Nvidia is reportedly planning on making Arm-based chips for Windows PCs. According to Reuters, the company has “quietly begun” taking on Intel by designing processors compatible with Microsoft’s operating system. The report says Nvidia could begin selling the chips as early as 2025.

The move is part of Microsoft’s broader goal of challenging Apple with Arm-based processors for Windows PCs. According to preliminary Q3 data from research firm IDC, the iPhone maker has nearly doubled its market share since launching the Arm-based Apple Silicon three years ago. The company’s in-house Mac chips balance performance (including on-chip AI tasks) and battery life to a degree that Intel’s processors have yet to match.

When reached via email by Engadget, Nvidia spokesperson Hector Marinez said the company declined to comment.

Nvidia has recent experience with Arm-based chips, but they’re designed for data centers. Nvidia announced plans to buy Arm Holdings in 2020 for $40 billion. However, the company pulled the plug on the deal in early 2022. 

AMD is also reportedly preparing to launch Arm-based PC chips as early as 2025. The two companies will join Qualcomm, which has made Windows laptop processors since 2016. Reuters reports that Qualcomm’s exclusivity agreement with Microsoft for Arm-based Windows chip designs will expire in 2024, opening the door to new challengers soon after. Windows on Arm hasn’t exactly been a rousing success to date.

“Microsoft learned from the 90s that they don’t want to be dependent on Intel again, they don’t want to be dependent on a single vendor,” Jay Goldberg, chief executive of consulting firm D2D Advisory, told Reuters. “If Arm really took off in PC (chips), they were never going to let Qualcomm be the sole supplier.”

As Microsoft (along with much of the tech industry) bets its future on generative AI, the upcoming chips will unsurprisingly focus heavily on it. The company has reportedly urged chipmakers to bake advanced AI-powered capabilities into their silicon. Microsoft recently launched Windows Copilot, which provides an OpenAI-powered chatbot in a persistent sidebar to respond to contextual queries anywhere in Windows.

Qualcomm is working with Google to develop a RISC-V platform that can power Wear OS devices. RISC-V is an open standard instruction set architecture (ISA), which Arm describes as "part of the abstract model of a computer that defines how the CPU is controlled by the software." Since it's open standard, it could lead to open-source chip designs and could lower the costs for companies that want to build their own chips. That means more companies entering the industry and more devices on the market, which could foster innovation and give rise to more affordable options. 

Google came out in support of RISC-V earlier this year in hopes that it could compete on equal footing with Arm one day. However, the technology still has a long way to go to reach that status. According to Ars Technica, Qualcomm's could be the first mass-market RISC-V Android chip ever, if all goes according to plan. As the publication notes, a RISC-V platform represents a big threat to Arm, whose CPU architecture dominates the smartphone industry. Things had been pretty unsteady for Arm over the past couple of years, with its parent company Softbank trying to sell it to NVIDIA until regulators stepped in and prevented the acquisition from happening. The company ended up registering for a US IPO instead and raised $4.87 billion in September.

In its announcement, the chipmaker said its RISC-V Snapdragon Wear platform will power the next generation of Wear OS devices. It's planning to commercialize the technology for wearable products meant for global release, but it doesn't have a launch date yet. Qualcomm said it has already started the new chip's development and will be sharing more information about its timeline at a later date. 

ERNIE, Baidu’s answer to ChatGPT, has “achieved a full upgrade,” company CEO Robin Li told the assembled crowd at the Baidu World 2023 showcase on Tuesday, “with drastically improved performance in understanding, generation, reasoning, and memory.”

During his keynote address, Li demonstrated improvements to those four core capabilities on-stage by having the AI create a multimodal car commercial in a few minutes based on a short text prompt , solve complex geometry problems and progressively iterate the plot for a short story on the spot. The fourth-gen generative AI system “is not inferior in any aspect to GPT-4,” he continued.

ERNIE 4.0 will offer an “improved” search experience resembling that of Google’s SGE, aggregating and summarizing information pulled from the wider web and distilled into a generated response.The system will be multimodal, providing answers as text, images or animated graphs through an “interactive chat interface for more complex searches, enabling users to iteratively refine their queries until reaching the optimal answer, all in one search interface,” per the company’s press. What’s more, the AI will be able to recommend “highly customized” content streams based on previous interactions with the user.

Similar to ChatGPT Enterprise, ERNIE’s new Generative Business Intelligence will offer a more finely-tuned and secure model trained on each client’s individual data silo. ERNIE 4.0 will also be capable of, “conducting academic research, summarizing key information, creating documents, and generating slideshow presentations” as well as enable users to search and retrieve files using text and voice prompts.

Baidu is following the example set by the rest of the industry and has announced plans to put its generative AI in every app and service it can manage. The company has already integrated some of the AI’s functions into Baidu Maps, including navigation, ride hailing and hotel bookings. It is also offering “ow-threshold access and productivity tools” to help individuals and enterprises develop API plugins for the Baidu Qianfan Foundation Model Platform.

Baidu had already been developing its ERNIE large language model for a number of years prior to the debut of ChatGPT in 2022, though its knowledge-base is focused primarily on the Chinese market. Baidu released ERNIE Bot, it’s answer to ChatGPT, this March with some 550 billion facts packed into its knowledge graph, though it wasn’t until this August that it rolled out to the general public.

Baidu’s partner startups also showed off new product series that will integrate the AI’s functionality during the event, including a domestic robot, an All-in-One learning machine and a smart home speaker.

Microsoft’s new Copilot AI has wormed its way into nearly every aspect of Windows 11. However, there’s a bit of a learning curve, but don’t worry. We’ve got you covered. We've put together a primer on the company's new AI assistant, along with step-by-step instructions on how to both enable and disable it on your Windows computer.

What does Microsoft Copilot do?

Microsoft’s Copilot is a suite of AI tools that work together to create a digital personal assistant of sorts. Just like other modern AI assistants, the tech is based on generative artificial intelligence and large language models (LLM.)

You can use Copilot to do a whole bunch of things to increase productivity or just have fun. Use the service to summarize a web page or essay, write an email, quickly change operating system settings, generate custom images based on text, transcribe audio or video, generate a screenshot and even connect to an external device via Bluetooth. It also does the sorts of things other AI chatbots do, like creating lists of recipes, writing code or planning itineraries for trips. Think of it as a more robust version of the pre-existing Bing AI chatbot.

How to enable Microsoft Copilot

Update your computer to the latest version of Windows 11

First of all, you need the latest Windows 11 update, so go ahead and download that first. 

1. Head to Settings and look for the Windows Update option. 

2. Follow the prompts and reset your computer if required. 

You’re now ready to experience everything Copilot has to offer. If Microsoft just dropped an update, you may have to wait a bit before it reaches your region. Click the tab to automatically install the latest update when available.

Once your computer is updated, click the Copilot button

As for enabling the feature, click the Copilot button on the taskbar or press Win + C on the keyboard. That’s all there is to it.

How to disable Microsoft Copilot

Engadget/Terrence O'Brien

Microsoft Copilot isn’t an always-on feature. Once it shows up in the taskbar, it only works when you ask it something. However, if you want to disable or delete the feature entirely, you have a couple of options.

The easiest method is to remove it from the taskbar. Out of sight, out of mind, right? Open up Settings and click on Personalization. Next, tap the Taskbar page to the right side. Look for Taskbar Items and then click on the Copilot toggle switch to remove it from the line-up. This ensures you won’t ever accidentally turn it on via the Taskbar, but you can still call up the AI by typing Win + C.

If you want to delete the toolset entirely, the process is a bit more involved. Start by opening a PowerShell window. Search for Windows PowerShell, right-click on the results and select the option to run as an administrator. Next, click yes on the UAC prompt. This opens up a command prompt.

Paste the following into the window: reg add HKCU\Software\Policies\Microsoft\Windows\WindowsCopilot /v "TurnOffWindowsCopilot" /t REG_DWORD /f /d 1

That should do it. Every trace of Copilot will disappear from your system.

What are the limitations of Copilot?

This is new technology, so the limitations are extensive. Like all modern LLMs, Microsoft’s Copilot can and will make up stuff out of thin air every once in a while, a phenomenon known as hallucination. It also doesn’t retain information from conversation to conversation, likely for security reasons. This means it restarts the conversation from a blank slate every time you close a window and open another one. It won’t remember anything about you, your preferences or even your favorite order from the coffee shop down the street. Finally, it doesn’t integrate with too many third-party sources of data, beyond the web, so you won’t be able to incorporate personal fitness data and the like.

What's the difference between Github Copilot and Microsoft Copilot?

There is a primary difference between the two platforms, despite the similar names. Github Copilot is all about helping craft and edit code for developing software applications. Microsoft Copilot can whip up some rudimentary code but it’s far from a speciality. If your primary use case scenario for an AI assistant is code, go with Github. If you only dabble in coding, or have no interest at all, go with Microsoft.

Snapchat has rolled out two new features, including the ability to embed content from the platform into a website. Users can now embed Lenses, Spotlight videos and public stories or profiles through their computer browser by clicking the embed button under share options. This will automatically copy the code — just as competitors like Instagram and TikTok have long allowed users to do. 

Following years of trying to broaden from just a platform to send pictures back and forth with friends, the option to embed is a logical next step from Snapchat. It builds on other features like articles and discovering local places of interest and, in 2022, Snapchat for Web. 

Along with embeds, Snapchat has also launched an OpenAI-powered feature that lets users extend their snaps to include more of their possible surroundings. The tool is reminiscent of Photoshop's Content-Aware Fill but, in this case, estimates what the entire border area looks like versus one targeted piece. Engadget has confirmed this feature is available for Snapchat+ subscribers. 

The company has regularly been using AI tools as perks for its now five million-plus Snapchat+ subscribers. The company's AI-powered Dreams feature — which lets users generate eight packs of "fantastical" images — is limited to one time only for regular users or one set per month for Snapchat+ subscribers. Anyone can buy extra packs for $0.99 each.

Snapchat was quick to hop on the AI boom, rolling out a chatbot called My AI using "OpenAI's GPT technology that the authors have customized" back in February. Initially also available solely to Snapchat+ subscribers, My AI expanded to all global users two months later with everything from restaurant recommendations to photo responses (as has been the case for AI bots in 2023, not always appropriately). 

It used to be that if you dreamed of reaching 6GHz speeds with your hot rod desktop CPU, you'd have to try your luck with overclocking and all of the potential instability and cooling demands that required. Earlier this year, Intel released the Core i9-13900KS, which hit 6GHz right out of the box. Now, the company is doing it again with its fastest 14th-gen desktop CPU, the i9-14900K. That frequency is just a short-lived "Thermal Velocity Boost" speed, which isn't sustained for very long, but it's still something Intel can lord over AMD.

These 14th-gen chips, to be clear, are different from Intel's recently announced 14th-gen Core Ultra processors for notebooks. Understandably, Intel is focusing on efficiency for its mobile lineup, whereas its desktop chips are all about raw power (and 6GHz bragging rights). You can look at the 14th-gen desktop hardware as a last gasp for Intel's existing architecture, where the company doesn't mind pushing power demands to out-bench AMD. (The i9-14900K consumes as much as 253 watts, just like its predecessor.)

Intel's highest-end 14th-gen chip may get most of the attention, but discerning gamers may be more interested in the i7-14700K, which now features 20 cores (8 performance and 12 efficiency) and reaches up to 5.6GHz "Turbo Boost Max 3.0" speeds. Perhaps most importantly, its $409 price tag is well below the 14900K's $589.

When it comes to benchmarks, Intel claims the 14900K is up to 23 percent faster than AMD's Ryzen 9 7950X3D while playing Starfield in 1080p. The company also says that chip is up to 54 percent faster than the same AMD hardware while multi-tasking between After Effects and Premiere Pro. (That comparison may be a bit unfair, since Intel tested an Auto Reframe task in Premiere Pro that works together with its UHD graphics, something that AMD's graphics don't help with.)

Intel

Perhaps more useful than raw benchmark comparisons, Intel also says the i9-14900K was able to reach over 100fps in Total War: Warhammer III while playing, streaming and recording in 1080p with ultra graphics settings. That game is also optimized specifically for its 14th-gen hardware, so you can't expect the same results with every title.

If you're still eager to overclock, Intel is also making that easier with its new XTU AI Assist feature, which will only be available on the i9-4900K. In a demo for media, an Intel representative showed off how the XTU app can quickly determine the ideal processor core voltages, motherboard power settings and "other tuning knobs" to determine safe performance speeds. They noted that Intel had trained its AI overclocking model on hundreds of CPUs, as well as a variety of motherboard and cooler options (including a bit of liquid cooling). Once the AI tuning process is complete, you can roll with its suggested settings or use them as a baseline for further overclocking.

One feature you won't see in these new 14th-gen chips? An NPU (neural processing unit) for AI acceleration. Oddly enough, the company's upcoming Core Ultra mobile chips will feature an NPU, which enables things like Windows Studios Effects for AI-powered background blurs in video chats. It makes sense for Intel's first NPU to appear in its new architecture, whereas the 14th-gen desktop chips are beefed-up versions of last year's hardware. Still, it's a bit odd for its priciest desktop hardware to miss out on something mere laptops will see next year.

Intel's 14th-gen desktop chips will be available from retailers and system manufacturers on October 17th.

If Apple does release a cheaper Vision Pro headset, it’s likely to be a stripped-down version with fewer external cameras, no EyeSight feature and an iPhone chip in place of the current Mac chip. Bloomberg’s Mark Gurman reported in his Power On newsletter that the company is looking at these options as ways to bring the cost down to between $1,500 and $2,500 for a lower-end model. Apple Vision Pro, which is set to begin shipping in early 2024, currently starts at an eye-watering $3,500.

In an earlier report, Gurman said the cheaper Vision Pro would also use lower resolution displays inside the headset. The big indication now is that it may drop the external display that enables its signature EyeSight feature, according to Gurman. EyeSight projects the wearer’s eyes onto an outward-facing OLED display so anyone who approaches can see where their attention is focused. It was one of major selling points in Apple's Vision Pro announcement, though it isn’t exactly mission critical.

According to Gurman, a follow-up to the premium Vision Pro is in the works already, too. That will have EyeSight, and potentially a more lightweight design that opts for built-in prescription lenses for those who need it rather than the Zeiss inserts it’s working with now.

We’re not quite done with the busy fall season of tech events. It’s Baidu’s turn to reveal what it has been cooking up. The company will host its Baidu World showcase on October 16th at 10PM ET.

After several years of holding the event entirely virtually, Baidu is returning to an in-person format with "real-world demos and experiences" for the first time since 2019. Perhaps unsurprisingly, given the direction almost every major tech company has gone in recent times, there will be a significant focus on artificial intelligence.

Baidu says it will discuss “a range of AI-native applications and the latest advances in foundation models.” That suggests we’ll learn more about what’s next for ERNIE Bot, its large language model and generative AI chatbot. The company may also have more to share about the impact of artificial intelligence on its self-driving initiatives.

The two-hour event will open with a keynote address from Baidu co-founder and CEO Robin Li. You can watch a live stream of Baidu World 2023 below.

One of the biggest unanswered questions regarding Microsoft's $67.8 billion purchase of Activision Blizzard concerned the role of Bobby Kotick at the company. Now that the deal is finally complete and the publisher is officially a part of Microsoft, the future of the Activision Blizzard CEO is a little clearer.

In a memo to employees, Kotick wrote that he is "fully committed to helping with the transition." He'll remain at the helm "through the end of 2023" and he'll report to Microsoft Gaming CEO Phil Spencer. "We both look forward to working together on a smooth integration for our teams and players," Kotick wrote.

Kotick is widely expected to step down from the role he has held for over three decades as soon as January 1, though Activision Blizzard has declined to confirm an exact date for his departure. In any case, his contract runs until April.

Other changes are expected at the top of Activision Blizzard. In his own memo to the publisher's staff, Spencer wrote that "we will share more updates on our new organizational structure in the coming months."

Kotick shaped his company into a financially successful enterprise after leading a group of investors to buy Mediagenic for a bargain-basement price in 1991. He restructured the company and restored its previous name of Activision. In 2008, following a string of successful games, acquisitions and investments, Kotick engineered a merger with the games division of Vivendi, which owned Blizzard.

However, Kotick's stewardship of the company has long been a controversial one. As far back as 2010, Kotaku called him "the most hated man in video games," partly because of the perception that Kotick places a heavy focus on monetizing Activision Blizzard's games as much as possible. That's not to mention his reported treatment of employees at his various companies.

Over the last few years, observers have been paying more attention to Activision Blizzard's culture under Kotick. In 2021, the California Civil Rights Department, sued the company, alleging that it fostered a culture where discrimination and widespread harassment were rife. Later that year, a report by The Wall Street Journal indicated that Kotick had long been aware of sexual misconduct and assault allegations at Activision Blizzard, and that he neglected to share some of those (or details about settlements that were agreed with alleged victims) with the board.

The report led to many Activision Blizzard workers walking out and demanding Kotick's resignation. The following year, months after Microsoft made its blockbuster bid for Activision Blizzard, the publisher's shareholders voted to let Kotick keep his board seat.

In fact, the turmoil over Activision Blizzard's toxic workplace culture (and resulting pressure on the company that sent its stock tumbling) was what prompted Microsoft to buy the company. Kotick had claimed that Overwatch 2 and Diablo IV delays resulted in a fall of the stock price. But workers pushed back on those assertions.

Overwatch producer Tracy Kennedy claimed in early 2022 that Kotick pushed "random projects" onto the development team. Kennedy said the team worked overtime only to see those projects canceled and that "entire teams are turning over" and blaming Kotick.

It's not only at Activision where Kotick has spurred controversy. In 2007, the flight attendant of a private jet he was co-owner of sued him. The attendant alleged that, after she informed the plane's other owner that the pilot had sexually harassed her, Kotick fired her. The case resulted in a $200,000 settlement for the flight attendant.

Kotick has long been one of the highest-paid CEOs in North America and he's not exactly going to be retiring with only a nice watch to show for his time at Activision Blizzard. A report last year suggested that he stood to make $375.3 million from the sale to Microsoft. He's also expected to receive a golden parachute payment of $14.6 million when he departs.

While most Apple chatter tends to concern the iPhone, iPad, Apple Watch and Macbook lines, the company is hardly neglecting those who prefer a desktop Mac. The Mac mini is a capable piece of kit, especially now that the company is shoving its Apple Silicon chips into them. Those interested in picking up a Mac mini may be interested to learn that an M2-powered model is currently on sale at B&H. It has dropped by $100 to $499 to match a solid deal that popped up in August.

This model includes an M2 chipset with an eight-core CPU, 10-core GPU and 16-core neural engine, along with 8GB of unified RAM and 256GB of SSD storage. It has a pair of Thunderbolt 4/USB4 ports, two USB-A slots and one each for Ethernet and HDMI. The M2 Mac mini supports Wi-Fi 6E and Bluetooth 5.3. It has a 3.5mm headphone jack too.

We gave the M2 Mac mini a score of 86 in our review back in January. We appreciated the machine's performance and variety of ports, as well as its quiet operation. We liked the design too. While it largely looks similar to earlier models, this Mac mini has a slightly elevated base to improve airflow.

We didn't have any major reservations with this version specifically, though upgrades are expensive and the more advanced M2 Pro model is too pricey. Our main quibble was not having any front-facing ports. Ultimately, there's not a ton to dislike about the M2 Mac mini.

Follow @EngadgetDeals on Twitter and subscribe to the Engadget Deals newsletter for the latest tech deals and buying advice.