OpenAI partners with AMD to deploy 6 gigawatts of GPUs from 2026, boosting large-scale AI infrastructure with multi-year technology collaboration. Read...

In the ongoing war of AI investment, OpenAI has secured itself a new ally: AMD. The chip maker will trade millions of its upcoming Instinct MI450 GPUs for an investment by the AI company, worth up to 10 percent of its stock. The numbers, though, remain vague. The deal hinges on AMD’s ability to deliver “6 gigawatts” worth of Instinct MI450 GPUs, a rack-scale enterprise GPU chip that the chip manufacturer hasn’t begun shipping yet. If it begins shipping the MI450 by an undisclosed milestone and in undisclosed amounts, then OpenAI has warrants to buy the company’s stock, worth up to 160 million shares. That would be about 10 percent of its current outstanding shares, according to CNBC. AMD must deliver its first tranche, or shipment, of MI450 GPUs by the second half of 2026, worth one gigawatt. The total deal encompasses six gigawatts, though the company’s announcement of the deal didn’t put a timetable to the final shipments. It also includes “multiple generations” of Instinct chips. Since AMD hasn’t formally announced the MI450 yet, it’s unclear how six gigawatts’ worth translates to in actual chips. Assuming that the MI350X draws a kilowatt of power apiece, and that the older MI300X drew a board power of 750 watts, I asked OpenAI’s ChatGPT for a projection. It returned a range of between three to six million GPUs, with a likelier target of between four and five million. That also assumes that the upcoming MI355X draws 1,400W, which hasn’t been confirmed. In the technology space, there’s one surefire domestic source of nearly unlimited cash — and no, it’s not the Trump administration. That administration has already agreed to convert its CHIPS Act investment into Intel and transfer it into a nearly 10 percent stake. OpenAI’s cash reserves aren’t publicly known, but it’s in the process of raising a $40 billion funding round this year, and CNBC reports that the AI company is already pulling in between $10 billion and $13 billion per year. Is AMD headed for the cloud, and not the PC? But there’s a very uneasy subtext in all this, too. AMD chief executive Lisa Su now has a very loud, dynamic, and persuasive voice telling her to invest in high-end GPUs for the cloud, and not the PC. Every business, from Intel to Nvidia to AMD, has to decide how to spend their capital allotment and negotiate for production output inside TSMC and other fabs. “We are thrilled to partner with OpenAI to deliver AI compute at massive scale,” Su said in a statement. “This partnership brings the best of AMD and OpenAI together to create a true win-win enabling the world’s most ambitious AI buildout and advancing the entire AI ecosystem.” When a report was published last week that AMD might use Intel as a production partner, no one from Intel would comment. Charlie Demerjian at SemiAccurate reported that the rumor was simply not true. It most likely isn’t, but the problem in the breakneck world of AI, where truckloads of money are backed up to anyone who can use “AI” in a press release, is that most anything is somewhat plausible these days. Intel has already built tiles inside its Core Ultra PC processors at both its own fabs as well as at TSMC, of course. The second half of 2026 is far away, but enthusiasts do have to grumble and worry. With more and more emphasis being placed on GPU training and inferencing in the cloud, how much will be left for PCs? Nvidia already controls more than 90 percent of all PC GPU shipments, even after AMD had made waves about trying to aim at the mainstream PC market instead of the high end. If it can’t succeed in PCs, why wouldn’t it simply turn to the more lucrative enterprise market instead? Sure, Nvidia GeForce 5000-series GPUs may be near MSRP once again. But remove a source of competition, and who knows how long that will last? Read...

In the ongoing war of AI investment, OpenAI has secured itself a new ally: AMD. The chip maker will trade millions of its upcoming Instinct MI450 GPUs for an investment by the AI company, worth up to 10 percent of its stock. The numbers, though, remain vague. The deal hinges on AMD’s ability to deliver “6 gigawatts” worth of Instinct MI450 GPUs, a rack-scale enterprise GPU chip that the chip manufacturer hasn’t begun shipping yet. If it begins shipping the MI450 by an undisclosed milestone and in undisclosed amounts, then OpenAI has warrants to buy the company’s stock, worth up to 160 million shares. That would be about 10 percent of its current outstanding shares, according to CNBC. AMD must deliver its first tranche, or shipment, of MI450 GPUs by the second half of 2026, worth one gigawatt. The total deal encompasses six gigawatts, though the company’s announcement of the deal didn’t put a timetable to the final shipments. It also includes “multiple generations” of Instinct chips. Since AMD hasn’t formally announced the MI450 yet, it’s unclear how six gigawatts’ worth translates to in actual chips. Assuming that the MI350X draws a kilowatt of power apiece, and that the older MI300X drew a board power of 750 watts, I asked OpenAI’s ChatGPT for a projection. It returned a range of between three to six million GPUs, with a likelier target of between four and five million. That also assumes that the upcoming MI355X draws 1,400W, which hasn’t been confirmed. In the technology space, there’s one surefire domestic source of nearly unlimited cash — and no, it’s not the Trump administration. That administration has already agreed to convert its CHIPS Act investment into Intel and transfer it into a nearly 10 percent stake. OpenAI’s cash reserves aren’t publicly known, but it’s in the process of raising a $40 billion funding round this year, and CNBC reports that the AI company is already pulling in between $10 billion and $13 billion per year. Is AMD headed for the cloud, and not the PC? But there’s a very uneasy subtext in all this, too. AMD chief executive Lisa Su now has a very loud, dynamic, and persuasive voice telling her to invest in high-end GPUs for the cloud, and not the PC. Every business, from Intel to Nvidia to AMD, has to decide how to spend their capital allotment and negotiate for production output inside TSMC and other fabs. “We are thrilled to partner with OpenAI to deliver AI compute at massive scale,” Su said in a statement. “This partnership brings the best of AMD and OpenAI together to create a true win-win enabling the world’s most ambitious AI buildout and advancing the entire AI ecosystem.” When a report was published last week that AMD might use Intel as a production partner, no one from Intel would comment. Charlie Demerjian at SemiAccurate reported that the rumor was simply not true. It most likely isn’t, but the problem in the breakneck world of AI, where truckloads of money are backed up to anyone who can use “AI” in a press release, is that most anything is somewhat plausible these days. Intel has already built tiles inside its Core Ultra PC processors at both its own fabs as well as at TSMC, of course. The second half of 2026 is far away, but enthusiasts do have to grumble and worry. With more and more emphasis being placed on GPU training and inferencing in the cloud, how much will be left for PCs? Nvidia already controls more than 90 percent of all PC GPU shipments, even after AMD had made waves about trying to aim at the mainstream PC market instead of the high end. If it can’t succeed in PCs, why wouldn’t it simply turn to the more lucrative enterprise market instead? Sure, Nvidia GeForce 5000-series GPUs may be near MSRP once again. But remove a source of competition, and who knows how long that will last? Read...

At a glanceExpert`s Rating Pros Decent performer for PCIe 4.0 HMB Super-stylish and beefy heatsink Affordable given the heatsink Cons Runs off of secondary cache in PS5 Our Verdict You can use Orico’s PCIe 4.0 OS5 NVMe SSD in the PS5 it’s marketed towards, however PS5 doesn’t support HMB so it will operate off secondary cache. Workable, but not as quick as a DRAM design. Price When Reviewed This value will show the geolocated pricing text for product undefined Best Pricing Today Best Prices Today: Orico OS5 PCIe 4.0 SSD Retailer Price $71.99 View Deal $79.99 View Deal Price comparison from over 24,000 stores worldwide Product Price Price comparison from Backmarket Best Prices Today: Check today’s prices The OS5 NVMe SSD from Orico is a looker thanks to its stylish and beefy heatsink. It’s moderately fast for its ilk and will work fine, if not optimally in the PS5 it’s marketed for. That gaming console doesn’t support the host memory buffer (HMB) technology that the OS5 uses for primary caching duty on a PC. Read on to learn more, then see our roundup of the best SSDs for comparion. What are the Orico OS5’s features? You’ll notice from the photo at the head of the article that the OS5 ships with a rather hefty heatsink to fit its 2280 (22mm wide, 80mm long) form. It’s a PCIe 4.0, NVMe SSD using a host memory buffer (HMB) for primary caching in place of DRAM. The OS5’s controller is a Maxio 1602A and its NAND is 144-layer TLC (Triple-Level Cell/3-bit). Seemingly around 35 percent of that can be dedicated to secondary cache (Orico claims 20 percent), i.e., writing the NAND as single-bit SLC (Single-Level Cell). Another view of the OS5 and its handsome heatsink. Orico warranties the OS5 for five years, which is mitigated by a 600TBW (terabytes that may be written before read-only commences) per terabyte of capacity. That’s about average for the industry and 144-layer TLC. Should you ever want to know how close you are to your TBW limit, download CrystalDiskInfo or a similar utility and check the “percentage used” number. As this only increases with writes, it’s likely going be a lot less than you might imagine. My last main computer was only at 1 percent after two years, albeit with heavy use of external storage for many tasks. How much is the Orico OS5? The OS5 is $90 in the 1TB capacity, $150 for 2TB, and $280 for 4TB (prices on Amazon are about $20 less than that). That’s quite a bit of capacity for a reasonable price given that the included heatsink is generally a $10 to $15 option with most SSDs. I wouldn’t call OS5 an outrageous bargain, but it’s certainly competitive. How fast is the Orico OS5? Using our new test MO and equipment, the OS5 turned in the fourth fastest PCIe 4.0/HMB score out of six SSDs. Not great, though the difference isn’t vast in most tasks, as you’ll see below. The upshot is that even slower NVMe SSDs are really, really fast. The OS5 we tested was hurt by the fact that it’s only 1TB and nearly all the SSDs we test are 2TB or 4TB. Because of that, it ran out of secondary cache more quickly. This really took a toll in the 450GB write. But the OS5 was certainly competitive in CrystalDiskMark 8, including the sequential transfer tests shown below. The OS5 offers quite a bit of capacity for a reasonable price given that the included heatsink is generally a $10 to $15 option with most SSDs. The OS5 was competitive in CrystalDiskMark 8, including the sequential transfer tests shown here. Longer bars are better. CrystalDiskMark 8’s 4K numbers were much the same story. Hardly awesome, but good enough for government work. CrystalDiskMark 8’s 4K numbers were much the same story, not ground-shaking but close enough for rock n’ roll. Longer bars are better. When shuffling only 48GB around, the Orico OS5 was largely on par with the more capacious Lexar 2280 Play SE (another heatsink/HMB design marketed for the PS5), Teamgroup MP44Q, and WD Blue SN5100. Faster in some cases, a tad slower in others. Note that FastCopy is a highly recommended file transfer utility that operates far nearer the pace you see in synthetic benchmarks than Windows Explorer. When shuffling only 48GB around, the Orico OS5 was competitive with the more capacious PCIe 4.0/HMB SSDs. Faster in some cases, a tad slower in others. Shorter bars are better. The 450GB write is where the OS5 was body-slammed by its competitors. The result of the already discussed lower capacity and lack of secondary cache. However… Arrgh! With only 1TB on board, the OS5 was at the mercy of other 2TB/4TB SSDs with more secondary cache. Shorter bars are better. …the long write news isn’t all bad. Once secondary cache is exhausted, the TLC still transcribes data to cells at a livable pace of around 1GBps. Believe me, I’ve seen worse. How about 75MBps in the early days of QLC? This really isn’t that bad a pace for native writing. Twice SATA. Not ideal, but not one of the 75MBps to 150MBps horror stories we used to see. Should you buy the Orico OS5? Though not optimal for the PS5, like other HMB designs, the OS5 will get the job done adequately and for a lot less than a DRAM model. I like the look of the heatsink, so if you find the right price, have at it. But to be honest, slapping a heatsink on HMB and then marketing the SSD as ideal for PS5 is not my favorite marketing schtick. That goes for Orico, but also Lexar with its Play 2280 series. How we test Drive tests currently utilize Windows 11 24H2, 64-bit running off of a PCIe 4.0 Samsung 990 Pro in an Asus Z890-Creator WiFi (PCIe 4.0/5.0) motherboard. The CPU is a Core Ultra i5 225 feeding/fed by two Crucial 64GB DDR5 4800MHz modules (128GB of memory total). Both 20Gbps USB and Thunderbolt 5 are integrated and Intel CPU/GPU graphics are used. Internal PCIe 5.0 SSDs involved in testing are mounted in a Asus Hyper M.2 x16 Gen5 adapter card. We run the CrystalDiskMark 8.04 (and 9), AS SSD 2, and ATTO 4 synthetic benchmarks (to keep article length down, we only report one) to find the storage device’s potential performance, then a series of 48GB and 450GB transfers tests using Windows Explorer drag and drop to show what users will see during routine copy operations, as well as the far faster FastCopy run as administrator to show what’s possible. A 20GBps two-SSD RAID 0 array on the aforementioned Asus Hyper M.2 x16 Gen5 is used as the second drive in our transfer tests. Formerly the 48GB tests were done with a RAM disk. Each test is performed on a NTFS-formatted and newly TRIM’d drive so the results are optimal. Note that in normal use, as a drive fills up, performance may decrease due to less NAND for secondary caching, as well as other factors. This issue has abated somewhat with the current crop of SSDs utilizing more mature controllers and far faster, late-generation NAND. Note that our testing MO evolves and these results may not match those from previous articles. Only comparisons inside the article are 100% valid as those results are gathered using the current hardware and MO. Read...

They say the proposed change to technology, hospitality, and food and nutrition subjects will narrow students` options and cause job cuts. Read...

TL;DR: For one week only, get lifetime access to the PDF Converter & Editor for just $24.99 (MSRP $99.99) with code PDF15. Here’s a little productivity upgrade you don’t want to miss: through October 12, you can grab lifetime access to the PDF Converter & Editor for just $24.99 (MSRP $99.99) when you use code PDF15 at checkout. That’s a one-time payment for a tool that makes working with PDFs way less painful — no recurring costs, no annoying limitations. Whether you’re merging pages, converting files to Word or Excel, or extracting text from scanned documents, this software gives you all the PDF power you need in one place. It’s fast, reliable, and trusted by individuals and businesses to make daily document workflows way smoother. Highlights at a Glance Convert to & from PDF — Word, Excel, PowerPoint, HTML, PNG, JPG & more Edit & Organize Pages — Merge, split, rotate, or delete in a few clicks OCR Technology — Extract and edit text from scanned PDFs Form Tools — Fill out or build interactive forms with full customization Compress, Protect & Unlock — Control file sizes and secure your documents Annotate & Watermark — Add comments, shapes, or branding to your PDFs If PDFs are part of your daily grind, this is an easy one-time investment that will pay for itself the next time you need to edit or convert a document. Get a lifetime of PDF Converter & Editor for just $24.99 (MSRP $99.99) when you use code PDF15 at checkout through October 12. PDF Converter & Editor: Lifetime LicenseSee Deal StackSocial prices subject to change. Read...

Whether you’re a small business trying to promote your services, a content creator looking to polish your latest YouTube post, or simply someone who wants to experiment with making cool home videos, using the right software is essential. If it’s too complex, then it can get in the way of your creativity. Alternatively, not having enough tools to fully realise your vision is a very frustrating experience.  Adobe understands this and provides a variety of solutions so you can get just what you need. Plus, the incorporation of the latest AI technology means you can easily do a lot more than you might have thought possible. So, which is the right package for you? Adobe Firefly makes dreams a reality Thanks to its powerful generative AI capabilities, Adobe Firefly is the perfect tool for creating short videos that can boost a presentation or social media post. Using ordinary language prompts or an existing image, you can quickly transform your ideas into professional-looking clips with just a few words. There’s even a built-in prompt enhancer to help produce the results you’re after with the minimum of fuss.  Adobe You don’t need any pro video skills, as Adobe Firefly takes care of everything, but you do get a great level of creative control over aspects such as lighting, camera motion, animation timings, cinematic effects, and video style. There’s also the ability to instantly translate the audio or text in the video to another language and open up the global appeal of your content. Adobe Firefly is one of the fastest ways to generate video ideas, making it perfect for B-roll, storyboards, or just to bring a bit of sparkle to your work. Videos run for 5 seconds, all in high-fidelity 1080p, are designed to be safe for commercial use and can be downloaded as MP4 files that are easy to share or import to other software. Adobe Express brings AI power to your existing videos If you already have video you’ve shot, then Adobe Express is a great platform on which to assemble, polish and output your finished mini-masterpiece. That doesn’t mean it’s old-school though, as Adobe Express is enriched with AI-powered tools that make editing smarter, faster and more inventive. At the heart of the software are one-click tools that streamline edits such as trimming, splitting, and adjusting the speed of footage. These are joined by the option to enhance speech, add captions automatically, or record vocals over the video – all directly within the app. Clip Maker is an amazingly useful new tool that goes through an existing video and pulls out the key moments to create short-clips for sharing on social media. Something that would have taken hours before is now done in minutes. Adobe As Firefly is baked into Adobe Express, you can also use the generative AI capabilities to convert a static scene into animated video, again without the need for any expert skills. When you’re done, there are templates available to format and post videos to social media platforms, with outputs available as GIFs or MP4. Invent new worlds without astronomical prices You might think that the vast array of creative and editing options on offer in Adobe Express and Adobe Firefly would make them out of reach for normal people, but that’s far from the truth. Adobe Firefly has a free tier where you can experiment with its amazing capabilities, and if you want to create more videos then there’s the Standard plan that costs only $9.99/£9.98 p/m. Sign up to Adobe Firefly today! Adobe Express also has a free tier that includes many of the editing tools mentioned above, with a Premium plan that adds advanced features and generative AI capabilities for $9.99/£9.98p/m. If you’re still not sure, then there’s a 30-day free trial so you can get to grips with all it has to offer before you sign up to a subscription. Try Adobe Express for Free Video is the communication method of the modern internet, so make yours the best they can be, all without having to learn loads of editing commands or investing in expensive software. Let Adobe Firefly and Adobe Express take care of the hard work so you can concentrate on being creative. Read...

The policy `is not just about technology - it`s about empowering our people`, the country`s acting information and communications technology minister says. Read...

At a glanceExpert`s Rating Pros Beginner-friendly setup (when its GPS works correctly) Traction Mode mapping is a great innovation Impressive obstacle avoidance Cons Very low maximum cutting height (for U.S. grasses) Persistent GPS issues Unrealistic advertised LoRa communication range (LoRa depends on line-of-sight) Our Verdict Price When Reviewed This value will show the geolocated pricing text for product undefined Best Pricing Today Best Prices Today: UBHome M10 Retailer Price $1,299 View Deal Price comparison from over 24,000 stores worldwide Product Price Price comparison from Backmarket Best Prices Today: Check today’s prices We’ve seen our share of robot mowers from large companies and well-capitalized startups, but very few crowd-funded ones. UBHome is an exception, raising HK $1.6 million (~$200,000 USD) this spring to bring its M10 robot lawn mower to market. The UBHome M10 features artificial intelligence combined with RTK (real-time kinetics) positioning, which reminds me of the Segway X3 series mowers, albeit with a different price tag and maximum yard size. The M10 costs just $1,499, which makes it one of the lower-priced robot mowers we’ve tested. The M10 uses a reference station and a LoRa (Long Range) gateway to determine position and communicate with the mower. In theory, no mower should have any problems maintaining a connection anywhere in my yard, but its unusual shape and plethora of obstructions make it a good test location. Mapping my entire yard with the UBHome M10 was so user friendly I was able to do it for the first time in many reviews. Installation and setup My yard quickly looked well-manicured thanks to the UBHome M10’s attention to detail. It spent extra time on the faster-growing portions.Ed Oswald/Foundry Getting the UBHome M10 set up wasn’t easy, not because the hardware was difficult to deal with or that the robot’s mapping controls were deficient—in fact, I loved the latter and will discuss them in more detail later—but because the M10 had persistent issues acquiring GPS signals. This occurred even in areas where the robot has a clear view of the sky. The M10 is super-reliant on GPS positioning, requiring 20 satellites just to install. That is a very high number of satellites to expect—I was maxing out at 15 to 17 at my location in Pennsylvania, which initially meant I couldn’t move forward with the review. I encountered a similar issue with a Yarbo mower, an issue that has plagued that platform for years: a refusal to connect to GPS satellites that are reliably in view. A view of the sky from the UBHome M10’s reference station, looking due south and tilted approximately 120 degrees. As you can see, apart from the horizon and a portion of the southwestern sky, there is a reasonably clear view at this location.Ed Oswald/Foundry The good news? Yarbo’s problems came down to a software issue, and I highly suspect the same thing is happening here. I eventually found another five or so satellites (although this part of the install was done a day later) and was therefore able to map my yard using the M10’s excellent in-app controls. Unlike many other mowers I’ve tested, the UBHome M10 doesn’t lurch with the slightest tap of its in-app controls. In areas where I wasn’t experiencing issues with GPS connectivity, I was able to map the terrain relatively quickly. It also has a slightly bizarre yet novel mapping method it calls Traction Mode, which involves looping a rope around the front grille of the M10 and pulling it behind you to create a map. If you struggle with remote control operation, this might be an attractive alternative (and it really works). UBHome deserves credit for thinking out of the box here. Mapping was so user friendly that I was able to use it to map my entire property for the first time in many reviews. I was also curious to see how well the LoRa technology could handle the more distant parts of our yard. UBHome says LoRa support endows its mower with a communication range of 576,000 square yards (138,000 square meters), a claim I was rightly skeptical of. What the company’s marketing materials don’t point out is that LoRa networks are heavily dependent on line of sight. So, that range might be achievable if you’re mowing a pasture, but having buildings in the picture quickly changes things. The UBHome app has a great user interface, and the ultra-precise positioning can tell you exactly where your mower has been. The barebones scheduling options, on the other hand, need a makeover.Ed Oswald/Foundry UBHome’s infrastructure consists of a gateway, an antenna, and a base station, which isn’t all that beginner friendly, but the company’s instructional materials clearly show all the steps you need to get started. While the production value of its video won’t earn an Emmy, it makes it abundantly clear what you need to do. UBHome’s app will also walk you through the process with helpful text directions. Using the UBHome M10 The UBHome M10 is a daytime-only mower due to the absence of LiDAR navigation or any headlamps. I assume most people reading this review aren’t likely to mow their lawns in the middle of the night, like I do, so I don’t see this as a negative. What’s most important is the cut, and even with all the installation difficulty, the UBHome redeemed itself in that regard. The M10 has a very low maximum cutting height of just 2.36 inches (6cm). That might be too low for the types of grass commonly found in American lawns. I generally keep our grass at 2.5 inches or higher, as anything lower tends to expose brown patches. I’ve noticed this quirk in many mowers that initially launched overseas. The M10’s front wheels move freely, so its rear wheels try to compensate, causing the mower to “waddle” slightly on open straightaways.Ed Oswald/Foundry It’s also because the U.S. itself is an aberration: The grasses that make up most of our lawns are a result of homesick American colonists favoring European grass species—and importing their seeds—instead of planting native North American grasses. The M10 does have a high-grass mode for when you need to cut grass that is significantly above the mower’s maximum cutting height, but UBHome warns that this setting could impair the mower’s obstacle avoidance. That said, I saw no substantial effects, even when the mower got close to the slope in my yard. That slope has been proven to be the Achilles’ Heel of so many rear-wheel drive mowers I’ve tested here. The mower has three mowing methods, the last of which is optional. The first is your traditional mow, where the mower takes long passes to cover as much ground as possible. In oddly shaped or narrow areas, you can use “spot mowing,” where the mower moves in a manner that covers spots close to the edge of the mowing area or where it may have missed. Finally, the optional “perimeter mowing” rides the edge of the mowing area to provide that finished look. You can monitor all of this in the app in real time. LoRa helps to ensure that the app’s positioning is exact, and it also results in a significantly longer mowing time than most robot mowers, but the results are worth it. There are two situations where you’ll encounter trouble with the M10: Roughly 2 to 3 hours before sundown, and in low-light conditions. To avoid this, I recommend mowing in the late morning or early afternoon.Ed Oswald/Foundry I experienced issues with sun blindness being misinterpreted as an obstruction when the sun was low in the sky. This is typical of robot mowers that aren’t equipped with LiDAR navigation. To get around the problem, program the mower to cut any portions of the lawn that will have the mower moving directly into the sun at least three hours before sunset. Mowing the easiest part of my yard went so well that I decided to tackle the more problematic areas next. I tempered my expectations given the amount of trouble I’ve had with other rear-wheel drive (RWD) mowers. Things went well, for the most part. I set the M10 to its slowest setting while cutting narrow strips of grass, which kept it on the median rather than in the street, where so many other mowers have ended up (the mower did fall off the curb while running at its standard speed). On slopes, the M10 seemed to correct the drift of its front wheels to either side much faster than any other RWD mower I’ve tested. The M10, however, lost its connection to the Reference Station about 300 feet through my detached garage. This wasn’t entirely surprising, given LoRa’s dependence on line-of-sight communication; in fact, it performed better than most of its competitors in this area. That said, it didn’t come anywhere close to the 138,000 square meters of communication range that’s advertised. The mower also had difficulty staying within its assigned boundaries while it was in that section, confusing a small strip down the alley as an edge, likely due to connectivity issues. Using the UBHome app UBHome’s app has an excellent user interface; it’s well-organized and clean, and I appreciate how the most critical actions are assigned to buttons at the bottom of the screen, so I can use the app with one hand. The scheduling portion of the app, on the other hand, is too barebones; it shows scheduled mow days and times, but not the areas to be mowed. The UBHome M10 is an AI+RTK mower that uses a trio of cameras and seven sensors to navigate your yard and avoid obstacles.Ed Oswald/Foundry A log show displays all recent actions and alerts, and there’s a remote control feature if you feel like driving the mower yourself. I experienced sporadic connectivity problems from time to time, which were usually remedied by restarting the app. While you can set mowing direction within the app, there doesn’t appear to be any type of automatic adjustment for each successive mow. That said, it might not be necessary, as the M10 will switch between modes based on the height of the grass and if it spots areas where grass is growing faster than others. Should you buy a UBHome M10 robot mower? With a price tag of less than $1,500, the UBHome M10 punches above its weight. Its dependence on GPS for navigation, however, comes with risks. The mower’s sporadic inability to find satellites in areas where it had a clear view of the sky is disappointing. Apart from that, the mower is surprisingly flawless and fun to use. I initially thought such out-of-the-box ideas as Traction Mode were silly, but when I used them, I discovered they’re an innovative way to speed up the most irritating part of robot mower setup. The Ml10 delivers outstanding cut quality, and its obstacle avoidance works well even in high-grass mode. This rear-wheel drive mower’s ability to successfully mow my lawn without falling down its slope is also notable. And for the most part, it stayed within its assigned boundaries, something many other mowers have failed to do, especially on my yard’s median strips. If you’re in the market for a robot lawn mower, most of your yard has a clear view of the sky, and you have a limited budget, the UBHome M10 is an option to consder. I do hope UBHome solves this machine’s GPS issues with a firmware update, because that’s holding back an otherwise excellent robot lawn mower. This review is part of TechHive’s in-depth coverage of the best robot lawn mowers. Read...

Microsoft has redesigned all the icons for its various Office applications (officially known as Microsoft 365 apps), this time focusing on color gradients and contrasts. The result? These new Microsoft Office icons are now more colorful, more curvy, and more approachable. We’ve known for some time that Microsoft Office apps were getting new icons (thanks to a leak). But they’re official as of today, according to Jon Friedman (CVP of Design and Research for Microsoft 365) in this Microsoft Design blog post. Microsoft is rolling out the new design to users with immediate effect. In the post, Friedman explains the redesign’s significance: When it comes to outsized impact, it’s hard to debate the almighty icon. No bigger than a postage stamp, these tiny symbols are gateways to entire experiences, distilling complex ideas, product abilities, and brand identities into a single, memorable image. By evoking emotion, sparking curiosity, and giving intuitive guidance, they make technology more accessible and approachable. The last time Microsoft updated the Office icons was in 2018. Below, you can see all the redesigned icons side by side: The 2025 redesigned icons for Microsoft 365 apps.Microsoft We find the following illustration from Microsoft particularly interesting, which shows the evolution of these icons over the years: The evolution of Microsoft Office app icons from 2001 to 2025.Microsoft And here’s a direct comparison of how the Word icon changed: Old versus new icon for Microsoft Word.Microsoft Microsoft stuck to four main principles in this redesign: Delightfully simple: To maintain familiarity while streamlining the visual experience, we graphically simplified the icons for clarity and reduced visual noise. Whereas Word’s icon previously used four horizontal bars, the new version uses just three, improving legibility at small sizes and creating more visual concision. Fluid shapes: We’ve moved away from bold, static solidity to embrace softer, more fluid forms. Sharp edges and crisp lines are replaced by smooth folds and curves, giving the icons a sense of playful motion and approachability. Rich and colorful: The color palette has been dramatically refined. Where gradients were once subtle, they’re now richer and more vibrant, featuring exaggerated analogous transitions that improve contrast and accessibility. This shift makes the icons feel brighter, punchier, and more dynamic. Instantly recognizable: Letter plates were much debated because they’re valuable real estate and icons following the core 10 Office ones no longer use them. Their brand equity is so strong, however, that we decided to keep them—maintaining our heritage while also modernizing them through a more cohesive visual integration with the overall design. The reference to the omnipresent AI is almost inevitable as well, with the design of the new Microsoft 365 icons clearly influenced by the Copilot icon. The new look is intended to represent a more connected design system and the influence of Copilot on Microsoft 365. Read...