As the 2026 Olympic Winter Games begin today, news articles are swelling with juicy claims that male ski jumpers have injected their penises with fillers to gain a flight advantage.

As the rumor goes, having a bigger bulge on a required 3D body scan taken in the pre-season could earn jumpers extra centimeters of material in their jumpsuits—and a suit's larger nether regions provide more surface area to glide to the gold. Even a small increase can make a satisfying difference in this sport. A 2025 simulation-based study published in the journal Frontiers in Sports and Active Living suggested that every 2 cm of extra fabric in a ski jumpsuit could increase drag by about 4 percent and increase lift by about 5 percent. On a jump, that extra 2 cm of fabric amounts to an extra 5.8 meters, the simulations found.

Elite ski jumpers are aware of the advantage and have already crotch-rocketed to scandal with related schemes. Last year, two Norwegian Olympic medalists, Marius Lindvik and Johann Andre Forfang, and three of their team officials were charged with cheating after an anonymous video showed the head coach and suit technician illegally restitching the crotch area of the two jumpers' suits to make them larger. The jumpers received a three-month suspension, while the head coach, an assistant coach, and the technician faced a harsher 18-month ban.

Injections are alleged to be a new, more drastic strategy. Rumors that jumpers were internally padding their peckers first came to light in January, when German newspaper Bild reported that there were "whispers" of jumpers using hyaluronic acid or possibly paraffin injections.

Bild quoted Dr. Kamran Karim, a specialist at Maria-Hilf Hospital in Krefeld, Germany, as saying (translated): "There is the possibility of obtaining a temporary, optical thickening of the penis with the injection of paraffin or hyaluronic acid. ... Such an injection, however, is not medically indicated and involves risks."

On Thursday, the injection claims sprang up again at a press conference in which journalists asked officials of the World Anti-Doping Agency about the claims. The agency's director general said they were not aware of any claims and that non-doping means of enhancing performance are not in their purview. But according to the BBC, WADA President Witold Banka, who is from Poland, was "clearly entertained" by the questions, responding: "Ski jumping is very popular in Poland, so I promise you I'm going to look at it."

Bruno Sassi, the communications director for FIS, the international ski and snowboard federation, seemed less amused, telling the BBC, "There has never been any indication, let alone evidence, that any competitor has ever made use of a hyaluronic acid injection to attempt to gain a competitive advantage."

But what if they did? Here's what we know about hyaluronic acid and paraffin for penis augmentation.

Hyaluronic acid

While some news outlets have played up the "acid" part of its name, hyaluronic acid is not some nefarious flesh-melting hazard. It's a common filler used for various clinical purposes.

Hyaluronic acid is a polysaccharide that is naturally found in a wide variety of tissues in the human body, including the skin, eyes, and connective tissue. It's a chief component of the extracellular matrix. It attracts water molecules to itself, creating volume that can provide structural support. In a pure form, it has no tissue or even species specificity and therefore is considered to have little risk of sparking immune responses.

As such, hyaluronic acid gel fillers are used in a variety of medical procedures, with approval from the Food and Drug Administration. Hyaluronic acid (HA) fillers are injected into joints, particularly knees, to relieve pain from mild to moderate arthritis, which can decrease the natural amount of HA in joints. Age also decreases natural levels of HA, and one of the main uses of HA fillers is for cosmetic purposes—plumping lips and cheeks, and minimizing the appearance of wrinkles and fine lines in the face. HA fillers can also be used inside the eye in a variety of surgeries, including cataract extraction and corneal transplants. It can also be used topically for wound care and to relieve skin pain and itching.

For these purposes, the most common adverse effects are pain, bruising, redness, itching, and swelling, which usually last for just a few days. In extremely rare cases, there can be more serious side effects from injections, such as bacterial infections, tissue death (from blocked blood flow), and a granulomatous foreign body reaction, in which the immune system tries to clear a foreign substance, such as bacterial impurities, leading to a collection of immune cells.

A notable feature of HA fillers is that they have something of an antidote. For procedures that go wrong for whatever reason, clinicians can use hyaluronidase, an enzyme that breaks down hyaluronic acid, to dissolve the filler. Hyaluronidase is then rapidly deactivated and degraded in the body. But even without any dissolving, HA fillers offer only temporary effects, lasting between eight weeks and up to six months, for the various established uses.

Plumping penises

What is not on the list of established clinical purposes for HA fillers is penis augmentation. But that's not from a lack of trying. Studies have looked at the possibility of using HA fillers for adding girth and length—and some of have reported positive results. But so far, there's a lack of evidence to support safety and efficacy. Still, that hasn't stopped cosmetic clinics around the world from offering it anyway.

In a 2022 study in the Asian Journal of Andrology, researchers in China followed 38 patients who had HA filler injections, which seemed to provide desired results. One month after injections, the men saw an average increase of 3.4 cm in girth and 2.5 cm in length when flaccid. After a year, the measurements had decreased, but were still an increase of 2.4 cm in girth and 1.65 cm in length. Erect girth ended at an average increase of 0.8 cm. The researchers reported that only three patients had complications: two had swelling, and one had bleeding under the skin. All three cases resolved on their own.

However, not all reports are so rosy. In a 2021 case study in BMC Urology, researchers in Australia reported the experience of a 31-year-old man who received penis HA fillers at a cosmetic clinic. He developed a severe penis infection soon after that led to sepsis and multi-organ failure, landing him in the intensive care unit. Doctors ended up surgically removing pus-stained filler from his penis, which laboratory tests found teeming with Streptococcus pyogenes.

Another case study in 2021, published in a Japanese urology journal, reported that a 65-year-old man had to have part of his penis surgically removed after having HA fillers injected into the head of his penis. When he arrived at the hospital, black necrotic lesions and ulcers were readily visible.

Paraffin

While things can go wrong with HA fillers, despite their legitimate clinical uses, the outcomes for paraffin injections are much darker. Amid the Olympics scandal, most of the focus has been on hyaluronic acid. But the initial Bild article introduced speculation that skiers could also use paraffin injections to jack up their junk, which would be a terrible idea.

The use of injections of mineral oil or paraffin wax for cosmetic procedures dates back to at least 1899, when it was almost immediately found to have horrifying results. The injections lead to what's called paraffinoma, in which the body tries to encapsulate and sequester the oil it can't degrade. The result is tissue that looks like swiss cheese—with large spaces filled with oil. Over time, the tissue becomes thickened and scarred and can block lymphatic drainage. From the outside, it results in disfiguring and debilitating lesions. Some people initially ignored these disturbing results because, in some cases, the lesions can take years to develop.

While such injections have largely been abandoned, cases occasionally crop up. That includes a 2002 report of a 64-year-old Michigan man who self-injected his penis with mineral oil to increase its girth. He used a series of shots over the course of 18 months. Two years after the last shot, he went to urologists because of an increasing mass in his penis that was causing erectile dysfunction and making it difficult to urinate. He reluctantly admitted to the oil injections only after his doctors suspected cancer. In the end, the doctors had to surgically remove the mass of oily nodules from all around his penis and circumcise him.

Whatever is going on in the trousers of Olympic ski jumpers, let's hope it doesn't involve paraffin.

Read full article

Comments

Amid a push toward AI agents, with both Anthropic and OpenAI shipping multi-agent tools this week, Anthropic is more than ready to show off some of its more daring AI coding experiments. But as usual with claims of AI-related achievement, you'll find some key caveats ahead.

On Thursday, Anthropic researcher Nicholas Carlini published a blog post describing how he set 16 instances of the company's Claude Opus 4.6 AI model loose on a shared codebase with minimal supervision, tasking them with building a C compiler from scratch.

Over two weeks and nearly 2,000 Claude Code sessions costing about $20,000 in API fees, the AI model agents reportedly produced a 100,000-line Rust-based compiler capable of building a bootable Linux 6.9 kernel on x86, ARM, and RISC-V architectures.

Carlini, a research scientist on Anthropic's Safeguards team who previously spent seven years at Google Brain and DeepMind, used a new feature launched with Claude Opus 4.6 called "agent teams." In practice, each Claude instance ran inside its own Docker container, cloning a shared Git repository, claiming tasks by writing lock files, then pushing completed code back upstream. No orchestration agent directed traffic. Each instance independently identified whatever problem seemed most obvious to work on next and started solving it. When merge conflicts arose, the AI model instances resolved them on their own.

The resulting compiler, which Anthropic has released on GitHub, can compile a range of major open source projects, including PostgreSQL, SQLite, Redis, FFmpeg, and QEMU. It achieved a 99 percent pass rate on the GCC torture test suite and, in what Carlini called "the developer's ultimate litmus test," compiled and ran Doom.

It's worth noting that a C compiler is a near-ideal task for semi-autonomous AI model coding: The specification is decades old and well-defined, comprehensive test suites already exist, and there's a known-good reference compiler to check against. Most real-world software projects have none of these advantages. The hard part of most development isn't writing code that passes tests; it's figuring out what the tests should be in the first place.

The compiler also has clear limitations that Carlini was upfront about. It lacks a 16-bit x86 backend needed to boot Linux from real mode, so it calls out to GCC for that step. Its own assembler and linker remain buggy. Even with all optimizations enabled, it produces less efficient code than GCC running with all optimizations disabled. And the Rust code quality, while functional, does not approach what an expert Rust programmer would produce. "The resulting compiler has nearly reached the limits of Opus's abilities," Carlini wrote. "I tried (hard!) to fix several of the above limitations but wasn't fully successful. New features and bugfixes frequently broke existing functionality."

Those limitations may actually be more informative than the successes. Carlini reports that toward the end of the project, fixing bugs and adding features "frequently broke existing functionality," a pattern familiar to anyone who has watched a codebase grow beyond the point where any contributor fully understands it.

And that limitation is even more common when dealing with AI coding agents, which lose coherence over time. The model hit this wall at around 100,000 lines, which suggests a practical ceiling for autonomous agentic coding, at least with current models.

The human work behind the automation

Anthropic describes the compiler as a "clean-room implementation" because the agents had no Internet access during development. But that framing is somewhat misleading. The underlying model was trained on enormous quantities of publicly available source code, almost certainly including GCC, Clang, and numerous smaller C compilers. In traditional software development, "clean room" specifically means the implementers have never seen the original code. By that standard, this isn't one.

On Hacker News, the distinction drew sharp debate, reflective of a controversial reception to the news among developers. "It was rather a brute force attempt to decompress fuzzily stored knowledge contained within the network," wrote one commenter.

The $20,000 figure also deserves some context. That number covers only API token costs and excludes the billions spent training the model, the human labor Carlini invested in building the scaffolding, and the decades of work by compiler engineers who created the test suites and reference implementations that made the project possible.

And that scaffolding was not trivial, which makes any claim of "autonomous" work on the C compiler among the AI agents dubious. While the headline result is a compiler written without human pair-programming, much of the real work that made the project function involved designing the environment around the AI model agents rather than writing compiler code directly. Carlini spent considerable effort building test harnesses, continuous integration pipelines, and feedback systems tuned for the specific ways language models fail.

He found, for example, that verbose test output polluted the model's context window, causing it to lose track of what it was doing. To address this, Carlini designed test runners that printed only a few summary lines and logged details to separate files.

He also found that Claude has no sense of time and will spend hours running tests without making progress, so he built a fast mode that samples only 1 percent to 10 percent of test cases. When all 16 agents got stuck trying to fix the same Linux kernel bug simultaneously, he used GCC as a reference oracle, randomly compiling most kernel files with GCC and only a subset with Claude's compiler, so each agent could work on different bugs in different files.

"Claude will work autonomously to solve whatever problem I give it," Carlini wrote. "So it's important that the task verifier is nearly perfect, otherwise Claude will solve the wrong problem."

None of this should obscure what the project actually demonstrates. A year ago, no language model could have produced anything close to a functional multi-architecture compiler, even with this kind of babysitting and an unlimited budget. The methodology of parallel agents coordinating through Git with minimal human supervision is novel, and the engineering tricks Carlini developed to keep the agents productive (context-aware test output, time-boxing, the GCC oracle for parallelization) could potentially represent useful contributions to the wider use of agentic software development tools.

Carlini himself acknowledged feeling conflicted about his own results. "Building this compiler has been some of the most fun I've had recently, but I did not expect this to be anywhere near possible so early in 2026," he wrote. He also raised concerns rooted in his previous career in penetration testing, noting that "the thought of programmers deploying software they've never personally verified is a real concern."

Read full article

Comments

Yesterday afternoon, following the end of trading on Wall Street for the day, Tesla published its financial results for 2025. They weren't particularly good: Profits were almost halved, and revenues declined year on year for the first time in the company's history. The reasons for the company's troubles are myriad. CEO Elon Musk's bankrolling of right-wing politics and promotion of AI-generated revenge porn deepfakes and CSAM has alienated plenty of potential customers. For those who either don't know or don't care about that stuff, there's still the problem of a tiny and aging model lineup, with large question marks over safety and reliability. Soon, that tiny lineup will be even smaller.

The news emerged during Tesla's call with investors last night. As Ars and others have observed, in recent years Musk appears to have grown bored with the prosaic business of running a profitable car company. Silicon Valley stopped finding that stuff sexy years ago, and no other electric vehicle startup has been able to generate a value within an order of magnitude of the amount that Tesla has been determined to be worth by investors.

Musk's attention first turned away from building and selling cars to the goal of autonomous driving, spurred on at the time by splashy headlines garnered by Google spinoff Waymo. Combined with ride-hailing—a huge IPO by Uber took the spotlight off Tesla long enough for it to become a new business focus for the automaker too—Musk told adoring fans and investors that soon their cars would become appreciating assets that earned money for them at night. And as an intermediary, Tesla would take a hefty cut for connecting the rider and the ridee.

Autonomy remains a stated goal for Tesla: last week it announced it's moving to a purely subscription-based approach for its FSD partially automated driving assist for all new Teslas purchased from mid-February. At the same time, it's doing away with Autopilot entirely, it's a less-capable, partially automated driving assist. The company's goal is more ambitious, involving selling two-seat robotaxis that lack even a steering wheel or pedals, but only once it has proven the technology, something it's currently attempting with little success on the streets of Austin, Texas.

But even robotaxis have lost some of their shine. For Musk, the real action is with his humanoid robot. These will sell in the billions, Musk has claimed, adding up to $20 trillion to Tesla's market capitalization at some point in the future. And he needs factory space to build these Optimus robots, which Tesla claims will go on sale in 2027. And that means an end to the Model S sedan and Model X SUV.

Once a world-beater

The Model S wasn't Tesla's first car, but it was the first one it built from scratch. At the time, it was simply revolutionary. EVs from traditional automakers were still firmly in the realm of the compliance car—hasty conversions of internal combustion engine-powered models with the batteries and control electronics crammed wherever possible, plus maybe some aerodynamic smoothing to try to eke out a little more range.

By contrast, the Model S was designed as an EV from the ground up, with a big enough battery to go 265 miles (426 km) on a full charge—something unthinkable in any other EV on sale. It didn't hurt that the car was seriously quick in a straight line and came with infotainment that made anything else on the road seem dated. When Ars tested one in 2013, we were impressed.

Over the years, the Model S got more power and more advanced driver assists—and eventually, a cosmetic facelift. But as rivals responded with vehicles like the Porsche Taycan and Lucid Air, the Model S stagnated rather than being replaced.

Similar neglect was shown to the Model X, the brand's SUV-cum-minivan. The lengthy and troubled gestation for the Model X was a forerunner of the problems Tesla has faced developing each successive product; in this particular case, the "falcon wing" doors, created as an alternative to the minivan's traditional sliding door, proved particularly problematic to get right. Indeed, I still remember being smacked in the head by one at my first introduction to the ungainly people-mover. And yet, compared to the other SUVs on sale in 2016, the Model X still stood out.

A decade later, it's fair to say the Model X still stands out, but like a sore thumb. Its looks never became more gainly, and there is now vast competition for large, luxurious electric SUVs, whether that's from Chinese startups like BYD and Xiaomi, American startups like Rivian and Lucid, or the traditional automakers that now have a handle on electrification.

That has been reflected in the sales. Right-hand drive cars for markets including the UK, Australia, New Zealand, and Japan ended production in 2023. And each quarter, production and sales of the Model S and X slipped more and more. Even lumped together with the poorly selling Cybertruck—which is only offered in the US—these deliveries fell by more than half in Q4 2025, and by 40 percent for the year.

Is there much reason to expect that the development of the Optimus robot will be any smoother than the "development hell" that beset the Model X, 3, Y, and Cybertruck? On last night's call, Musk admitted—contrary to previous claims—that the robots are not doing any useful work at the Tesla factory, and the idea that the company will build 10,000 robots this year seems in conflict with Optimus still being "very much at the early stages" and "still in the R&D phase," to use Musk's own words.

Read full article

Comments

Humans can’t live without lungs. And yet for 48 hours, in a surgical suite at Northwestern University, a 33-year-old man lived with an empty cavity in his chest where his lungs used to be. He was kept alive by a custom-engineered artificial device that represented a desperate last-ditch effort by his doctors. The custom hardware solved a physiological puzzle that has made bilateral pneumonectomy, the removal of both lungs, extremely risky before now.

The artificial lung system was built by the team of Ankit Bharat, a surgeon and researcher at Northwestern. It successfully kept a critically ill patient alive long enough to enable a double lung transplant, temporarily replacing his entire pulmonary system with a synthetic surrogate. The system creates a blueprint for saving people previously considered beyond hope by transplant teams.

Melting lungs

The patient, a once-healthy 33-year-old, arrived at the hospital with Influenza B complicated by a secondary, severe infection of Pseudomonas aeruginosa, a bacterium that in this case proved resistant even to carbapenems—our antibiotics of last resort. This combination of infections triggered acute respiratory distress syndrome (ARDS), a condition where the lungs become so inflamed and fluid-filled that oxygen can no longer reach the blood.

In this case, the infections were necrotizing—the cells in the lungs were dying, turning his lung tissue into a liquid. The surgeons faced a seemingly impossible choice. The patient needed a transplant to survive, but he was in refractory septic shock. His kidneys were shutting down, and his heart was failing to the point where it completely stopped shortly after hospital admission. The doctors had to bring him back with CPR.

He was too sick for a transplant, yet the very organs that needed replacing were the source of the infection fueling his decline. "When the infection is so severe that the lungs are melting, they’re irrecoverably damaged," Bharat explained. "That’s when patients die."

But this patient did not die.

The empty chest problem

To save him, Bharat’s team had to remove the infected lungs, a procedure called a bilateral pneumonectomy, to remove the source of the sepsis. We have machines that can oxygenate the blood. But removing both lungs creates a lethal mechanical problem for the heart.

The human heart is two pumps in one. The right side, called the pulmonary circuit, pumps oxygen-poor blood returning from the body into the lungs, which remove its carbon dioxide and load it with a fresh supply of oxygen. The left side, known as the systemic circuit, receives freshly oxygenated blood and pumps it to the rest of the body. The pulmonary vascular bed, all these miles of tiny vessels inside the lungs, facilitates this gas exchange. But it also acts as a capacitor, absorbing the pressure and volume of the blood ejected by the right ventricle.

If you remove the lungs and simply shut the pulmonary arteries, the right ventricle has nowhere to pump—it would experience an immediate, massive pressure spike, distend like a balloon, and fail within minutes. At the same time, the left side of the heart would have no blood returning to it, leading to a total collapse of blood pressure and systemic circulation.

This is the reason why most double-lung transplants are performed sequentially: The surgeons replace one lung, get it up and running, and then move on to the second. But desperate times require desperate measures.

The last line of defense

In rare cases where both lungs must be removed at once, a patient can still be supported by the Extracorporeal Membrane Oxygenation (ECMO), a mechanical lung that takes blood out of the body, removes the carbon dioxide, adds oxygen, and pumps it back in. The problem is that while ECMO can support a person for up to over a year when their lungs are still in their body, the risks of using it skyrocket when the lungs are removed.

The empty chest cavity creates a void where blood and fluids can pool, leading to huge internal bleeding. The heart, which relies on the physical presence and pressure of the lungs to maintain its proper anatomical position, can flop around or collapse. Finally, circulating blood through complex machinery significantly increases the risk of stroke or clotting.

Because of all these risks, surgeons always treat removing both lungs as the last line of defense, one they can only rely on for a short time—the shorter the better. Bharat and his colleagues, though, had to keep their patient this way until the sepsis had been dealt with.

It wasn’t enough for the team to keep the patient alive without lungs in his body until the transplant arrived. In this lungless state, somehow, they had to improve his health.

Synthetic lungs

To make this happen, Bharat’s team engineered a device they called the “flow-adaptive extracorporeal total artificial lung system" (TAL), a complex circuit designed to mimic the physics of the missing organs. At its core was a pump and an oxygenator borrowed from the standard ECMO setup, but it also used four new components to replace biological functions.

The first was a dual-lumen cannula, essentially a single tube with two separate channels inside. Inserted through the internal jugular vein, this tube acted as the primary drain. It allowed the team to pull deoxygenated blood directly from the right side of the heart, unloading the right ventricle to prevent it from distending.

The second component was something the researchers called a flow-adaptive shunt, which connected the right pulmonary artery back to the right atrium. When the right ventricle pumped out more blood than the external pump could handle, the excess blood would safely recirculate back into the atrium through this low-resistance path, protecting the heart and the surgical staples from pressure spikes. During the 48 hours the patient was supported by the TAL, this shunt self-regulated its flow to anywhere between 1.1 and 6.3 liters per minute, based on the patient's needs.

To ensure the left side of the heart stayed full and active, the team used a device called dual left atrial return. It comprised two 10 mm grafts that returned oxygenated blood from the ECMO artificial lung directly into the left atrium. This, the team said in a paper that describes the hardware, maintained what’s called Starling physiology: the principle that the heart pumps better when it is properly filled. It also prevented blood from stagnating and forming clots inside the heart chambers.

Finally, to prevent the heart from physically shifting and damaging vital vessels in the empty chest cavity, the surgeons used bovine pericardium to reconstruct the heart's protective sac and filled the empty space with tissue expanders and surgical sponges.

Lungless recovery

The results of hooking the patient up to TAL were immediate.

Within hours of the surgery, the patient’s septic shock began to resolve. His lactate levels, a key marker of tissue oxygen starvation, dropped from a dangerous 8.2 mmol/L to a normal level of less than 1.0 mmol/L within 24 hours. The medications used to keep his blood pressure up were discontinued after just 12 hours.

For two days, the patient lived as a human being with no lungs, stabilized by a machine that breathed and buffered his circulation with surgical precision. When donor lungs became available 48 hours later, the patient's body was no longer suffering from sepsis.

He was ready for the transplant, which the team successfully performed. And after putting in the new lungs, they focused on the lungs they had removed.

When lungs die

Conventionally, patients with ARDS do not get transplants because doctors hope that with the right treatment and support, the diseased lungs will eventually heal. But the examination of the infected lungs removed from Bharat’s patient told a different story and helped define the clinical point of no return, when a lung is truly dead.

Based on the spatial transcriptomics, a set of techniques that let scientists see which genes are active at different sites in the tissue, the team built a high-resolution molecular map of the removed lungs. What they found was a landscape of total devastation.

The lungs were filled with aberrant basaloid cells—a signature of failed regeneration. The stem cells required to rebuild the lungs were almost entirely gone. The architecture was uniformly destroyed and replaced by cells that were laying down scar tissue.

“People think if you get severe ARDS, you keep supporting them and ultimately the lungs will get better,” Bharat says.  The data collected by his team suggested no amount of waiting or life support would have brought this patient’s lungs back to life. “In my practice, young patients die almost every week because no one realized that transplantation was an option,” Bharat added.

Tough choices

In many hospitals, patients with severe, acute lung infections are often allowed to die because they are considered too unstable for surgery. While Bharat’s study offers some hope that this situation might improve in the future, the team admits that their approach currently requires immense expertise and access to a highly specialized medical center with donor lungs. Making expertise and resources more accessible will take some time.

And that’s not the only thing we have to wait for. Bharat and his colleagues note in their paper that one key challenge in ARDS is determining whether the injury is reversible. His study offers some initial insights into diagnosing irreversible damage, but the researchers note that their conclusions were based on a single case. The features of lungs that are beyond repair may differ across various pathogens, stages of disease, or the way individual patient’s body responds to the disease.

For Bharat’s patient, though, all possible stars aligned. The paper says that two years after the procedure, he has returned to a normal, independent life with excellent lung function.

Med, 2026. DOI: 10.1016/j.medj.2025.100985

Read full article

Comments

Complaining about Windows 11 is a popular sport among tech enthusiasts on the Internet, whether you're publicly switching to Linux, publishing guides about the dozens of things you need to do to make the OS less annoying, or getting upset because you were asked to sign in to an app after clicking a sign-in button.

Despite the negativity surrounding the current version of Windows, it remains the most widely used operating system on the world's desktop and laptop computers, and people usually prefer to stick to what they're used to. As a result, Windows 11 has just cleared a big milestone—Microsoft CEO Satya Nadella said on the company's most recent earnings call (via The Verge) that Windows 11 now has over 1 billion users worldwide.

Windows 11 also reached that milestone just a few months quicker than Windows 10 did—1,576 days after its initial public launch on October 5, 2021. Windows 10 took 1,692 days to reach the same milestone, based on its July 29, 2015, general availability date and Microsoft's announcement on March 16, 2020.

That's especially notable because Windows 10 was initially offered as a free upgrade to all users of Windows 7 and Windows 8, with no change in system requirements relative to those older versions. Windows 11 was (and still is) a free upgrade to Windows 10, but its relatively high system requirements mean there are plenty of Windows 10 PCs that aren't eligible to run Windows 11.

Windows 10's long goodbye

It's hard to gauge how many PCs are still running Windows 10 because public data on the matter is unreliable. But we can still make educated guesses—and it's clear that the software is still running on hundreds of millions of PCs, despite hitting its official end-of-support date last October.

StatCounter, one popularly referenced source that collects OS and browser usage stats from web analytics data, reports that between 50 and 55 percent of Windows PCs worldwide are running Windows 11, and between 40 and 45 percent of them run Windows 10. StatCounter also reports that Windows 10 and Windows 7 usage have risen slightly over the last few months, which highlights the noisiness of the data. But as of late 2025, Dell COO Jeffrey Clarke said that there were still roughly 1 billion active Windows 10 PCs in use, around 500 million of which weren't eligible for an upgrade because of hardware requirements. If Windows 11 just cleared the 1 billion user mark, that suggests StatCounter's reporting of a nearly evenly split user base isn't too far from the truth.

Although Windows 11's reputation could be part of the reason for Windows 10's tenacity, it's easy to explain away the older OS's continuing popularity in other ways. Just a few years ago, Windows 10 was the most ubiquitous version of Windows since XP, and when Windows 11 launched, its requirements mostly excluded PCs older than three or four years.

Because it would be catastrophic if all of those Windows 10 PCs completely stopped getting security patches or anti-malware protection, Microsoft has built a three-year off-ramp for the operating system—one year of free (but opt-in) security updates for consumer PCs, up to three years of paid security updates for businesses and other large institutions, and updates to and basic support for core apps like Windows Defender, Microsoft Office, and Microsoft Edge through at least 2028.

Microsoft executives are at least paying lip service to Windows 11's real and perceived reputational issues. Pavan Davuluri, Microsoft's president of Windows and devices, said in a statement to The Verge that the company would be "swarming" engineers over the next few months to "urgently fix Windows 11's performance and reliability issues" and to modernize more of the barely buried parts of the operating system that still look a lot like they did in Windows 7 or XP or NT.

Assuming those issues are addressed—and that Microsoft can fix them without adding fun new bugs along the way—it would fix some of what ails Windows 11. But Davuluri didn't acknowledge the aspects of Windows 11 that make it annoying even when it's working as designed: the mandatory Microsoft account sign-in prompts, the continual reminders and upsell notifications about OneDrive, Game Pass, and other Microsoft services, and the periodic reminders to use Edge and Bing that come back no matter how many times they're dismissed.

People are definitely switching to and using Windows 11. For people already used to Windows 10, it's still the software update of least resistance. What Microsoft needs to do now is ensure that Windows 11 doesn't frustrate people so much that it's the last version of Windows they use.

Read full article

Comments