A group of South Taranaki rangatahi are reconnecting with their culture in a first-of-its kind collaboration with the Western Institute of Technology. Read...

If you’re wondering what effect Intel’s blockbuster deal with Nvidia will have on its existing product roadmaps, Intel has one message for you: it won’t. “We’re not discussing specific roadmaps at this time, but the collaboration is complementary to Intel’s roadmap and Intel will continue to have GPU product offerings,” an Intel spokesman told my colleague, Brad Chacos, earlier today. I heard similar messaging from other Intel representatives. Nvidia’s $5 billion investment in Intel, as well as Nvidia’s plans to supply RTX graphics chiplets to Intel for use in Intel’s CPUs, have two major potential effects: first, it could rewrite Intel’s mobile roadmap for laptop chips, because of the additional capabilities provided by those RTX chiplets. Second, the move threatens Intel’s ongoing development of its Arc graphics cores, including standalone discrete GPUs as well as integrated chips. We’re still not convinced that Arc’s future will be left unscathed, in part because Intel’s claim that it will “continue” to have GPU product offerings sounds a bit wishy-washy. But Intel sounds much more definitive on the former point, in that the mobile roadmap that you’re familiar with will remain in place. So far, Intel’s public roadmap calls for Intel’s “Panther Lake” processor to debut this fall, probably shipping in early 2026. Intel’s been talking about that chip for months and months, and there’s no reason to believe those plans will change. Intel has also publicly disclosed Nova Lake, the next-next-generation mobile processor for laptops, which is also due in late 2026 and will probably enter laptops in early 2027. According to a leaked roadmap from a Spanish PC maker, Wildcat Lake might be a 2026 part, too. What we’ve been told, however, implies that any work that comes out of the Nvidia-Intel partnership will be additive. Essentially, there will be additional products that will be added to the roadmap: premium products, attached to markets like consumer, gaming, creator, and business. To me, that sounds like Intel could be adding a premium version or option to its established lines. Remember, we don’t know what Panther Lake or Nova Lake will be designed as. We do know, however, that Meteor Lake, the first-generation Core Ultra chip, was designed with a specific GPU tile. One might imagine that Intel could ship a processor with a GPU tile that could either be Intel’s own Arc chip, or a replacement architected by Nvidia. Whether that would be possible or not is unknown — that’s just speculation. Though Intel and Nvidia have been working on this partnership for about a year, according to Nvidia chief executive Jensen Huang, we also don’t expect to get that much information about future products anytime soon. Though, with a rabid technology press corps eager to follow up on the question financial reporters didn’t ask during the Nvidia-Intel press conference, who knows what will emerge? Read...

Google is tying Gemini and Chrome closer together, allowing Gemini broader access to your Chrome tabs while quietly turning the address bar into an entry point for its AI Mode. Eventually, it’ll add agentic browsing to Chrome as well. The latter point is likely what Google wants to signal to the broader market, since keeping pace with (or surpassing) other browser makers deploying agentic AI is seen as a leadership move. But agentic browsing will debut in the coming months, while Gemini’s tighter integration with Chrome arrives soon, even today. A day ago, you could type “best laptops” in Microsoft Edge on your PC and receive a summary of Copilot’s findings above a list of links. In Chrome, provided that you didn’t have AI Mode enabled, Google would return just that list of search results. Later this month, Google is making the “omnibox” a repository for AI Mode: what was once the “address bar” over time became a search box, and now it’s being transformed into something more. Essentially, Google seems a step away from merging Gemini and Search within Chrome, as it said in May–it just hasn’t quite gotten there yet. Then, users were able to try out Gemini in Chrome with a paid subscription, allowing them to see the Gemini “sparkle” hovering above the content on the page. Today, Google has removed that paid limitation. Now that Gemini is being added to Chrome, users will be able to “ask” queries about the page via a sidebar. If you “search” via the omnibox, you may receive an “AI Mode” result. AI Mode in the omnibox will roll out later this month. Users can toggle between a traditional search and an AI Mode request by clicking the small AI Mode “chip”. It’s still unclear why Google has both an “omnibox” for web pages and search queries, plus a separate “box” dedicated to Gemini queries. Google Gemini in Chrome, however, can now draw on more of your own information for context: not just what’s on the page, but also your other open Chrome tabs, your browser history, and even Google apps like Gmail and Calendar. Querying your browser history sounds a bit like Microsoft Recall, but without the fanfare or controversy. Here, Google suggests that you use a prompt like “what was the website that I saw the walnut desk on last week?” or “what was that blog I read on back to school shopping?” Here, Gemini within Chrome is not only summarizing the YouTube video but also creating an event.Google If you have a question about a page you’re viewing, Google will supply some suggestions. Google’s road to agentic browsing Like Microsoft in Edge and Opera with its own version, Google has also demonstrated agentic browsing. However, it looks like every other agentic browser demo I’ve seen: give it a task and off goes the AI agents to complete it. In this case, Google already offered a sneak peek of agentic technology in May as “Project Mariner.” When the process (in this case, a shopping task) completes, you’re given a chance to look it over and then make a final decision to pay, or not. The idea is that agentic AI could be used to plan trips, handle shopping, or even combine the two. In this case, however, Google’s agentic AI will be limited to English-language web pages, according to company executives. The feature will roll out “in the coming months,” Google said. Google AI-based security, too Google’s AI is being applied to personal web security. It’s not just detecting scams that might trick you into downloading software, it’s also blocking sites that push fake contests or sweepstakes, and even cutting down on low-quality sites that request unusual permissions, like camera access. Google also said that it will use AI to detect compromised passwords that were leaked in a data breach. Today, it simply alerts you and points you to the site to change them. On certain sites, Google says it can now reset and securely store the updated password for you. There are some smart uses of AI here, but it also feels like Google is slowly easing us into a future where AI answers our queries instead of showing a list of links. What will that make Chrome, which began life as a web browser and now is evolving into a showcase for Google’s AI? The jury’s still out on that one —well, not legally. Read...

At a glanceExpert`s Rating Pros $300 MSRP seems reasonable Thunderbolt 5 Three-display capability, or two displays plus an SSD Thunderbolt Share is included Stable Cons You’ll probablly need to buy display adapter cables No active cooling, but it didn’t seem to need it Our Verdict Plugable’s 11-in-1 Thunderbolt 5 (TBT-UDT3) docking station is a solid all-around TB5 dock with a great mix of features and ports. Pair it with a TB5 SSD and you’ve got impressive storage performance. Price When Reviewed This value will show the geolocated pricing text for product undefined Best Pricing Today Best Prices Today: Plugable 11-in-1 Thunderbolt 5 Docking Station Retailer Price $299.95 View Deal Plugable $299.95 View Deal Price comparison from over 24,000 stores worldwide Product Price Price comparison from Backmarket Best Prices Today: Check today’s prices Plugable’s 11-in-1 Thunderbolt 5 (TBT-UDT3) docking station is a moderately priced Thunderbolt 5 dock that can future-proof your PC for years to come. While it might not offer the dedicated display ports of older docks, its integrated Thunderbolt Share delivers file sharing and a KVM-like experience, for free. What Plugable doesn’t offer is integrated storage or active cooling, saving your wallet some additional cash. Just keep in mind that you may have to make up for that by buying some additional display cables. If you’re interested in future-proofing your PC, the combination of the high-speed display options Thunderbolt 5 offers, plus Thunderbolt Share, and the additional performance a high-speed external TB5 SSD offers makes this dock really intriguing. Plugable TBT-UDT3: Design and build Plugable calls this dock the Plugable Thunderbolt 5 Dock with 3x Thunderbolt 5 ports, 140W Laptop Charging, or TBT-UDT3. It’s a rather compact Thunderbolt 5 docking station, measuring 6.9 x 1.6 x 3.1 inches. I’ve always been somewhat partial to docks which utilize vertical space, such as the HP Thunderbolt G4 Dock, simply because my desk doesn’t have that much space on it. Plugable’s dock can fit into a vertical stand included in the package. This maximizes your available desk space even more. The TBT-UDT3 is made of aluminum and ABS plastic. These two materials weave their way in and out of the chassis. You’ll find metal coating the top and bottom (if mounted flat, not vertically) which feels necessary. The dock was fairly warm inside my air-conditioned office, using two 4K displays connected via the dock’s included Thunderbolt 5 cable. (That cable measures 39 inches, or 1 meter long.) Some TB5 docks include active cooling, with an external fan; the TBT-UDT3 does not. That’s possibly a corner Plugable cut, but it doesn’t seem to have affected its stability at all. The rear of Plugable’s 11-in-1 Thunderbolt 5 (TBT-UDT3) docking station, with two downstream Thunderbolt 5 ports, and upstream TB5 port to your computer, plus 2.5Gbps Ethernet, USB-A, and locking slots.Mark Hachman / Foundry I usually refer to docking stations without dedicated display ports as hubs, not docks. In this case, Plugable’s TBT-UDT3 includes two Thunderbolt 5 ports on the rear of the dock, and one in front. All three can be used for display connections. Make sure you choose the proper cable for the job. A Thunderbolt 4 dock at 60Hz can use a USB-C to HDMI adapter that supports 4K60 output. That works fine with this Thunderbolt 5 dock, too. But a TB5 dock (like this one) should output a 4K display at 144Hz per port–you’ll need a slightly more expensive cable (about $25 apiece). Specifically, Plugable’s dock puts its power button on front, lit by a bright white LED which, accidentally or not, leaks into the internal ports, giving them a faint glow and making them easier to insert connections inside a dim room. The front of the dock also includes both a microSD and SD card slot at 312MB/S UHS-II speeds, a 10Gbps USB-A port, a Thunderbolt 5 port, plus a 3.5mm headphone jack. On the rear are two more Thunderbolt 5 ports, both a 5Gbps and 10Gbps USB-A port, the downstream TB5 connection to the PC, a 2.5Gbps Ethernet jack, and two lock ports. The front of Plugable’s 11-in-1 Thunderbolt 5 (TBT-UDT3) docking station is simpler, with just an external TB5 port, plus SD/MicroSD and a headphone jack.Mark Hachman / Foundry Plugable’s dock doesn’t really offer charging capabilities — one rear USB-A port supplies 7.5W — but you can certainly plug in a phone to an unused USB-C/Thunderbolt 5 port, which is rated for 15W of power for a phone or an external device of some sort. It actually provided 13.9W under load. That used to be enough to fast-charge a smartphone like a Samsung Galaxy S19, but can’t really keep up with the high-speed charging power used by recent iPhones or Android phones, which fast-charge at 45W or higher. Keep in mind that the full charging capabilities of Thunderbolt 5 go up to 240W. This dock taps out at 140W. On the other hand, Thunderbolt 5 is (for now) confined to gaming laptops, and those laptops generally pull close to 400W or more while gaming under load. Put another way, even 240W isn’t going to cut it for gaming, right now. That power supply might be suitable for tomorrow’s content-creation/light gaming notebook, but not now. Plugable’s 11-in-1 Thunderbolt 5 (TBT-UDT3) docking station, deep into testing. Your desk will hopefully be more organized than this is. Note how the vertical orientation saves space, however.Mark Hachman / Foundry Plugable’s dock also includes an unexpected bonus: Thunderbolt Share, a technology that came and went without a lot of fanfare from the mobile community. Using Thunderbolt Share, two PCs can share files over a Thunderbolt connection, or two PCs can share a single screen. However, Thunderbolt Share requires that the PCs download the Thunderbolt Share app. One of the devices must also have a Thunderbolt Share license — or, in this case, the dock. Only then is Thunderbolt Share allowed to work. (You can see our video demonstration of Thunderbolt Share, here.) Plugable TBT-UDT3: Performance For these tests, I used Razer’s Blade 18, which includes a Thunderbolt 5 port as well as a separate Thunderbolt 4 connection. On that laptop, Plugable’s dock seemed almost perfectly stable. It connected to a pair of 4K160 displays at the dock’s rated speed of 144Hz, my test bed’s default configuration. It also connected perfectly to a third 4K160 display at 144Hz, too, as it should. In this scenario, however, my test laptop’s display wouldn’t light up until I rebooted. After it did so, all three external displays lit up at 144Hz, plus the laptop’s display. The laptop/dock combination couldn’t handle streaming 4K60 video to all displays, but static Web pages loaded with no issue. I also connected it to my daily laptop, with a Thunderbolt 4 port, and I had no issues using Plugable’s same dock with the same 4K displays at 60 Hz. When I accidentally powered off the dock when trying to insert it into its vertical stand, there was a bit of “panic,” where the displays cycled through and flipped on and off for a few seconds. That was user error, however, and the dock and the connected displays worked quite well thereafter. While the dock had some issues bringing up the displays when connected to an older TB4 laptop that was resuming from sleep, that problem did not manifest on the Blade 18 and its TB5 port. I’ve been testing Thunderbolt docks for several years using a standardized methodology. Thunderbolt 5, however, requires an update to my test procedures. I stream 4K video at 60Hz across two displays, then three — Plugable’s dock handled it like a champ without dropping more than a handful of frames. Streaming data from an attached SSD, though, is a bit more challenging with a higher-bandwidth Thunderbolt 5 dock. OWC kindly provided us with an Envoy Ultra SSD, rated at Thunderbolt 5 speeds. That puts more pressure on the dock itself to keep up. To date, there just haven’t been that many Thunderbolt 5 docks available. Most of my reviews cover Thunderbolt 3 and 4, so this dock, along with the Sonnet Echo 13 Thunderbolt 5 Dock, represent a small cadre of the fastest docks available. As it happens, the performance of the two is roughly comparable. I ran PCMark’s storage test against the Envoy Ultra, both directly connected and also connected to a Thunderbolt 5 port on the dock. Directly connected, the Envoy Ultra returned a bandwidth score of 469MB/s or a score of 3,202. Connected to the dock’s TB5 port, performance dropped to 437MB/s or a score of 2,920. That’s a 7 percent drop, and basically identical to the 436MB/s bandwidth score that the Sonnet Echo 13 yielded when the Envoy Ultra was connected to its Thunderbolt 5 port. While streaming the two 4K videos across the integrated Ethernet port, performance dropped to 402.77 MB/s, since some of the bandwidth was taken up by the Ethernet port. Only my folder copy test, which measures how long it takes to copy a bundle of files from an SSD through the dock to the desktop, showed any real difference from the Sonnet: 13.9 seconds for the Sonnet, and 16.9 seconds for the Plugable dock, or 14.2 seconds vs 18.96 seconds while streaming. I tested Thunderbolt Share, which crashed the first time I tried it. (I didn’t notice that OneDrive was syncing in the background.) Using one laptop to control another worked fine. I was able to transfer my folder of files in about 54 seconds, slightly faster than my Thunderbolt 4 dock tests. That’s reasonable, given that both a Thunderbolt 4 laptop and a Thunderbolt 5 laptop were connected to the TBT-UDT3. Mark Hachman / Foundry It might be worth noting that that the rival Sonnet Echo 13 includes a 2TB integrated SSD and costs $439 at press time; Plugable’s dock does not include an SSD, and OWC’s Envoy Ultra (2TB) costs about $300 alone at press time. On the other hand, the internal bandwidth of the Sonnet’s internal SSD was 279.84MB/s, substantially less than the Plugable + OWC SSD combination. The bandwidth of Sonnet’s internal SSD is similar to what you might expect of external gaming SSDs. On the other hand, the read and write speeds of the OWC Envoy Ultra plus the Thunderbolt 5 connection push upwards into the speeds of a good internal PCIe4 SSD, and that’s worth something, too. The one thing I didn’t test is how well this dock accommodates an external GPU. That’s a capability that’s built in (again) to Thunderbolt 5, but I don’t think an eGPU makes a compelling argument yet if TB5 ports are only found within gaming PCs already equipped with discrete GPUs. Plugable TBT-UDT3: Conclusion I would, yes. Generally I hope for premium Thunderbolt docks to be in the $250 range or a little lower, and $299 seems pretty reasonable for a premium dock — though you may have to add display adapter cables to that price. A two-year warranty is included. Though this dock does offer access to three displays, you might find that connecting two displays plus a high-speed SSD works best for you. Interestingly, all of the Thunderbolt 5 docks I’ve seen do not add dedicated display ports, as their TB4 and TB3 offerings did. Don’t forget about Thunderbolt Share, either. It’s not a technology you might use often; after all, you can always connect a hard drive to the dock, copy a file to the drive, replace the laptop with another, and download the file. Still, it’s an interesting twist that most docks don’t offer. If you’re interested in future-proofing your PC, though, the combination of the high-speed display options Thunderbolt 5 offers, plus Thunderbolt Share, and the additional performance a high-speed external TB5 SSD offers makes this dock really intriguing. I really liked the flexibility the SSD inside the Sonnet Echo 13 offered, but Plugable offers an alternative with a different but very viable perspective. Read...

Adjustable actuation is what lets you choose a custom “trigger” point for each key, and it’s the buzziest feature in gaming keyboards. Logitech is hopping on the train now, even if it had to do a lot of jogging to catch up. The newest version of the G515 keyboard, now with magnetic switches, is christened the Rapid TKL—and it’s shipping now for $170. Ouch. The G515 Rapid TKL looks a whole lot like the existing G515 designs, with a distinctive low-profile tenkeyless layout that’s just 22mm tall. Each one of those “analog” switches can be adjusted at 0.1mm increments in the travel (down to 2.5mm total), including “rapid trigger” capabilities, a feature that many competitive gamers want. (Razer calls it “snap tap,” if you’ve heard that one.) Two different functions can be bound to each key at different actuation points. Logitech Other features include keycaps made out of premium PBT keycaps (“Nice,” says my inner keyboard nut), a steel top plate, and a complete lack of wireless. That might feel like a step backwards for a top-of-the-line design, but adjustable actuation switches can drink down a lot of battery power. Even so, it’s an omission that hurts with a keyboard this expensive. Logitech Logitech also announced new gaming mice. The Pro X2 Superstrike features a “haptic inductive trigger system,” which Logitech says is a combination of adjustable actuation in the mouse switch and rapid trigger capabilities. That’s a lot of buzzwords, but the technology underneath uses copper coils to generate an electromagnetic field with a 0.6mm switch that can be user-adjusted at 10 different points, including rapid trigger reset. Logitech What does that mean for PC gamers? In addition to adjustable actuation points on a mouse—which seems like something that’s almost impossible to feel to me, but I haven’t gotten this thing in hand—Logitech says that the mouse can reduce click latency by “9 to 30ms.” Other highlights for the wireless, shooter-style mouse include Logitech’s 44,000 DPI Hero 2 sensor, 90 hours of battery life, and 8,000Hz polling. The mouse is 60 grams, impressively light for all that tech. Be prepared to pay for it. The Pro X2 Superstrike mouse will cost a hefty $180 when it launches in the first quarter of 2026. Read...

Mark Zuckerberg recently unveiled the Meta Ray-Ban Display and Neural Band, which are smart glasses with a color high-resolution display and wristband controls. It sounds like science fiction, but it’s more of an inconspicuous integration of tech into everyday life. Ray-Ban Display and Neural Band The glasses look like classic Ray-Bans, but they have a discreet display on the side of the right lens. It supposedly doesn’t get in the way when looking through the glasses and only appears when needed to display messages, photos, translations, or AI responses. Meta / Ray-Ban The smart display’s resolution is 600×600 pixels with a 20-degree field of view and 42 pixels per degree—sharp enough for everyday use by everyday people. The display’s brightness adjusts from 30 to 5,000 nits and the refresh rate goes up to 90Hz, according to Meta. A new feature here is control via the supplied Meta Neural Band. This wrist-worn device recognizes muscle movements on the wrist and converts them into commands—a swipe of the thumb is enough. Meta / Ray-Ban Based on four years of research with 200,000 participants, the AI glasses should be intuitive to control. The glasses themselves have a battery life of up to 6 hours, and up to 30 hours with a charging case. The Vectran material is as strong as steel but flexible, and Meta also claims that the AI-powered glasses are waterproof with an IPX7 rating. Other features include WhatsApp and Messenger integration, video calls, navigation, live subtitles, and music control. The 12MP camera films in 1440p at 30 FPS, while the internal 32GB storage can store up to 500 photos or 100 30-second videos. Two open-ear speakers and five microphones ensure good audio and recordings. The AI glasses are compatible with iOS 15.2 and Android 10, and corrective lenses from -4 to 4 diopters are also possible. Meta offers the Ray-Ban Display in black and sand colors and with transition lenses, in two sizes: Standard and Large (144mm to 150mm width). Pricing and availability The price of the Meta Ray-Ban Display at launch will be $799 including the Neural Band, and the AI glasses will initially only be available to buy in the US starting September 30th, 2025. Europe is to follow in early 2026. Meta sees these smart glasses as an intermediate step between camera glasses and holographic AR models—technology that you wear inconspicuously in day-to-day life instead of constantly staring at your smartphone. It remains to be seen whether Meta’s AI glasses will catch on, but neuro-control is an exciting development. Read...

Surprise! We woke up this morning to a blockbuster mashup between Intel and Nvidia. Team Green invested a cool $5 billion into Intel, and in exchange, the two companies will be co-creating consumer and data center x86 processors interwoven with Nvidia’s RTX graphics. Human sacrifice, dogs and cats living together… MASS HYSTERIA! It’s simultaneously a shocking shakeup of the PC chip triumvirate (AMD must be fuming), a much-needed lifeline for struggling Intel, and a recipe for a potentially exciting future – the world’s foremost graphics pioneer joining forces with the company formerly known as Chipzilla. Imagine the possibilities! But I also have to ask myself at the same time: What does this mean for the future of Arc, Intel’s own in-house graphics project? Intel Arc’s short history shows promise… Arc is still in its infancy. Intel famously canceled its early “Larrabee” graphics architecture in the 2000s, which became a liability after the rise of Bitcoin and AI demonstrated the powerful potential of GPUs. Intel realized it missed the boat and rushed – slowly, at times – to orchestrate both the Arc brand and the Xe graphics architecture girding it. The first Arc graphics cards launched in just October 2022, delivering great value for its price despite an onslaught of annoying bugs. Intel diligently fixed those bugs over time, and by the time the second-gen Arc B580 launched in late 2024, we called it “the first worthy budget GPU of the decade.” And Arc’s underlying Xe graphics architecture now powers the integrated graphics in Intel’s CPUs too, bringing a notable spike in laptop gaming performance. …but potentially shaky foundations Software bugs aren’t the only problem that rears its head when you’re trying to break into a new field where Nvidia and AMD have a decades-long lead. Intel’s GPU hardware prowess isn’t up to par with its rivals yet either; this shows in the size of the discrete Arc GPU die sizes. Bigger dies are much more expensive to make. The $250 Arc B580’s die size is a relatively massive 292 mm. By contrast, Nvidia’s RTX 4060 was around 150 mm, while the RTX 5060 is around 181 mm. That matters. Intel’s Tom Peterson (a frequent guest on our Full Nerd podcast!) admitted last year that the Arc B580 is a “loss leader” – a product that costs the manufacturer money to sell, in the hopes of attracting customers. Intel figured it was worth eating that loss to build for a stronger GPU future. Intel’s partnership with Nvidia suddenly throws that into question, even though the company says no major changes are currently planned. “We’re not discussing specific roadmaps at this time, but the collaboration is complementary to Intel’s roadmap and Intel will continue to have GPU product offerings,” an Intel spokesperson told me. Intel needs strong GPUs to battle Nvidia in the data center, because AI is where the real money is. The consumer Arc cards are stepping stones to that goal. Now Nvidia is investing $5 billion into Intel – roughly a 5 percent stake, if the recent government investment is any indicator — to integrate RTX graphics into at least some Intel consumer CPUs, and to create data center solutions that interweave Intel’s x86 chips and Nvidia’s class-leading graphics. If a major investor is bringing GPU technology to Intel’s chips, spanning from consumer to enterprise, and gifting Intel a lifeline in the data center where AMD has been eating Intel’s lunch. Is Arc worth investing in separately anymore? Intel Arc’s certain yet uncertain future From a strategic standpoint, there’s certainly a case to keep Arc around. What if the Nvidia relationship suddenly goes sour despite the big money? The company is known to be a ruthless partner. Keeping Arc and Xe in motion protects against a potential future where Nvidia pulls the rug out from underneath Intel, especially since Xe (and seemingly this Nvidia partnership) touches everything from laptops to data center GPUs. Continued investment into internal GPUs makes so much sense for Intel’s future. But I’m not sure that’s what’s going to happen. Bright futures have a way of bumping into ugly realities. Part of the reason this Intel-Nvidia mashup even happened is because Intel lost its manufacturing lead and has been hemorrhaging cash (and CEOs) ever since. Nvidia’s deal follows in the footsteps of the U.S. government taking a 10 percent stake in the company to help it stay afloat, and Intel selling off subsidiaries like RealSense cameras and Altera’s FPGA chips. Christian Wiediger / Unsplash Intel is scrambling to stay relevant, and Nvidia’s partnership is a major shot in the arm – not least by likely infusing Intel’s beleaguered 14A process, the current crown jewel of Intel’s foundry arm, with work from Nvidia and other companies inspired by Nvidia’s faith. Either way, don’t expect major announcements from Intel (who I’ve asked for comment) any time in the near future. “I don’t expect these platforms for 2-3 years,” Patrick Moorhead, an analyst who founded Moor Insights & Strategy and formerly served as an AMD executive, told me via direct message. “Both companies said there are no roadmap changes… on either side. “Now… what will the demand be for Arc be once these are in market by customers remains to be seen.” Xe and Arc have driven much-needed competition in the entry-level graphics card market this turbulent decade. I hope they manage to stick around. If not, Nvidia’s $5 billion investment could not only get the company a foothold in the x86 markets, but also drive a competitor out of the market. If that happens, that sky-high price tag will wind up looking like a downright bargain in the rearview mirror. In the meantime, the Intel Arc B580 remains the best budget GPU of the decade. Read...

Nvidia is on top of the world right now, riding waves of investment in “AI” and becoming one of the most powerful and most profitable companies on the planet. Intel? Not so much. The company has been struggling in sales and performance for more than a year. So color me shocked when Nvidia and Intel announced a joint venture this morning that could be huge for both. Nvidia is investing $5 billion into Intel stock—a comparatively small slice for both companies—and approximately half of what the United States government invested in it after president Trump and other politicians demanded action on Intel’s CEO. But the bigger news coming from the press release is that Nvidia and Intel will partner on new chips for both data centers and consumers. The so-called “x86 RTX” chips will integrate Nvidia-designed graphics and AI chiplets into Intel CPUs. “This historic collaboration tightly couples Nvidia’s AI and accelerated computing stack with Intel’s CPUs and the vast x86 ecosystem—a fusion of two world-class platforms,” said Nvidia CEO Jensen Huang. Nvidia has been unassailable on the discrete graphics card front for years, now commanding over 90 percent of sales for desktop add-in boards in an effective monopoly, and dominating both sales and discussion for anything related to graphics, gaming, and “AI” industrial processing. Its longtime rival AMD has struggled to hold onto what market share it had, slipping even as it reportedly sells every GPU it can make. Things are better for AMD on the CPU side, where it’s gaining ground against Intel on the back of strong sales for laptop and desktop chips, especially its well-regarded X3D gaming series. Intel has been trying to enter the discrete graphics market for the last three years. But despite impressive gains in performance right out of the gate, the Arc series of desktop graphics cards has made barely a blip, falling to nearly zero percent market share. Exactly what will happen to Arc if Intel starts to co-brand Nvidia integrated graphics isn’t quite known. Nvidia’s meteoric rise to the top of both the chip market and the technology world has recently hit a highly visible snag, as reports indicate that the Chinese government is blocking purchases of its chips. The incredibly lucrative market was already showing some speed bumps as embargoes limit the performance of exported chips and China invests in domestic production to leverage its incredible energy and industrial infrastructure to compete with rivals like the United States and Taiwan. Intel and Nvidia will hold a joint press conference this afternoon, broadcast live on the web. Read...

“Click To Do” is Microsoft’s latest AI feature and selling point for Copilot+ PCs. Now that Windows Recall has taken a backseat after so much privacy criticism, Microsoft is turning the page—and this time it’s all about a special shortcut for accessing contextual AI actions. To use Click To Do, you just hold down the Windows key on your keyboard and click once with the left mouse button. You can also press Windows key + Q if you’d rather use a keyboard shortcut. (If nothing happens, that means you aren’t on a Copilot+ PC.) When Click To Do is activated, an outline will appear around your screen and Windows will highlight all text and images on your screen, making them selectable and allowing you to perform actions with them. Not only that, but Microsoft keeps adding more actions to this menu! Click To Do started with Windows Recall Funnily enough, Click To Do began its life as a feature built into Windows Recall. It let you take actions on text that appeared in the snapshots that Recall automatically took of your screen. However, after Microsoft pivoted from Recall, Click To Do became its own standalone feature. As far as privacy goes, Click To Do feels like the anti-Recall: it doesn’t do anything in the background, and you must choose to use it. Chris Hoffman / Foundry When you activate Click To Do, it takes a screenshot and lets you interact with it. When you select actions like “Summarize Text,” all of it happens right on your PC with your PC’s neural processing unit. But while most actions happen right on your PC, there are some actions—like “Search the web,” “Visual search with Bing,” and “Ask Copilot”—that will send data to Microsoft’s servers for processing. Fortunately, nothing leaves your PC unless you intentionally use such actions. Click To Do lets you feed text to AI models Click To Do uses optical character recognition (OCR) technology to scan your current screen and make text selectable. Basically, it’s taking a screenshot and letting you interact with elements within it. For example, if you select an email address, the Click To Do menu presents “Send email” to compose an email. If you select a website URL, you can choose “Open website” to launch it in your default web browser. (Thankfully, this doesn’t always use Microsoft Edge!) If you select over 10 words of text, things get more interesting. You’ll get a variety of actions that use the Phi Silica language model that runs on your Copilot+ PC’s NPU to perform actions like summarizing text, creating a bulleted list, or rewriting the text in different tones. Chris Hoffman / Foundry This is one of Microsoft’s first attempts at integrating NPU-powered text actions on a Copilot+ PC. Because it’s based on a screenshot, you can only send so much text to the language model at once. That should improve the experience, because let’s be honest: those local language models that run on your Windows PC are nowhere near as powerful as a cloud-based large language models like ChatGPT. (If you’re using a chatbot app in your browser, you’ll wonder why you’re bothering to use a worse app that runs entirely on your PC. That’s a core problem with so many Copilot+ PC local AI features.) Chris Hoffman / Foundry There’s also an “Ask Copilot” action that will send your selected text to Microsoft’s Copilot AI chatbot, and a “Draft with Copilot in Word” action to start working on a Word document with Microsoft’s AI. You can send information to a chatbot in the cloud here, too—but only if you’re using Microsoft’s Copilot (for home users) or Microsoft 365 Copilot (for businesses). Yeah, I know, it can be confusing. Click To Do exposes AI image tools, too Click To Do isn’t just for doing things with text—it also takes actions with images. This is where Microsoft’s other goals for Click To Do start to become clear. When you click on an image, you can select actions like “Blur background with Photos,” “Erase objects with Photos,” or “Remove background with Paint.” Chris Hoffman / Foundry These are all AI-powered image actions that would normally be scattered throughout Windows across different applications, but Microsoft is exposing them here through the Click To Do interface. And there’s also an “Ask Copilot” action here, so you can send an image to Microsoft’s Copilot AI and start a conversation about it there. There’s still no shortcut for the impressive Super Resolution feature, though. That would be really useful! You can disable Click To Do if you want If you don’t want Click To Do for whatever reason—and let’s be honest, that “hold the Windows key and click” shortcut could easily get in the way when playing some PC games—you can turn it off. Chris Hoffman / Foundry To do so, head to Settings > Privacy & security > Click to Do. Flip the switch here to turn it off. (While Click To Do is activated by default, it’s only available on Copilot+ PCs, so you won’t see it on your average Windows 11 PC.) What’s coming in the future? Microsoft has been spending a lot of time adding feature after feature to Click To Do. For example, Click To Do will soon have an integrated Copilot prompt box where you can select text/images, type a prompt, and then click “Ask Copilot” to send the selected content to Copilot along with the prompt you typed. You will also be able to select a mix of text and images, describe an image with an on-device large language model, and send text to the Microsoft Reaching Coach app. Click To Do will also soon be able to detect tables on your screen so you can send them right to Microsoft Excel. This mix of features shows what Microsoft wants Click To Do to become: the “one click to access AI anywhere in Windows” action. Microsoft hasn’t added file integration to Click To Do yet, but lots of actions for right-clicking files and sending them to Copilot and other AI tools are popping up in File Explorer’s context menu. As for me? I’d rather copy and paste After using so many half-baked Copilot+ PC features, it’s nice to see something with long-term potential. However, even though I use a Surface Laptop when I’m away from my powerful desktop PC, I must confess: I never actually use Click To Do, just like I never use Recall. Click To Do is an interesting idea, but I still find myself copy-and-pasting text and images instead. I can copy-paste text and images into any app, with or without AI tools. And Windows 11’s built-in Snipping Tool is already great for extracting text from screenshots and capturing images for me to send to other apps or whatever else. Maybe this will change in the future. Maybe not. But one thing’s for sure: Microsoft’s move away from Recall towards Click To Do is a smart one, and if it ends up paying off, it will pay off big. Further reading: Should you buy a Copilot+ PC? What to know Read...

New Zealand pledges NZD $70 million over seven years to advance AI research, boost innovation, and create high-value jobs in health, agriculture, and technology. Read...