MicroLEDs, with pixels just micrometers across, have long been a byword in the display world. Now, microLED-makers have begun shrinking their creations into the uncharted nano realm. In January, a startup named Polar Light Technologies unveiled prototype blue LEDs less than 500 nanometers across. This raises a tempting question: How far can LEDs shrink?

We know the answer is, at least, considerably smaller. In the past year, two different research groups have demonstrated LED pixels at sizes of 100 nm or less.

These are some of the smallest LEDs ever created. They leave much to be desired in their efficiency—but one day, nanoLEDs could power ultra-high-resolution virtual reality displays and high-bandwidth on-chip photonics. And the key to making even tinier LEDs, if these early attempts are any precedents, may be to make more unusual LEDs.

New Approaches to LED

Take Polar Light’s example. Like many LEDs, the Sweden-based startup’s diodes are fashioned from III-V semiconductors like gallium nitride (GaN) and indium gallium nitride (InGaN). Unlike many LEDs, which are etched into their semiconductor from the top down, Polar Light’s are instead fabricated by building peculiarly shaped hexagonal pyramids from the bottom up.

Polar Light designed its pyramids for the larger microLED market, and plans to start commercial production in late 2026. But they also wanted to test how small their pyramids could shrink. So far, they’ve made pyramids 300 nm across. “We haven’t reached the limit, yet,” says Oskar Fajerson, Polar Light’s CEO. “Do we know the limit? No, we don’t, but we can [make] them smaller.”

Elsewhere, researchers have already done that. Some of the world’s tiniest LEDs come from groups who have foregone the standard III-V semiconductors in favor of other types of LEDs—like OLEDs.

“We are thinking of a different pathway for organic semiconductors,” says Chih-Jen Shih, a chemical engineer at ETH Zurich in Switzerland. Shih and his colleagues were interested in finding a way to fabricate small OLEDs at scale. Using an electron-beam lithography-based technique, they crafted arrays of green OLEDs with pixels as small as 100 nm across.

Where today’s best displays have 14,000 pixels per inch, these nanoLEDs—presented in an October 2025 Nature Photonics paper—can reach 100,000 pixels per inch.

Another group tried their hands with perovskites, cage-shaped materials best-known for their prowess in high-efficiency solar panels. Perovskites have recently gained traction in LEDs too. “We wanted to see what would happen if we make perovskite LEDs smaller, all the way down to the micrometer and nanometer length-scale,” says Dawei Di, engineer at Zhejiang University in Hangzhou, China.

Di’s group started with comparatively colossal perovskite LED pixels, measuring hundreds of micrometers. Then, they fabricated sequences of smaller and smaller pixels, each tinier than the last. Even after the 1 μm mark, they did not stop: 890 nm, then 440 nm, only bottoming out at 90 nm. These 90 nm red and green pixels, presented in a March 2025 Nature paper, likely represent the smallest LEDs reported to date.

Efficiency Challenges

Unfortunately, small size comes at a cost: Shrinking LEDs also shrinks their efficiency. Di’s group’s perovskite nanoLEDs have external quantum efficiencies—a measure of how many injected electrons are converted into photons—around 5 to 10 percent; Shih’s group’s nano-OLED arrays performed slightly better, topping 13 percent. For comparison, a typical millimeter-sized III-V LED can reach 50 to 70 percent, depending on its color.

Shih, however, is optimistic that modifying how nano-OLEDs are made can boost their efficiency. “In principle, you can achieve 30 percent, 40 percent external quantum efficiency with OLEDs, even with a smaller pixel, but it takes time to optimize the process,” Shih says.

Di thinks that researchers could take perovskite nanoLEDs to less dire efficiencies by tinkering with the material. Although his group is now focusing on the larger perovskite microLEDs, Di expects researchers will eventually reckon with nanoLEDs’ efficiency gap. If applications of smaller LEDs become appealing, “this issue could become increasingly important,” Di says.

What Can NanoLEDs Be Used For?

What can you actually do with LEDs this small? Today, the push for tinier pixels largely comes from devices like smart glasses and virtual reality headsets. Makers of these displays are hungry for smaller and smaller pixels in a chase for bleeding-edge picture quality with low power consumption (one reason that efficiency is important). Polar Light’s Fajerson says that smart-glass manufacturers today are already seeking 3 μm pixels.

But researchers are skeptical that VR displays will ever need pixels smaller than around 1 μm. Shrink pixels too far beyond that, and they’ll cross their light’s diffraction limit—that means they’ll become too small for the human eye to resolve. Shih’s and Di’s groups have already crossed the limit with their 100-nm and 90-nm pixels.

Very tiny LEDs may instead find use in on-chip photonics systems, allowing the likes of AI data centers to communicate with greater bandwidths than they can today. Chip manufacturing giant TSMC is already trying out microLED interconnects, and it’s easy to imagine chipmakers turning to even smaller LEDs in the future.

But the tiniest nanoLEDs may have even more exotic applications, because they’re smaller than the wavelengths of their light. “From a process point of view, you are making a new component that was not possible in the past,” Shih says.

For example, Shih’s group showed their nano-OLEDs could form a metasurface—a structure that uses its pixels’ nano-sizes to control how each pixel interacts with its neighbors. One day, similar devices could focus nanoLED light into laser-like beams or create holographic 3D nanoLED displays.

The head of the antitrust division is out at the US Department of Justice. Gail Slater, a former JD Vance adviser and Fox Corp VP, reportedly clashed with Attorney General Pam Bondi. Their longstanding feud is said to have centered around Slater’s skepticism of corporate mergers.

“It is with great sadness and abiding hope that I leave my role as [Assistant Attorney General] for Antitrust today,” Slater posted on X. “It was indeed the honor of a lifetime to serve in this role.”

Although Slater technically resigned, The Guardian reports that she was forced out. The fallout was said to be over her differences with Bondi (who just yesterday yelled, insulted and deflected her way through a hearing over the DOJ’s stonewalling of the Epstein files). In recent weeks, Bondi reportedly reiterated to the White House that Slater’s views on the antitrust division’s direction made the pair’s relationship irreconcilable.

The tensions reportedly began simmering last summer, when Slater sought to block the merger between Hewlett-Packard Enterprise and Juniper Networks. She opposed the deal out of concerns that it would create a duopoly in cloud computing and wireless networking. In addition, Slater reportedly told Bondi that US intelligence hadn’t raised any concerns about blocking the merger. However, CIA Director John Ratcliffe later claimed that blocking it would pose national security risks because it could lead to the loss of business to China. The Trump administration’s merger-friendly DOJ ultimately approved the deal.

Alongside Bondi, Slater was overseeing the DOJ’s review of Netflix’s proposed acquisition of Warner Bros. Discovery. In December, Trump said he would be involved in the regulatory review. That followed intense lobbying by Netflix and Paramount, the latter of which launched a hostile takeover bid. Earlier this month, The Wall Street Journal reported that the department was investigating whether Netflix was involved in anticompetitive practices during the process.

Slater’s ousting also comes weeks ahead of the DOJ’s antitrust trial against Ticketmaster owner Live Nation. The department’s lawsuit was filed during the Biden administration. It claims that Live Nation is operating as a monopoly, harming competition, fans, industry promoters and artists.

The report found that in an economy close to full employment, failing to fully utilise the advanced digital capability of women in the labour market is unsustainable.

The Economic and Social Research Institute (ESRI), which produces independent research with the objective of informing policies that support a healthy economy and social progress, has released a new report exploring gender and the digital skills gap in Europe. 

What was discovered is that Ireland currently has Europe’s largest gender gap in advanced digital skills usage in the workplace. According to the research, 44pc of men in Ireland use advanced digital skills in their jobs, compared to just 18pc of women. This shows a difference of 26pc, close to double the European average. 

The report – which was developed in partnership with Block W – noted women in Ireland use advanced digital skills at rates broadly comparable to women elsewhere in Europe, with the large gap instead reflecting the particularly high rates of advanced digital skill use among men in Ireland.

Prof Joyce O’Connor, the co-founder and chair of Block W, stated: “In an economy close to full employment, and one that relies heavily on international ICT talent, failing to fully utilise the advanced digital capability of women already in the labour market is inefficient and unsustainable.”

While differences in the types of jobs men and women do can explain a relatively larger share of the gap in Ireland than in other European countries, the report stated that a substantial portion remains unexplained, exposing the possible influence of unobserved structural, cultural and organisational factors specific to Ireland’s labour market. 

The report stated: “The evidence shows that closing the gender gap in digital skill use at work will require more than increasing women’s participation in science, technology, engineering and maths (STEM) education or occupations. While education and access to digital jobs are important, the results highlight the need for further research into other factors that may shape opportunities to develop and apply advanced digital skills.”

Closed doors

ESRI’s research found that one such factor limiting women in the widespread use of advanced digital skills is their underrepresentation in roles considered digitally intensive. The report stated: “An understanding of who gets access to and/or performs digitally-intensive work is therefore crucial for designing policies that ensure women are not left behind as the digital economy advances.”

Age was also determined to be a contributing factor, with the report finding that across Europe, women categorised as younger – under 35 – face larger digital skill gaps than their older peers.

“Gender gaps in advanced digital skills use are larger among younger workers and are less easily explained by differences in observable characteristics such as education or occupation. This suggests the issue is not a legacy problem among older cohorts and will not resolve automatically over time,” explained the report. 

Commenting on the research, Dr Adele Whelan, a senior research officer at ESRI, said: “These gender gaps persist even among women and men with similar education levels, fields of study and occupations. This indicates that encouraging women into STEM education and occupations, while essential, will not on its own close the divide. Women are underrepresented in the most digitally-intensive roles, pointing to a potential ‘digital glass ceiling’ within workplaces.

“The finding that younger women already face large gaps is a particular concern for policymakers, as it suggests the problem will not resolve on its own and requires targeted action. Addressing these issues is important not only for gender equality, but also for productivity, innovation and inclusive economic growth in Ireland.”

O’Connor added: “For Ireland, these findings should give us pause. Competitiveness, innovation and resilience depend not only on investment and infrastructure, but on what happens inside workplaces: how advanced digital work is designed and allocated, whose expertise is trusted and who gets access to high-value opportunities. 

“In an economy facing skills shortages, failing to fully utilise women’s advanced digital capability is an avoidable constraint on growth. Further research is needed on task allocation and progression within firms, and on what interventions positively impact outcomes. This report provides a timely evidence base to inform the Updated National Digital and AI Strategy and wider policy action.”

Don’t miss out on the knowledge you need to succeed. Sign up for the Daily Brief, Silicon Republic’s digest of need-to-know sci-tech news.

From automating in industry and manufacturing to space-tech, Irish robotics start-ups making a name for themselves.

Robotics can automate, make warehouses cheaper to run and safer to work in. Whether it’s a long arm helping in the production line, a tiny machine designed to work up in space, or a more “robot-looking” machine helping people in their day-to-day life, Ireland has a growing number of start-ups innovating in this space.

And in honour of automation, AI and robotics – our focus for this month – SiliconRepublic.com has put together a list of Irish robotics start-up making a name for themselves.

Akara

This 2020-founded Trinity College Dublin spin-out made it to Time Magazine’s best inventions list not once – but twice.

In 2019, the health-tech’s robot Stevie shot to fame as a Time Magazine cover star (while also making it to the best inventions list that year). While last year, the company’s AI sensor graced that list under the medical and healthcare category.

Based out of Dublin 8’s Digital Hub, Akara uses AI and robots to assist in monitoring and disinfecting critical areas such as hospital operating theatres. The start-up won the Irish leg of the 2025 KPMG Global Tech Innovator competition.

Eiratech

This Dublin-based start-up develops a “complete goods-to-person robotics automation platform” across e-fulfilment, materials handling, kitting and retail. The company offers automated guided vehicle robots, picking stations and shuttle racks in a “user-friendly” system.

Eiratech was founded in 2014 by CEO Alexey Tabolkin. The company opened its first UK office in 2021.

Eiratech says it delivers a range of services, from concept design to project installation and management. It also claims to not outsource any core services, including electrical and mechanical engineering, safety certification, software development, operational, maintenance and post-sales customer support.

Fabtech Robotics

Co Monaghan-based Fabtech Robotics provides robotic solutions to the manufacturing industry across applications including welding, coating, blasting and material preparation.

Incorporating robots into manufacturing provides faster and consistent results, an increase in production, a safer work environment and better product quality, Fabtech argues.

Founded in 2018, Fabtech was previously known as DesignPro Robotics.

Forge Robotics

Galway-based AI robotics start-up Forge Robotics was picked up by Y Combinator in last September, joining a long list of Irish start-ups that have been backed by the accelerator, such as Luminate Medical, Protex AI and Solidroad.

Forge Robotics, founded by CEO Eoin Cobbe and chief technology officer Robert Cormican, is building an AI-powered intelligence layer to improve the welding capabilities of industrial robots. Its system allows robots to scan a part, interpret its geometry and execute welds even when the set-up is imperfect.

Despite just being inducted into the accelerator, the start-up had already incorporated a US entity and is on track to launch their first product in July this year. They received $500,000 in pre-seed funding as part of acceptance into the San Francisco-based programme.

HomeBot Ireland

HomeBot Ireland is a 2024-founded start-up from Cork that makes robotic vacuums, mops and lawnmowers.

The businesses was founded by Clara Mulligan and her husband Alan. HomeBot Ireland’s main mission, Mulligan says, is to make machines that are “really simple to use, with excellent customer service”.

The flagship product in the company’s growing range is the AI-enabled robot mower, Buddy, which is a wireless robot equipped with cameras and sensors. Buddy is programmed to recognise anything that isn’t grass, such as flower beds, paths and water.

Icarus

Space-tech start-up Icarus was co-founded by Co Tyrone-born Jamie Palmer. The New York-headquartered start-up raised $6.1m in a seed round last September led by Soma Capital and Xtal, with participation from Nebular and Massive Tech Ventures.

The start-up wants to create an intelligent robot force that take over the time-consuming and menial tasks. Icarus isn’t aiming for a humanoid model, rather, a fan-propelled robot with arms fitted with grippers.

The less than two-years-old start-up’s first robot will unpack and stow cargo, the founders told TechCrunch last year.

Nitrexo

Cork-based space-tech Nitrexo specialises in delivering cost-effective solutions for thermal and analytical problems.

Founded in 2019, Nitrexo has already completed seven projects with customers engaged in manufacturing satellites and launchers, spacecraft and other instruments.

In 2021, Nitrexo launched a Europe-wide product called ‘Digital Engineer’, backed by the European Space Agency (ESA). Digital Engineer is an AI robot designed for distributed learning and working environments. The project won €500,000 in support from the ESA in space-based funding to bridge the gap between engineering education and industry.

Volta Robotics

Also a Cork start-up, the 2023-founded Volta Robotics provides technology that helps reuse battery packs to prolong their life cycle. This start-up is working on technology to extract battery cells from EV batteries, in order to repackage them into residential or utility energy storage.

Founder Elvis Seporaitis participated in the 2023 New Frontiers programme hosted by Munster Technological University, where he was awarded for his business idea.

Don’t miss out on the knowledge you need to succeed. Sign up for the Daily Brief, Silicon Republic’s digest of need-to-know sci-tech news.

OPPO has officially announced that the Find X9s is coming to India. This is an addition to the Find X9 and the X9 Pro. Although the Chinese smartphone maker has not announced a launch timeline yet, industry reports suggest that the device could debut in China in the second quarter of 2026. The device is expected to reach India and other regions after its China debut.

Design and Processor

Rumors and leaks are whispering about a 6.3-inch OLED LTPS display with an approximate 1.5 K resolution. The smartphone is also rumored to feature an ultrasonic in-display fingerprint scanner and IP68 and IP69 certifications for dust and water resistance.

OPPO has officially revealed that the Find X9s will use the MediaTek Dimensity 9500 SoC. The chipset arrived recently with performance-focused upgrades. It is expected to boost overall performance while reducing power consumption. This makes the phone suitable for demanding apps and gaming.

Battery and Camera

According to leaks, the OPPO Find X9s is expected to feature a massive 7,000 mAh battery. It is also expected to support wireless charging, which is quite an exciting feature.

On the camera front, the Find X9s is rumoured to feature a triple-rear-camera setup. The phone may include a 200MP primary camera, a 200MP periscope telephoto lens, and a 50MP ultra-wide sensor. Some leaks also mention a multispectral lens.

Availability

OPPO has so far only confirmed the launch of the Find X9s in India. Alongside, the device is expected to launch in multiple regions, including Southeast Asian countries such as Vietnam, Singapore, and Malaysia. OPPO is yet to confirm the global timeline.

For decades, VPI Industries owned the record-cleaning conversation. The HW-16.5 was the default answer: brutally effective, industrial-strength, and loud enough to wake the neighbors. Then the vinyl revival (and the pandemic) hit and VPI raised the price from roughly $650 to $1,200. That single move cracked the door wide open. Opportunity rushed in. Brands like Pro-Ject, Record Doctor, Humminguru, and Okki Nokki didn’t just notice—they pounced. VPI still brings unmatched suction power and a tank-like motor, but it also brings noise and a price tag that now invites competition. Enter the Okki Nokki ONE: quieter, re-engineered from the ground up, and priced at $749; clear evidence that VPI’s long-standing dominance created the very opening now being exploited. The market shifted. This is what happens when you leave money on the table.

Okki Nokki ONE Record Cleaning Machine

Okki Nokki is back in full production and for the first time in many years, its record cleaning machines are once again being manufactured in the Netherlands. After a difficult stretch defined by supply-chain disruptions and uneven overseas production, manufacturing has returned to Europe with a renewed emphasis on build quality, long-term reliability, and hands-on craftsmanship.

Following the passing of founder Johan Bezem, his daughter Sanne Bezem stepped in to carry forward the company he built more than 30 years ago. With clear respect for her father’s vision and a pragmatic eye on what the brand needs to survive and grow, Sanne has taken Okki Nokki back to its roots: tighter control over engineering and manufacturing, direct involvement in product development, and a renewed commitment to doing things the right way rather than the fast way.

Okki Nokki isn’t simply back on its feet. It’s operational again, evolving by design, and now firmly in dedicated hands for the long road ahead.

Vana Ltd, a specialty distributor supplying the North American market with carefully selected audio components from international manufacturers, is now delivering the Okki Nokki ONE record cleaning machine, the successor to the long-standing MK II.

“The ONE is an example of Okki Nokki’s commitment to overseeing every step of the design and manufacturing process,” said Roy Feldstein, Chief Technology Officer of Vana Ltd and exclusive North American importer of Okki Nokki. “Instead of simply offering a next-generation update to a consumer favorite, Okki Nokki’s engineers started with a clean slate, ultimately creating a solid, consistently dependable cleaning machine that will perform to spec, year after year.”

The Okki Nokki ONE is engineered around simplicity, durability, and genuinely low-noise operation, positioning it among the quietest record cleaning machines in its class. Day-to-day use is intentionally straightforward, with all functions controlled by a single knob for fast, intuitive operation. At the core of the design is a molded high-grade ABS unibody chassis, the first of its kind in a record cleaning machine—which reduces the number of internal components and integrates the fluid reservoir directly into the structure.

This approach is intended to improve long-term reliability while eliminating common failure points found in older designs. A newly engineered air and fluid suction path, inspired by high-end vacuum technology, further contributes to quieter operation and helps prevent leaks.

The ONE also removes much of the friction from the cleaning process itself. A configurable universal vacuum arm supports 7-inch, 10-inch, and 12-inch records without requiring arm changes, while a stainless steel record clamp fully protects the label during cleaning.

Okki Nokki includes both full-size and label-only platters to minimize the risk of transferring debris when cleaning both sides of a record. A dust cover is included as standard, along with a bottle of Okki Nokki cleaning fluid concentrate and a goat-hair brush, making the ONE a complete, ready-to-use solution right out of the box.

Adds Feldstein: “Record cleaning is a necessary part of enjoying a vinyl collection, but it doesn’t have to be a chore. The Okki Nokki ONE makes it a quick and easy process—and at a much lower volume than competitive products.”

The Bottom Line

The real news here is not just that Okki Nokki has a new machine, but that the brand has taken full control of its future again. Manufacturing is back in the Netherlands, the design has been rethought from the ground up, and the ONE is positioned squarely against long-established heavyweights in the category. Okki Nokki claims meaningfully quieter operation than traditional vacuum-based machines, and in practice the redesigned suction path and enclosed unibody construction suggest that noise reduction was a genuine engineering priority rather than a marketing bullet.

Reservoir capacity and absolute suction power have not been disclosed in hard numbers, which leaves open the question of whether it can fully match the raw vacuum force of the long-dominant New Jersey powerhouse. That said, brute force has never been the only path to effective cleaning.

At $749, the Okki Nokki ONE lands in a highly competitive sweet spot. If its quieter operation and simplified workflow deliver the deeper, more consistent cleaning promised, it becomes a compelling alternative to louder, more industrial designs that now cost considerably more. This is not aimed at casual vinyl dabblers or those content with manual cleaning. The ONE is for serious record listeners and collectors who want a deeper clean, less noise, fewer compromises, and a machine designed to live in a listening room rather than a garage.

Where to buy: $749 at Okki Nokki USA

Related Reading:

Welcome to 2026. It’s the year we all hoped PC hardware would finally get cheaper, so we could build new PCs without breaking the bank. Unfortunately, none of this has come true. Our beloved AI companions, like ChatGPT, have bought up most of the RAM that will be produced over the next few years, increasing prices and limiting accessibility. This means that if you were thinking about building a PC, your budget will need to increase, or you’ll need to cut corners on accessories. That’s exactly where HyperX’s Cloud Jet Dual Wireless headphones come into play. These gaming headphones cost ₹4,999 (or $50) and offer wireless connectivity via both a dongle and Bluetooth, a claimed 25 hours of battery life, and 40mm drivers.

Those are quite decent specs for not a lot of money, and that made me wonder: where has HP cut corners? To find out, I called my friends, got the Cloud Jet Wireless in for a review, and used it as my primary gaming headset, with both my phone and the Asus ROG XBOX Ally.

Design & Comfort

HyperX hasn’t tried to revolutionize headphones with the Cloud Jet Wireless. In fact, they look exactly how you’d expect them to. And I quite like it. The blue headphones add a touch of uniqueness to a somewhat black-ish, stale world, all while remaining conventional. Inside the box, you get the headphones, along with the 2.4 GHz dongle, and that’s it. These don’t fold like others, so travelling with them could be a challenge.

The left side of the headphones houses all the controls. This includes a power/pairing button, a USB-C charging port, a toggle to switch between Wi-Fi and Bluetooth modes, and a physical volume adjuster. I can’t stress enough how much I love physical volume controls, as they work 100% of the time without needing to fiddle with annoying touch controls.

What HP has nailed out of the park is the comfort. The headphones are light, weighing under 300 grams. This allows them to feel natural while sitting on the head. Speaking of that, there’s a very niche elastic headband that distributes the weight perfectly, adding comfort. The mesh ear cushions are large and do a fantastic job of enveloping you in sound without clamping your ears too hard. I wore the Cloud Jet headphones for a three-hour BGMI marathon at night and never once felt uncomfortable. My ears also tend to heat up during intense battles, which can make over-the-ear headphones uncomfortable, but it wasn’t an issue here.

Sound Quality

Gamers are perhaps the most picky buyers on the market. They need the best specs at the lowest price and don’t want to compromise on anything. Well, that can’t be true for everything. Brands have to cut certain corners to achieve a lower price. However, with the Cloud Jet Wireless, sound isn’t where corners have been cut. In fact, the audio on these headphones punches way above their weight. In games like F1 2025, Forza Motorsport, and COD Modern Warfare, I noticed the sound was punchy with bass, which was necessary to feel the explosions. I could hear faint footsteps in games like Counter-Strike, which wasn’t enough to improve my trash skills, but did help me not get killed by a knife every time. Beyond that, the dialogues were clear, and the treble is decent as well.

Music listening sessions were better than many budget TWS earbuds I’ve tested, but not amazing. Songs with plenty of instruments will lack separation, but if you haven’t been testing headphones for a living like me, the difference isn’t that much.

If I had to pick one reason to buy the Cloud Jet Wireless, it would be the mic. The swivel-to-activate feature is what every single gaming headphone should adopt, simply because it’s convenient. Want to talk to your gaming buddies? Just swivel the mic down, and it’s activated. Swivel it back, and it’s disabled without needing to fiddle with the PC or game settings. Even the quality is really good. My teammates reported hearing me loud and clear, with little background noise.

Battery Life

HP claims the Cloud Wireless headphones can last 25 hours on Bluetooth 5.2 or 20 hours on 2.4 GHz mode. I found these claims to be plenty accurate. During my review period, when usage was split between a few hours of music listening at about 75% volume and gaming, I hit 23 hours. That translates to more than a week of use — more than enough for most people. Charging is handled via the USB-C port, which is a very nice feature if you don’t like carrying multiple cables, like me. The only gripe was the charging time, which, at 4 hours from empty to full, is a bit slow.

Verdict

At ₹4,999 ($50), the HyperX Cloud Jet Dual Wireless headphones are hard to beat. They serve as a great entry point for someone looking to get a dedicated pair of gaming headphones without breaking the bank. They connect with all the consoles, including PS5, Nintendo Switch, and Steam Deck, sound pretty decent, and fit snugly on your head without causing any discomfort. Combine all that with a battery life that lasts for weeks, and I’d be using them as my primary gaming headphones. If you’re in the market for such a thing, these HyperX headphones are hard to ignore.

At the end of last month, Ubisoft workers in the publisher’s native France threatened to strike in the wake of sweeping layoffs and cost-cutting measures. This week, they made good on those threats. According to GamesIndustry.biz, union members confirmed that at least 1,200 staff participated in the three-day strike, which was due to run from February 10 to February 12.

While the strike action primarily took place in France, GamesIndustry.biz was told that Ubisoft’s Milan office also took part. The union Solidaires Informatique, which represents French workers from a number of companies in the video game sector, including Blizzard and Ubisoft, had previously called for strikes to take place on January 27. Their demands included a 10 percent increase on all salaries and the implementation of a 4-day work week.

Some striking employees held up signs outside Ubisoft’s Paris headquarters, with one (pictured) wearing a Rabbids mask to hide their face. Their grievances are wide-ranging. As well as reportedly laying off hundreds of employees already in 2026, Ubisoft also introduced a mandate for its staff to return to work on site for five days a week. One employee who publicly voiced their disapproval of the new policy was reportedly fired for doing so.

Ubisoft has had a rocky start to 2026 on the software side too. The long-awaited Prince of Persia: The Sands of Time remake was among six games canceled by the struggling publisher last month, when it also confirmed several studio closures as part of the company’s organizational restructuring.

Update, Feb. 12 2026, 12:39PM ET: “We understand these changes, particularly those affecting work organization, are generating strong feelings,” Ubisoft wrote in a statement shared with Engadget. “Since the announcement, we have held a series of discussions and information sessions at multiple levels to help teams better understand the new organization and to give them the opportunity to share their questions and concerns.” The company added that it “remains committed to maintaining an open and constructive dialogue with employees and employee representatives.”

When an AI agent visits a website, it’s essentially a tourist who doesn’t speak the local language. Whether built on LangChain, Claude Code, or the increasingly popular OpenClaw framework, the agent is reduced to guessing which buttons to press: scraping raw HTML, firing off screenshots to multimodal models, and burning through thousands of tokens just to figure out where a search bar is.

That era may be ending. Earlier this week, the Google Chrome team launched WebMCP — Web Model Context Protocol — as an early preview in Chrome 146 Canary. WebMCP, which was developed jointly by engineers at Google and Microsoft and incubated through the W3C’s Web Machine Learning community group, is a proposed web standard that lets any website expose structured, callable tools directly to AI agents through a new browser API: navigator.modelContext.

The implications for enterprise IT are significant. Instead of building and maintaining separate back-end MCP servers in Python or Node.js to connect their web applications to AI platforms, development teams can now wrap their existing client-side JavaScript logic into agent-readable tools — without re-architecting a single page.

AI agents are expensive, fragile tourists on the web

The cost and reliability issues with current approaches to web-agent (browser agents)  interaction are well understood by anyone who has deployed them at scale. The two dominant methods — visual screen-scraping and DOM parsing — both suffer from fundamental inefficiencies that directly affect enterprise budgets.

With screenshot-based approaches, agents pass images into multimodal models (like Claude and Gemini) and hope the model can identify not only what is on the screen, but where buttons, form fields, and interactive elements are located. Each image consumes thousands of tokens and can have a long latency. With DOM-based approaches, agents ingest raw HTML and JavaScript — a foreign language full of various tags, CSS rules, and structural markup that is irrelevant to the task at hand but still consumes context window space and inference cost.

In both cases, the agent is translating between what the website was designed for (human eyes) and what the model needs (structured data about available actions). A single product search that a human completes in seconds can require dozens of sequential agent interactions — clicking filters, scrolling pages, parsing results — each one an inference call that adds latency and cost.

How WebMCP works: Two APIs, one standard

WebMCP proposes two complementary APIs that serve as a bridge between websites and AI agents.

The Declarative API handles standard actions that can be defined directly in existing HTML forms. For organizations with well-structured forms already in production, this pathway requires minimal additional work; by adding tool names and descriptions to existing form markup, developers can make those forms callable by agents. If your HTML forms are already clean and well-structured, you are probably already 80% of the way there.

The Imperative API handles more complex, dynamic interactions that require JavaScript execution. This is where developers define richer tool schemas — conceptually similar to the tool definitions sent to the OpenAI or Anthropic API endpoints, but running entirely client-side in the browser. Through the registerTool(), a website can expose functions like searchProducts(query, filters) or orderPrints(copies, page_size) with full parameter schemas and natural language descriptions.

The key insight is that a single tool call through WebMCP can replace what might have been dozens of browser-use interactions. An e-commerce site that registers a searchProducts tool lets the agent make one structured function call and receive structured JSON results, rather than having the agent click through filter dropdowns, scroll through paginated results, and screenshot each page.

The enterprise case: Cost, reliability, and the end of fragile scraping

For IT decision makers evaluating agentic AI deployments, WebMCP addresses three persistent pain points simultaneously.

Cost reduction is the most immediately quantifiable benefit. By replacing sequences of screenshot captures, multimodal inference calls, and iterative DOM parsing with single structured tool calls, organizations can expect significant reductions in token consumption. 

Reliability improves because agents are no longer guessing about page structure. When a website explicitly publishes a tool contract — “here are the functions I support, here are their parameters, here is what they return” — the agent operates with certainty rather than inference. Failed interactions due to UI changes, dynamic content loading, or ambiguous element identification are largely eliminated for any interaction covered by a registered tool.

Development velocity accelerates because web teams can leverage their existing front-end JavaScript rather than standing up separate backend infrastructure. The specification emphasizes that any task a user can accomplish through a page’s UI can be made into a tool by reusing much of the page’s existing JavaScript code. Teams do not need to learn new server frameworks or maintain separate API surfaces for agent consumers.

Human-in-the-loop by design, not an afterthought

A critical architectural decision separates WebMCP from the fully autonomous agent paradigm that has dominated recent headlines. The standard is explicitly designed around cooperative, human-in-the-loop workflows — not unsupervised automation.

According to Khushal Sagar, a staff software engineer for Chrome, the WebMCP specification identifies three pillars that underpin this philosophy. 

1. Context: All the data agents need to understand what the user is doing, including content that is often not currently visible on screen. 

2. Capabilities: Actions the agent can take on the user’s behalf, from answering questions to filling out forms. 

3. Coordination: Controlling the handoff between user and agent when the agent encounters situations it cannot resolve autonomously.

The specification’s authors at Google and Microsoft illustrate this with a shopping scenario: a user named Maya asks her AI assistant to help find an eco-friendly dress for a wedding. The agent suggests vendors, opens a browser to a dress site, and discovers the page exposes WebMCP tools like getDresses() and showDresses().  When Maya’s criteria go beyond the site’s basic filters, the agent calls those tools to fetch product data, uses its own reasoning to filter for “cocktail-attire appropriate,” and then calls showDresses()to update the page with only the relevant results. It’s a fluid loop of human taste and agent capability, exactly the kind of collaborative browsing that WebMCP is designed to enable.

This is not a headless browsing standard. The specification explicitly states that headless and fully autonomous scenarios are non-goals. For those use cases, the authors point to existing protocols like Google’s Agent-to-Agent (A2A) protocol. WebMCP is about the browser — where the user is present, watching, and collaborating.

Not a replacement for MCP, but a complement

WebMCP is not a replacement for Anthropic’s Model Context Protocol, despite sharing a conceptual lineage and a portion of its name. It does not follow the JSON-RPC specification that MCP uses for client-server communication. Where MCP operates as a back-end protocol connecting AI platforms to service providers through hosted servers, WebMCP operates entirely client-side within the browser.

The relationship is complementary. A travel company might maintain a back-end MCP server for direct API integrations with AI platforms like ChatGPT or Claude, while simultaneously implementing WebMCP tools on its consumer-facing website so that browser-based agents can interact with its booking flow in the context of a user’s active session. The two standards serve different interaction patterns without conflict.

The distinction matters for enterprise architects. Back-end MCP integrations are appropriate for service-to-service automation where no browser UI is needed. WebMCP is appropriate when the user is present and the interaction benefits from shared visual context — which describes the majority of consumer-facing web interactions that enterprises care about.

What comes next: From flag to standard

WebMCP is currently available in Chrome 146 Canary behind the “WebMCP for testing” flag at chrome://flags. Developers can join the Chrome Early Preview Program for access to documentation and demos. Other browsers have not yet announced implementation timelines, though Microsoft’s active co-authorship of the specification suggests Edge support is likely.

Industry observers expect formal browser announcements by mid-to-late 2026, with Google Cloud Next and Google I/O as probable venues for broader rollout announcements. The specification is transitioning from community incubation within the W3C to a formal draft — a process that historically takes months but signals serious institutional commitment.

The comparison that Sagar has drawn is instructive: WebMCP aims to become the USB-C of AI agent interactions with the web. A single, standardized interface that any agent can plug into, replacing the current tangle of bespoke scraping strategies and fragile automation scripts.

Whether that vision is realized depends on adoption — by both browser vendors and web developers. But with Google and Microsoft jointly shipping code, the W3C providing institutional scaffolding, and Chrome 146 already running the implementation behind a flag, WebMCP has cleared the most difficult hurdle any web standard faces: getting from proposal to working software.