and assistants. Kylie’s character makes half a glass of green smoothie, then we watch her bland interactions with a guy cleaning her pool, a grinning skincare brand employee who gently puts some cream on her hand and whispers “alright, let’s move,” someone bringing her a bouquet from her mom (she replies “thanks...”) and people moving a huge weird sculpture around her cavernous home.

The most emotion she displays in the ad is when she grabs a Persian cat and hoists it in a way I’d stop a toddler from doing. In a jarring transition away from the cat and the movers, we see her start inexplicably grabbing black spray paint from her massive closet (???) and jumping in an unbranded black SUV, then speeding to a billboard of her own face. In another unsettling transition that would work in an Ari Aster horror movie, the perspective is no longer from her own eyes, but from about 30 yards behind the car. We watch as she gets out, saunters to the blank space on the weirdly low-set billboard, and sprays “XO, KYLIE.”

Meta has endured years of brand crises with its smart glasses. In the years since Ray-Ban Meta glasses have been available to the public, we’ve almost exclusively seen them associated with cops, various gestapo-type stooges , unemployed creeps , and that guy at happy hour who wants to show you how the light turns on when it’s recording. During that time, 404 Media has documented all of this, and in the course of that reporting, heard time and time again from Meta that the glasses are NOT that creepy and definitely NOT cop-glasses.

When Jason broke the story about a Customs and Border Control (CBP) agent wearing Ray-Ban Metas to a raid , a Meta spokesperson asked 404 Media a series of questions about the framing of the article, stressing that Meta does not have a contract with CBP. The spokesperson asked why 404 mentioned Meta in this story — again, a story about Meta’s glasses seen on the face of an immigration officer. “I’m curious if you can explain why it is Meta will be mentioned by name in this piece when in previous 404 reporting regarding ICE facial recognition app and follow up reporting the term ‘smartphones’ or ‘phone’ is used despite ICE agents clearly using Apple iPhones and Android devices,” they said. I actually can’t parse this statement to this day. But Meta has seemed pretty stressed about the image of its smart glasses and slick Ray-Ban partnership for a long time. For them to sell to a mass market, the company desperately needed them to stop being associated with loser behavior, fast.

Each of us here have wondered aloud at various points in the last few years about whether Ray-Bans, the once-cool, hipster-coded, mid-luxury sunglasses brand, would keep tolerating this slow image suicide by association with Meta’s depraved quest for a more complete consumerist surveillance state. In October, Emanuel wrote : “I wonder how long Ray-Ban will want to be associated with this product, and if it’s going to tank the reputation of one of the most iconic fashion items in the world before it pulls out.” Ray-Ban hasn’t pulled out from its other offerings, as far as I can tell, and those glasses have reportedly made Meta and Ray-Ban’s parent company a boatload of money. There are many different brands and types of smart glasses being sold by Meta now, with other glasses brands. But it’s noteworthy that for this new It-Girl iteration, Ray-Ban is not along for the ride.

So: Meta’s in a coolness crisis, cops and stalkers are tainting the concept worse than Google Glassholes ever did, they had one chance. In comes Kylie. With the Starfire glasses, we have a few decades-long arcs coming full circle.

Podcast: Why Are DHS Agents Wearing Meta Ray-Bans?

CBP’s use of Meta Ray-Bans; the bargain that voice actors are having to make with AI; and how Flock tech is being essentially hacked into by the DEA.

Meta’s products have been accused and found guilty of damaging young people’s self esteem, especially girls’ confidence and mental health, year after year to the point that it can only be described as a business strategy. Just this week, the company lost its bid to dismiss a lawsuit brought by 29 attorneys general who claim that “research has shown that children's use of ​Facebook and Instagram could lead to depression, anxiety, insomnia, interference with education and daily life, and ​self-harm including suicide,” Reuters reported . Meta has built the world’s most profitable and popular panopticon; Smart glasses that turn everyone into an influencer, narc, and walking data collection apparatus are the logical next step. Meta’s original mistake was aiming for middle aged men, the primary market for Ray-Bans. No one thinks that demographic defines coolness; women aged 18-24, however, have been right there all along. They just had to make the glasses slightly less nerdy looking, and put them on the world’s most recognizable, aspirational, brand-safe face.

They had a lot of these to choose from, and I’m not saying other influencers, or the celebrity influencer pool in general, is pure of heart apart from Kylie Jenner and the Kardashian extended universe. And the Kardashian name has its own poisonous brand associations that would rule Kim or Kylie’s other sisters out of the running for Meta’s new glasses product spokesperson. The Kardashians, while controlling and at times subverting the ways reality television and our parasocial culture prey on women’s bodies, could shake hands with Meta’s influence on girls’ body image. They’ve dictated the course of beauty standards unsubtly and effectively for years, to the point of becoming an endless research topic for social scientists, psychologists and anthropologists. Being like Kylie is seen as being cool, effortless, skinny, rich, unopinionated and unproblematic. And lonely.

The lifestyle in Meta’s latest glasses ad portrays an unattainably wealthy and kind of bored existence where Kylie never sees another person who isn’t on her payroll. These Meta glasses are anti-social from the jump. If all we have is this ad to judge the product, they’re not about capturing memories with your friends at the bar or a party. They’re about packaging one’s life and then viewing it as an out-of-body experience, like you’re already dead. As a piece of media, the ad is the inverse of Chris Samra’s “stealth” Waves smart glasses video, which portrayed the social life of a total fuckhead .

Contrast these with projects like Jenny Zhang’s Computer Angel , a hair clip camera prototype that records her life from her point of view. She records nights out, passes it to other people, takes it off and sticks it on railings to record herself dancing. The important part is that it looks like a camera on the wearer’s head, and isn’t trying to disappear. It reminds me of the very early days of lifecasting, the early 2000s genre of online live stream pioneered by Jennifer Ringley’s Jennicam and also countless early webcam models. I won’t romanticize that time, however: Justin Kan’s Justin.tv , a 2007 experiment in livestreaming, famously caught him getting swatted in his San Francisco apartment while on stream and was funded by Y Combinator and sponsored by companies like Zipcar and Bawls energy drinks (RIP). It grew into what we now know as Twitch.

Maybe it doesn’t matter that the Starfire glasses, or the Ray-Ban Metas, are an attempt to hide the camera from the subject. We assume the camera is everywhere now, anyway, and we all act like it. The primary emotions I feel watching the new ad are sadness and a vague, quiet discontentment. Meanwhile, Instagram is already full of young people, primarily women, unboxing, reviewing, and promoting the glasses. With Kylie’s star power behind these glasses, we risk losing our grip on the last shred of privacy, autonomy, and control of our own images online.