Recently, AMD reached out to Neowin around that time frame and offered us the opportunity to interview Jason Banta, who is the Corporate Vice President and General Manager of Client OEM at the company. We figured that if anyone would be able to get into those nitty-gritty details about how AMD's product stack stacks up alongside the changing landscape of Windows, it would probably be him.

So we accepted the opportunity and in this write-up we share some of the interesting things Banta revealed to us, including AMD's complete hardware-readiness with the next generation of Windows, be it Windows 12 (?), or something else.

Q) With Windows 10 support ending this month (mid-October), how does AMD view the scale and urgency of this refresh cycle compared to past OS transitions? Estimates suggest hundreds of millions of devices are still on Windows 10. What are the some of the top challenges and opportunities this presents for AMD and its partners?

Jason Banta:

The end of Windows 10 support is one of the biggest refresh moments the PC market has seen in years.

As we approach the deadline, interest in upgrading is clearly building. Consumers and businesses are reassessing what a modern PC can do, which creates real room for innovation across the ecosystem. Our focus with Microsoft and OEM partners is to make the transition feel like a step forward change, not a sideways move.

For anyone coming from an older system, the difference is immediate. Moving to Windows 11 on a Ryzen platform brings faster performance, longer battery life, and advanced on-device AI features that were not available before. The combination of Windows 11 and Ryzen AI makes this a smart time to modernize and to choose a system that will carry you forward.

Q) Microsoft is promoting Copilot+ PCs as the natural upgrade path. From AMD’s perspective, what makes a Copilot+ PC fundamentally different from a traditional Windows 11 PC? How critical is the NPU (Neural Processing Unit) in a Copilot+ PC in terms of delivering the AI experiences Microsoft is highlighting, such as Recall, Copilot Vision, among others?

Jason Banta:

Copilot+ PCs are designed for on-device AI. The key is the Neural Processing Unit which runs AI workloads locally and helps power features like “Click-to-Do" and “Recall” respond quickly, protect data on the device, and use power efficiently.

AMD brought the first NPU to x86 PCs in 2023, and with the Ryzen AI 300 Series we deliver up to 50 TOPS of NPU performance. AI experiences also benefit from the CPU and RDNA graphics working alongside the NPU, which gives developers headroom to keep adding capabilities that run locally.

Traditional Windows 11 PCs remain excellent, but Copilot+ systems open the door to experiences that learn and improve over time. As more apps tap into local acceleration, users will feel a meaningful difference in responsiveness and personalization.

Q) Can you explain how AMD’s hardware roadmap (Ryzen AI, etc.) is optimized for these AI workloads compared to previous generations? How do AMD-powered Copilot+ PCs compare with Intel’s offerings in terms of AI performance, efficiency, and cost? Where do you expect AMD to have the clearest advantage over its rival?

Jason Banta:

Ryzen AI was built specifically for AI on the PC. Since introducing the first x86 NPU in 2023, we have advanced the architecture each generation to improve efficiency and coordination across CPU, GPU, and NPU. Our latest Ryzen AI MAX processors combine up to 16 Zen 5 CPU cores, an XDNA 2 NPU delivering up to 50 TOPS on x86, and up to 40 RDNA compute units for strong graphics performance.

This balanced design lets the system route work to the right engine, which helps battery life and overall responsiveness in Windows 11 Copilot+ workloads.

Our advantage is breadth and scalability. We offer AI-capable PCs across consumer, commercial, and creator segments, so users can choose the right performance and budget while OEMs can build consistent designs across product lines. That portfolio depth, paired with our x86 NPU performance and RDNA graphics, positions AMD well as AI PCs become the standard.

Q) How do you see the refresh cycle playing out differently in enterprise vs. consumer markets? Many enterprises may be cautious about AI adoption. How is AMD working with Microsoft to reassure businesses that Copilot+ PCs are not just “nice-to-have” but essential for productivity and security?

Similarly, for consumers, what are the most compelling reasons to upgrade from a Windows 10 PC to a Copilot+ PC? Also, how is AMD addressing concerns about price accessibility; will Copilot+ PCs be available across entry-level, mid-range and the more premium segment?

Jason Banta:

In enterprise, the conversation begins with security, compliance, and manageability. Windows 11 provides a stronger baseline with TPM 2.0 and Secure Boot as standard. AMD PRO technologies, available on Ryzen PRO and Ryzen AI PRO commercial SKUs, extend that with hardware-based protection, reliability features, and remote manageability that integrate with tools such as Microsoft Intune and Windows Autopilot.

Many organizations also view this refresh as a productivity upgrade. In addition to security advancements, AMD’s Zen 5 cores enable enterprise workers to get things done quickly and efficiently. On-device AI features like Copilot can streamline routine work, improve decision-making, and make using the PC easier.

For consumers, the benefits are straightforward: faster performance, longer battery life, stronger security, and new Copilot+ creative tools like Image Creator. We are working closely with OEM partners to ensure these advantages are widely accessible by offering Ryzen AI systems at multiple price points, from mainstream notebooks to flagship Ryzen AI MAX designs.

Q) Looking beyond Windows 11 24H2/25H2, how does AMD see the evolution of the AI PC category? How is AMD preparing for the possibility of the next gen Windows (12) and ensuring that today’s Copilot+ PCs can remain future-proof?

Jason Banta:

Everything we design at AMD is intended for longevity. Our high-performance cores, GPUs, and NPUs ensure your PC can handle the most complex applications. The socket architecture we use for desktop is built with upgradeability in mind. And our software solutions evolve with changes in the PC ecosystem.

AI PCs are now the foundation for the next era of personal computing.

We are working with Microsoft and the ecosystem on standards such as Windows ML, ONNX, and Model Context Protocol so AMD hardware, drivers, and software are ready for what comes next. Our goal is to ensure a Ryzen AI system you buy today is not only compatible with future Windows capabilities but optimized for them.

Overall, it looks like AMD is pretty upbeat about the opportunity it has on its hands, though it also wants to ensure compatibility. While Windows 10's end of support has caused a major disruption as many PCs are probably no longer officially able to run Windows 11 due to ineligibility, AMD seems to hint that the same would not be the case when Windows (12?) rolls out, at least in the case of the capable Copilot+ PCs.

We would sincerely like to thank AMD and Jason Banta for taking the time to consider and answer our questions.

Source

Hope you enjoyed this news post. Feedback welcome.

Posted Sunday 9 November 2025 at 5:48 pm AEST (my time).

News posts... 2023: 5,800+ | 2024: 5,700+ | 2025 (till end of October): 5,009

RIP Matrix

This kind of superintelligence, according to Suleyman, won’t be “an unbounded and unlimited entity with high degrees of autonomy” and will instead be “carefully calibrated, contextualized, within limits.” Suleyman joined Microsoft last year as the CEO of Microsoft AI, which has only recently launched its first in-house models for text, voice, and image generation.

Though Suleyman’s blog post says Microsoft AI will “reject narratives about a race to AGI,” the competition between Microsoft and OpenAI is about to get much more intense. Under a new deal with OpenAI, Microsoft can now “independently pursue AGI alone or in partnership with third parties.” And, as pointed out by my colleague Hayden Field, “Microsoft is perfectly within its legal rights to use OpenAI’s IP to develop its own AGI and attempt to win the race.”

But Suleyman has a vision for “humanist” superintelligence with three main applications, which include serving as an AI companion that will help people “learn, act, be productive, and feel supported,” offering assistance in the healthcare industry, and creating “new scientific breakthroughs” in clean energy.

“At Microsoft AI, we believe humans matter more than AI,” Suleyman writes. “Humanist superintelligence keeps us humans at the centre of the picture. It’s AI that’s on humanity’s team, a subordinate, controllable AI, one that won’t, that can’t open a Pandora’s Box.”

Source

The long-awaited addition of Housing in World of Warcraft is going to cost you. On top of your monthly subscription and the cost of the Midnight expansion itself, some Housing items will require a new in-game currency called “Hearthsteel” that players will have to pay for with real money. World of Warcraft already has hundreds of currencies, and at one point, the game’s main currency, gold, was worth more than the Venezuelan bolivar.

Blizzard said in its announcement blog post for Hearthsteel that it will only be for Housing, not the Trading Post, mounts, or other in-game items. Blizzard also clarified that not all of the Housing items will require Hearthsteel:

“The vast majority of Housing items are (and will continue to be) earnable in-game. Only a small fraction is available for purchase from the shop. Comparing this number against collectibles such as mounts is a good metric here, with more than a thousand mounts earnable in-game and only a few dozen from the shop.”

Source

Hope you enjoyed this news post. Feedback welcome.

Posted Saturday 8 November 2025 at 1:40 pm AEST (my time).

News posts... 2023: 5,800+ | 2024: 5,700+ | 2025 (till end of October): 5,009

RIP Matrix

Seven families across the US and Canada sued OpenAI on Thursday. Four claim the chatbot encouraged their loved ones to take their own lives; the remaining three say talking to ChatGPT led to mental health breakdowns.

The cases were all filed in California courts on the same day by the Tech Justice Law Project and Social Media Victims Law Center, according to The New York Times, in order to highlight the depth and breadth of the chatbot's alleged impact.

Amaurie Lacey, a 17-year-old from Georgia, discussed suicide with the AI for a month before taking his life in August. Joshua Enneking, 26, of Florida asked the bot if it would "report his suicide plan to the police."

Zane Shamblin, a 23-year-old from Texas, died by suicide after ChatGPT encouraged him to do so, his family says. Joe Ceccanti, 48, of Oregon became convinced the AI was sentient after using it without issue for years. He took his life in August after a psychotic break in June.

"This is an incredibly heartbreaking situation, and we're reviewing the filings to understand the details," OpenAI tells us in a statement. "We train ChatGPT to recognize and respond to signs of mental or emotional distress, de-escalate conversations, and guide people toward real-world support. We continue to strengthen ChatGPT’s responses in sensitive moments, working closely with mental health clinicians."

The three people who say ChatGPT spurred psychotic breakdowns include Hannah Madden (32, North Carolina), Jacob Irwin (30, Wisconsin), and Allan Brooks (48, Ontario, Canada).

Brooks says ChatGPT convinced him he had invented a mathematical formula that could break the internet and "power fantastical delusions," as the Times puts it. 

James, another man not included in the lawsuits, came across Brooks' story online and said it made him realize he was falling victim to the same AI-induced delusion. James is now seeking therapy and is in regular contact with Brooks, who is co-leading a support group called The Human Line Project for people going through AI-related mental health episodes, CNN reports.

All of the alleged victims in the lawsuits were using OpenAI's GPT-4o model. The company replaced it with a new flagship model, GPT-5, in August. After backlash from users who say they had a strong emotional attachment with GPT-4o, the company reintroduced it as a paid option.

OpenAI CEO Sam Altman has admitted multiple times that his product can be dangerous for people with poor mental health, and sycophantic to the point of encouraging delusions. After parents sued the company in August following their teen son's suicide, the company said its safety guardrails broke down over the months the boy was speaking to ChatGPT.

"If a user is in a mentally fragile state and prone to delusion, we do not want the AI to reinforce that," Altman wrote on X in August. "Most users can keep a clear line between reality and fiction or role-play, but a small percentage cannot." About a million users out of 800 million talk to ChatGPT about suicide each week, OpenAI says.

Altman added "I can imagine a future where a lot of people really trust ChatGPT’s advice for their most important decisions. Although that could be great, it makes me uneasy."

OpenAI says it's working on improving the chatbot's response during "sensitive conversations." It also introduced parental controls for teen users, and is working on a way to automatically identify teens to ensure none slip through the cracks.

Character.AI also faces a lawsuit from parents who say their son took his life after encouragement by an AI on the site. This month, it will ban teens from having unlimited chats on its platforms, and is working on a different way for them to engage with it.

Disclosure: Ziff Davis, PCMag's parent company, filed a lawsuit against OpenAI in April 2025, alleging it infringed Ziff Davis copyrights in training and operating its AI systems.

Source

Reading a person’s mind using a recording of their brain activity sounds futuristic, but it’s now one step closer to reality. A new technique called ‘mind captioning’ generates descriptive sentences of what a person is seeing or picturing in their mind using a read-out of their brain activity, with impressive accuracy.

The technique, described in a paper published today in Science Advances1, also offers clues for how the brain represents the world before thoughts are put into words. And it might be able to help people with language difficulties, such as those caused by strokes, to better communicate.

The model predicts what a person is looking at “with a lot of detail”, says Alex Huth, a computational neuroscientist at the University of California, Berkeley. “This is hard to do. It’s surprising you can get that much detail.”

Scan and predict

Researchers have been able to accurately predict what a person is seeing or hearing using their brain activity for more than a decade. But decoding the brain's interpretation of complex content, such as short videos or abstract shapes, has proved to be more difficult.

Previous attempts have identified only key words that describe what a person saw rather than the complete context, which might include the subject of a video and actions that occur in it, says Tomoyasu Horikawa, a computational neuroscientist at NTT Communication Science Laboratories in Kanagawa, Japan. Other attempts have used artificial intelligence (AI) models that can create sentence structure themselves, making it difficult to know whether the description was actually represented in the brain, he adds.

Horikawa’s method first used a deep-language AI model to analyse the text captions of more than 2,000 videos, turning each one into a unique numerical ‘meaning signature’. A separate AI tool was then trained on six participants’ brain scans and learnt to find the brain-activity patterns that matched each meaning signature while the participants watched the videos.

Once trained, this brain decoder could read a new brain scan from a person watching a video and predict the meaning signature. Then, a different AI text generator would search for a sentence that comes closest to the meaning signature decoded from the individual’s brain.

For example, a participant watched a short video of a person jumping from the top of a waterfall. Using their brain activity, the AI model guessed strings of words, starting with “spring flow”, progressing to “above rapid falling water fall” on the tenth guess and arriving at “a person jumps over a deep water fall on a mountain ridge” on the 100th guess.

The researchers also asked participants to recall video clips that they had seen. The AI models successfully generated descriptions of these recollections, demonstrating that the brain seems to use a similar representation for both viewing and remembering.

Reading the future

This technique, which uses non-invasive functional magnetic resonance imaging, could help to improve the process by which implanted brain–computer interfaces might translate people’s non-verbal mental representations directly into text. “If we can do that using these artificial systems, maybe we can help out these people with communication difficulties,” says Huth, who developed a similar model in 2023 with his colleagues that decodes language from non-invasive brain recording.

These findings raise concerns about mental privacy, Huth says, as researchers grow closer to revealing intimate thoughts, emotions and health conditions that could, in theory, be used for surveillance, manipulation or to discriminate against people.

Neither Huth’s model nor Horikawa’s crosses a line, they both say, because these techniques require participants’ consent and the models cannot discern private thoughts. "Nobody has shown you can do that, yet,” says Huth. 

Source

Gamers Nexus (2.52M subscribers)

November 6, 2025

Video length: 24m 13s

In this hardware news recap, we talk about GameTop, the ByteDance potential competitor to Steam, 5.25" drives still living and with new case options to go with them, AMD's rumored 7500X3D, Intel's financial situation improving, and more. We also cover the ongoing corporate obsession with the best two letters in the alphabet: "AI."

00:00 - Recap for the Week 

01:46 - AMD Earnings 

03:27 - Steam Competitor in China: GameTop 

06:15 - 5.25 Optical Drives Live - Sort Of 

08:01 - AMD Preps 7500X3D 

09:07 - Intel Layoffs Help Financials 

14:19 - Microsoft Fumbles Intro of Clippy's Bastard Son 

17:24 - Meta Lays Off 600 in AI Division 

19:10 - EA Announces AIAIAIAI O 

22:01 - Amiga 1200 Remake

Source

Hope you enjoyed this news post. Feedback welcome.

Posted Friday 7 November 2025 at 3:31 am AEST (my time).

News posts... 2023: 5,800+ | 2024: 5,700+ | 2025 (till end of October): 5,009

RIP Matrix

A new report has provided more information about Apple's rumored upcoming cheap MacBook, and it's not looking good for Microsoft and cheap Windows PCs. According to Bloomberg, Apple is spinning up production on a MacBook that's expected to cost "well under $1,000" that will ship in the first half of 2026 and feature Apple Silicon, a 13-inch display, and a premium aluminum chassis.

Murmurs about a cheap MacBook hit the internet earlier this summer, and at the time I called it the nightmare scenario for Microsoft and the mid-range Windows PC market. Up until now, Windows OEMs have gotten away with building uninspired, minimum effort PCs for the $500-700 price category, resulting in usually cheap feeling or poorly performing devices.

A new MacBook for anywhere near $700 is going to sell like hotcakes, potentially impacting the midrange Windows PC market considerably. It's fair to say that most people buying a PC in the $500-$700 price range aren't buying one because they love Windows, and any alternative that's good or better than what's on offer here is going to be preferred by many.

A cheap MacBook, running full macOS and complete with that premium Apple design is going to win out over any plastic or poor efficiency Windows laptop in the same price bracket. Bloomberg says the cheap MacBook will be powered by an A18 Pro SoC, which will give it performance similar to that of an M1 MacBook Air, but with the longest battery life ever achieved in a MacBook. That means it'll easily last over a day on a single charge.

It's also expected to feature a display slightly smaller than the one found in the MacBook Air. If I were a betting man, I'd wager it'll feature a 13-inch display, similar, if not identical, to the one found in the iPad Air. That's no slouch of a display, and while Apple is likely to remove the touchscreen, it'll still be good to look at, and that's all that matters to most people.

Simply put, there isn't a Windows laptop that you can buy new for $600-$700 that will match this MacBook. The closest competitor is Microsoft's Surface Laptop 13-inch, but even that is $899 with a significantly lower-resolution display. The only Windows laptops that even come close to competing are slightly older ones that are on sale, like the ASUS ZenBook A14, which has a 1080P OLED screen and all-day battery life for $699 (or less) on sale.

Ultimately though, given how dire Windows' reputation is these days, I don't think it matters if there are PCs that compete head to head with a cheap MacBook. A lot of people will simply choose a MacBook over a Windows PC simply because the MacBook doesn't run Windows, and more than ever people are scrambling to abandon Microsoft's enshittified OS.

AMD's third-quarter financial results for 2025 are in, and they're looking good for the NVIDIA and Intel competitor almost entirely across the board. In the age of astronomical AI-related spending, it's great to see Team Red making headway with its consumer graphics card and processor business, and not just AI chips.

Not only did AMD's revenue grow in the third quarter, but it broke a record at $9.2 billion with a gross margin of 52%. That revenue represents a 36% year-over-year increase, which AMD Executive Vice President Jean Hu says "generated record free cash flow."

With $1.3 billion in operating income and $2 billion of net revenue, diluted earnings per share work out to $0.75. Looking at the results with a non-GAAP eye — which can often give a better picture of core operations — AMD says its gross margin was 54% and diluted earnings per share were $1.20.

Before these earnings were made public, I probably would have placed my bet on AMD's specialized AI chips — known as Instinct — making the most money for the company. That is, however, not the case at all.

AMD says its data center segment's revenue was $4.3 billion, a 22% increase compared to the third quarter of 2024. The company says this growth was, unsurprisingly, driven by its specialized EPYC CPUs and Instinct MI350 GPUs.

It's the client and gaming segment that shows the most growth, and it's not close at all. Third-quarter revenue from this segment was $4 billion, which is an increase of 73% compared to the previous year.

Client revenue alone broke an AMD record at $2.8 billion, an increase of 46% year-over-year. AMD says these record sales were driven by its consumer Ryzen CPUs and "a richer product mix."

Considering the almighty AMD Ryzen 7 9800X3D — widely considered the best CPU for PC gaming and a nice upgrade over its predecessor, the 7800X3D — and its other new X3D siblings launched at the end of 2024, it's pretty clear that the demand for AMD's coveted CPUs is only climbing.

Ryzen is undoubtedly a hot buy, but it was AMD's Radeon gaming GPUs that grew the most in the third quarter. Gaming revenue hit $1.3 billion in the third quarter, which was a 181% year-over-year increase.

AMD has positioned itself on the launch pad ... will it take off?

Microsoft has let go more than 15,000 workers since May, and Amazon announced last week it was cutting 14,000 roles. Artificial intelligence has taken some of the blame for the layoffs, but not in an obvious way.

The companies aren't in financial distress—far from it. Rather, there's a desire now among the tech giants to trim their operations, to be lean and flexible enough to keep up with AI development. For the first time in tech's recent history, allocating capital toward employees isn't the top priority.

Since Nov. 30, 2022, when AI giant OpenAI launched ChatGPT, tech companies have poured hundreds of billions of dollars into computer chips, data centers and new talent. And that spending isn't slowing down, even after record investments in 2025.

While some startups are eschewing junior coders in favor of AI models their leaders can control, AI cannot yet replace most employees. Rather, companies like Microsoft and Amazon say they want flatter leadership structures and less bureaucracy.

As Amazon CEO Andy Jassy said, the layoffs aren't financially driven nor are they caused by artificial intelligence, at least not directly. Instead, tech companies are radically rethinking how they're structured, and which jobs they need to fill.

The opportunity to bring AI into both work and home is the promised land that almost every tech company hopes to reach. The promise of AI is simple: It will streamline tedious tasks that eat up hours, leaving more time to stretch creative muscles and enjoy life.

At work, that might mean a team developing a new product. At home, it's re-creating a pirate ship from a jumble of Legos.

AI usage

When Maggie Stringfellow uses artificial intelligence at work, she's usually analyzing reams of data that would have been almost impossible to sort without the technology.

When Stringfellow, a vice president of product management at Seattle cybersecurity giant F5, uses AI at home, she's playing with Legos.

It'd be more fair to say her children are playing with the toy bricks, as Stringfellow says she's able to take a photo of the mess of Legos her kids have and an AI model will spit out a design for something new.

Stringfellow never had much anxiety around AI.

"I might be a bit on the nerdier side, so I always thought this was going to be a really exciting tool," she said. "I would say it took awhile for the business tools to evolve and really become useful, but they certainly have."

Stringfellow is perhaps more optimistic than some tech workers, but her use of AI isn't uncommon among those at startups, midsize firms like F5 or tech giants. There's uncertainty about the future—will AI eventually replace jobs en masse? But for the time being, having AI handle some of your work is handy.

In her personal life, Stringfellow can use any AI product she wants. After finishing a young adult novel, she'll go back and forth with a few different chatbots to gather new recommendations. She's also using generative AI, which can produce sentences and images from a user's prompt, to design marble runs and Lego sets with her kids.

Microsoft, which sells AI services called Copilots attached to its existing offerings, said the technology is saving some of their teams days of work. Marketing teams use Copilot to cut down the time to create a campaign from 12 weeks to three weeks, according to Microsoft. Another highly technical team had to migrate apps from one platform to another; with Copilot it took them about half a day compared with four days.

But there's some skepticism. One Microsoft software engineer said he uses Copilot more than he thought he would. On a given day, he'll use it for an hour by assigning it relatively easy tasks. He ends up having to polish the results.

"Sometimes it goes well, sometimes not," he said, speaking on condition of anonymity to protect his job.

A data scientist at Microsoft, also speaking anonymously to protect her job, said she'll use Copilot from time to time, but the tools available right now aren't much help to her job. Even if there's pressure to use it.

A fever pitch

In Washington state, diffusion of AI is at a trickle in some counties while others embrace the technology. More than 30% of the working-age population uses AI in King County, according to data collected by Microsoft. Snohomish County leads the state with more than 35% using the technology.

That number plummets in rural areas like Ferry County, where less than 3% of the population uses AI. Seven counties in Eastern Washington have usage rates below 10%.

Still, AI may become unavoidable as companies increasingly include it in widely used products, like Google's search function.

In the future, we're going to be using AI and not even know it," Stringfellow said. "It's going to be embedded into all of our tools and all of the ways we work."

The hype of the technology is feverish.

Amazon, the longtime market leader in cloud computing, was on its heels this summer during an earnings cycle in which Microsoft's cloud division results thrilled Wall Street and Amazon's underwhelmed.

Amazon's share of the cloud market is starting from a larger position, so it's difficult to show quick growth, as CEO Andy Jassy pointed out during an earnings call. But there was a perception this summer that Microsoft was making gains in AI while Amazon fell behind.

Wall Street's sentiments almost flipped last week during the most recent earnings cycle. Microsoft's announcement Wednesday that it was spending more than expected on AI infrastructure worried investors.

Microsoft had been expected to slow on spending but ended up reporting a record $34.9 billion in capital expenditures for the quarter, building infrastructure to support a technology that has yet to show significant return on investment.

A day later, Amazon's cloud division sales outperformed expectations and the company's share price surged.

Microsoft's close relationship with OpenAI, which set off the AI race in 2022 when it launched ChatGPT, has given the tech giant a head start in introducing AI-powered tools to its customers.

"It seems to reflect more on where the companies themselves are in this race," said Jean Atelsek, a senior research analyst for S&P Global Market Intelligence. "Amazon Web Services is at the beginning while Microsoft is mature."

Bureaucracy

ChatGPT's launch date, Nov. 30, 2022, now serves as a dividing line. Is your startup a pre-GPT or post-GPT company? Are you selling standard software or developing AI tools?

Kirby Winfield, founder of AI-focused and Seattle-based venture capital firm Ascend, said startups founded after ChatGPT launched follow a different hiring pattern than tech companies had before.

Startups founded after ChatGPT hire less than those founded before, Winfield said. But he's also seeing revenue for those startups increase faster than before.

Winfield said his firm funded a company that started with six engineers before paring down three. It was faster to have senior people use AI products to code than hire junior employees.

"They just don't need as many people to write the code," he said.

He said the startups founded before ChatGPT have what he calls outdated technical and hiring models.

On the technical side, these startups are bolting AI onto their products and rushing to get customer-facing AI tools out. On the people side, they've filled roles that may fall out of favor in an AI age.

"It's very hard to unpack and turn around," Winfield said.

Big Tech is a collection of companies structured like those pre-GPT startups, but on a much grander scale. Most of the giants spent the 2010s hiring as many people as they could. They have established product lines with years of research and development poured into each. AI is disrupting much of that.

For Microsoft, one of the biggest shake-ups outside of layoffs was the company's decision in October to promote sales chief Judson Althoff to CEO of commercial business. The new role would put him in charge of a segment of Microsoft that represents 75% of the company's revenue.

The move isn't a succession plan for Microsoft CEO Satya Nadella, but rather a way to move Nadella back into "founder mode," according to Althoff. With the move, Nadella is spending more of his time with technical leadership during what Nadella terms a "generational platform shift."

Amazon's leadership ranks aren't shuffling as dramatically. The company is more focused on alleviating what it calls bloat.

Last year, Amazon created an email address where employees could flag bureaucracy, CEO Jassy told a Seattle crowd in September during a company conference.

"You can't wake up one day and decide you're going to move fast and have the organization move fast," Jassy said. "But if it's a priority to you and your leadership team to move quickly, you can make that happen."

Employees are encouraged to alert leadership of areas where things can move more quickly. Jassy said it's brought in about 1,500 emails, about 500 of which have been addressed.

Two days after Amazon's sweeping layoffs, Jassy again made his pitch for a more nimble company—or what he calls the world's largest startup.

When asked by an analyst about the layoffs, he said it wasn't a financial decision or even one driven by AI. Instead, it's about removing layers and putting everyone in the company into the hybrid player-coach role.

"I don't know if there's ever been a time in the history of Amazon, or maybe business in general, with the technology transformation happening right now, where it's important to be lean," Jassy said.

"It's important to be flat, and it's important to move fast. And that's what we're going to do.

Source

Artificial intelligence tools like ChatGPT struggle to distinguish belief from fact, a new study has revealed.

A team from Stanford University in the US found that all major AI chatbots failed to consistently identify when a belief is false, making them more likely to hallucinate or spread misinformation.

The findings have worrying implications for the use of large language models (LLMs) in areas where determining between true and false information is critical.

“As language models (LMs) increasingly infiltrate into high-stakes domains such as law, medicine, journalism and science, their ability to distinguish belief from knowledge, and fact from fiction, becomes imperative,” the researchers noted.

“Failure to make such distinctions can mislead diagnoses, distort judicial judgments and amplify misinformation.”

The researchers evaluated 24 LLMs – including Claude, ChatGPT, DeepSeek and Gemini – using 13,000 questions to analyse their ability of distinguishing between beliefs, knowledge and facts.

All models tested failed at recognising false beliefs and statements, demonstrating a fundamental limitation in being able to link knowledge to truth.

“These findings expose a structural weakness in language models: their difficulties in robustly distinguishing between subjective conviction and objective truth depending on how a given assertion is formulated,” said Pablo Haya Coll, a researcher at the Computer Linguistics Laboratory of the Autonomous University of Madrid, who was not involved in the study.

“Such a shortcoming has critical implications in areas where this distinction is essential, such as law, medicine, or journalism, where confusing belief with knowledge can lead to serious errors in judgement.”

One possible solution to this shortcoming, according to Dr Coll, could be to train the model to be more cautious in its responses. While this may reduce the likelihood of hallucinations, it may also affect their usefulness.

The Stanford researchers called for tech companies developing AI tools to “urgently” improve the models before deploying them in high-stakes domains.

The findings were detailed in a study, titled ‘Language models cannot reliably distinguish belief from knowledge and fact’, which was published in the scientific journal Nature Machine Intelligence on Monday.

Source

At launch, and for the past 10 years, Pillars of Eternity has been a real-time-with-pause experience, letting players fight large numbers of foes while planning things out using the combat pause mechanic. Soon, an almost fully featured turn-based combat mode will be available in beta form.

"This is a big undertaking. This has been a lot of work primarily from Nick Carver and Jesse Bureno. Special shout out to Kurt Decker, who's also been adding a lot of work of his own recently," said Obsidian Studio Design Director John Sawyer. "This turn-based mode is something that we wanted to do because turn-based mode was received so well in Deadfire. That being said, there were a number of very reasonable criticisms of how turn-based mode was implemented in Deadfire."

The new mode will be a toggle found in the isometric RPG's settings menu, inside the Difficulty section. The studio is planning to make this a button on the HUD in a later update as well, letting players switch between real-time with pause and turn-based combat systems in between encounters while playing.

Taking in the lessons learned from Pillars of Eternity II: Deadfire, Obsidian has made these changes to the original game's new turn-based combat system:

• Unbound Turns: All turns are now influenced by a character’s overall speed, improving the value of Real-Time with Pause attributes and mechanics.
• Smarter Free Actions: Actions like swapping weapons, drinking potions, or enabling modals are now free actions. Each free action is limited to one per turn per type for balance.
• Faster Pacing: Combat lethality has been significantly increased for both enemies and players to keep battles tense and impactful.
• Mode Flexibility: You can choose Turn-Based Mode alongside difficulty when starting a new game — or swap freely between Real-Time with Pause and Turn-Based Mode in the game options. We’re also investigating a direct HUD toggle for even more convenience.

Pillars of Eternity's new public beta featuring the turn-based mode will kick off on November 5, with Steam and Xbox on Windows players being invited to try it out.

Source

Hope you enjoyed this news post. Feedback welcome.

Posted Tuesday 4 November 2025 at 12:51 pm AEST (my time).

News posts... 2023: 5,800+ | 2024: 5,700+ | 2025 (till end of October): 5,009

RIP Matrix

The news sent shares of IREN (IREN.O) up more than 20% in premarket trading on Monday. AI-server maker Dell rose 5% as it would provide IREN with Nvidia's GB300 chips and other equipment Microsoft (MSFT.O) will use for about $5.8 billion.

The five-year deal shows the AI industry's growing hunger for computing power to run applications such as ChatGPT. It follows earnings from major tech companies last week that underscored capacity shortages were limiting their ability to fully benefit from the boom.

Partnering with IREN would allow Microsoft to expand computing capacity without building new data centers or securing additional power - two of the biggest hurdles slowing its ability to meet surging AI demand.

It will also sidestep heavy capital spending on chips that will lose value as newer, more powerful processors arrive.

IREN, which has a market value of $16.52 billion after a more than six-fold surge in its shares this year, has multiple data centers across North America with a total capacity of 2,910 megawatts. Its facilities are powered fully by renewable energy.

IREN said the Nvidia processors are scheduled for phased deployment through 2026 at its 750-megawatt Childress, Texas, campus, alongside new liquid-cooled data centers designed to deliver about 200 megawatts of critical IT capacity.

Cash from Microsoft's prepayment will help finance part of its $5.8 billion Dell deal, IREN said in a filing. Its Microsoft contract could be terminated if it fails to meet the delivery timelines.

Microsoft CFO Amy Hood said last week the company now expects its AI capacity crunch to stretch into at least mid-2026, after previously predicting they would improve later this year.

Source

Gamers Nexus (2.52M subscribers)

November 3, 2025

Video length: 20m 03s

AMD threatened to limit GPU driver support for products it is still ACTIVELY SELLING, then made errors in its driver notes, then fixed them, "clarified" its statement, then it clarified the clarifying statement, then it said you're all just confused. AMD continues to squander every opportunity it is handed, managing to screw-up its driver roll-out and communications in a span of just a few days. The backdrop to this is NVIDIA's partnership with Palantir, continual screwing of its gaming consumers, and recent investment in Intel, which has recently shown promise with Battlemage (Arc GPU) driver updates.

00:00 - Ya Done F'd Up Now, A-AMD 

03:24 - What Happened 

07:54 - Gaslighting 

11:44 - RDNA 2 is Still Sold 

14:26 - Intel Puts AMD to Shame

17:04 - Conclusion

Source

Hope you enjoyed this news post. Feedback welcome.

Posted Monday 3 November 2025 at 6:09 pm AEST (my time).

News posts... 2023: 5,800+ | 2024: 5,700+ | 2025 (till end of October): 5,009

RIP Matrix

Valve published the results of the October 2025 Steam Hardware & Software Survey, showing Windows 11's continuous growth. Right now, the operating system holds 63.57% of all Steam users. In one month, it managed to increase its share by 3.18 points. Windows 10, on the other hand, lost 3.94 points, and it currently holds 31.14% of all who participated in the survey (remember that participation is optional). Finally, even though Windows 7 support on Steam has long been discontinued, and users already have problems accessing certain features, 0.09% of Steam users reported still holding to what many still consider the best Windows version ever released.

Overall, Windows dominates Steam with its overwhelming 94.84% share (-0.75 points). Linux is second with 3.05%, and macOS is third with 2.11%. Arch Linux is the most popular distro on Steam (0.31%), while macOS 15.6.1 is tied with macOS 26, having 0.51% and 0.50%, respectively.

As for hardware preferences, the desktop Nvidia RTX 3060 is once again the most popular graphics card on Steam. It currently has a 4.3% market share (-0.32 points), while the laptop variant of the RTX 4060 is second with 4.1% (-0.33 points). The RTX 3050 is third with a 2.96% (+0.05 points) share. Other hardware parts still include the classic 16GB of RAM (41.49%), a 6-core processor (28.69%), 8 GB of video memory (33.46%), and a 1080p display.

You can find more information on the official Steam website.

Source

Hope you enjoyed this news post. Feedback welcome.

Posted Monday 3 November 2025 at 4:03 am AEST (my time).

News posts... 2023: 5,800+ | 2024: 5,700+ | 2025 (till end of October): 5,009

RIP Matrix

Quick links:

1. Windows 10 and 11
2. Windows Insider Program
3. Updates are available
4. Reviews are in
5. Gaming news

Windows 11 and Windows 10

Here, we talk about everything happening around Microsoft's latest operating system in the Stable channel and preview builds: new features, removed features, controversies, bugs, interesting findings, and more. And, of course, you may find a word or two about older versions.

This week, Microsoft released the latest non-security updates for Windows 11 (not for Windows 10 anymore, as it is now unsupported). The biggest and most interesting changes are in KB5067036 for Windows 11 versions 24H2 and 25H2. There, you will find the new Start menu, redesigned battery indicators, more AI-powered features, etc. Windows 11 23H2 received KB5067112.

The latest non-security updates for Windows 11 moved certain parts of the legacy Control Panel to the Settings app, but there are plenty of important applets still waiting for their updates. Here are five parts of the old Control Panel that Microsoft should move to the Settings app as soon as possible.

Over the last four years, Windows 11 has received plenty of improvements and new features, but its dark mode remains critically undercooked. Fortunately, with the latest builds and PowerToys updates, Microsoft is finally taking dark mode more seriously, as more classic UI elements now support proper dark mode.

In addition to improving dark mode in Windows 11, Microsoft is updating Windows Update. For one, the company fixed error 0x800f0983 and the bug where "Update and Shutdown" resulted in a system restart (very annoying). Additionally, Microsoft simplified update names so that users can better understand what updates are getting installed.

Speaking of bugs... Users discovered a very strange issue in Task Manager where closing it results in process duplication. Unaware users could end up with dozens of Task Manager copies eating up CPU and RAM. Microsoft quickly acknowledged the problem and said it would release a fix soon.

Windows 11 version 26H1 is reportedly on the horizon. This week, a new report emerged about Microsoft preparing version 26H1 for the launch of Snapdragon X2-powered devices. The update will include certain new features and changes that will later make it to all systems with the release of Windows 11 version 26H2.

We had plenty of guides published this week. For Windows 11 24H2 and 25H2 users, there is a guide detailing how to enable all the new features in the latest non-security update, including the new Start menu. Before you turn it on, though, beware of some very odd bugs. Additionally, if you plan to upgrade to Windows 11 and you do not know if Secure Boot is on, check out this guide.

If your system does not support Windows 11 officially, you can use third-party bypass tools like Flyoobe, which was recently updated with the ability to update systems with unsupported CPUs. Just keep in mind that the only safe place to download it is GitHub.

If your system is compatible with Windows 11, you can use the official Media Creation Tool app to update. It was recently updated to fix compatibility issues with Windows on ARM and Windows 10.

Windows Insider Program

Here is what Microsoft released for Windows Insiders this week:

In addition to new builds, Microsoft released a new version of the Copilot app, allowing insiders to use Copilot Vision with text input.

Updates are available

This section covers software, firmware, and other notable updates (released and coming soon) delivering new features, security fixes, improvements, patches, and more from Microsoft and third parties.

On October 30, Microsoft announced financial results for the Q1 FY2026, reporting $27.7 billion in profit and $77.7 billion in revenue.

Microsoft recently increased the price of the Microsoft 365 subscription, and certain governments are not happy about that. An Australian watchdog is suing Microsoft for deliberately misleading customers about the ability to stay on the cheaper Microsoft 365 plan.

Microsoft has revealed a new Teams agent called Facilitator. It integrates into Teams meetings to track topics with visual timeline markers, take notes, answer questions, track tasks, draft documents, and a lot more, saving you plenty of time and effort on tedious tasks. Also, you may have noticed that Office apps on iOS (and Android) now feature the new icons. For Apple's operating systems, these apps also received liquid glass support and template filters.

The month of October 2025 is now over, and, as usual, we have several recaps of all the new features and changes. Here is one for Excel, here is one for Teams, and here is one for Windows 11. The latter highlights five cool features and changes that the operating system received in one month.

This week's browser updates include Firefox 144.0.2, which fixed quite a lot of various bugs on Windows and macOS, including startup crashes, OneDrive compatibility issues, and more. Additionally, Firefox is getting smart suggestions in the address bar so that you can get quick information by simply typing something, like your flight number.

Samsung is now in the desktop browser market as well, with the new beta version of the Samsung Internet browser, which is now available on PC. However, the current beta test is limited only to South Korea and the USA.

Finally, Microsoft released Edge 142 with improved autofill, scareware blocker updates, and other changes.

Here are other updates and releases you may find interesting:

• New Microsoft 365 Copilot agent builds apps from simple prompts.

• Anthropic is giving Claude a boost to solve fintech with a new Excel add-in and more.

• Copilot in Microsoft 365 companion apps will get stuff done in Windows 'lightning fast'.

• You can now connect Meta Quest to your PC and have multiple virtual monitors.

• The October 2025 update for Visual Studio 2022 (17.14) is now available.

• Microsoft 365 Copilot's Researcher agent can now take action with 'Computer Use'.

• Microsoft makes it easier to add iCloud accounts to Outlook.

   

Here are the latest drivers and firmware updates released this week:

• AMD Adrenalin Edition 25.10.2 dropped Windows 10 support, adds Battlefield 6 optimizations.

• Intel released new Wi-Fi and Bluetooth drivers with Windows 11 25H2 support.

Reviews are in

Here is the hardware and software we reviewed this week

We had a bunch of reviews this week, particularly in the NAS segment. Steven Parker reviewed the TerraMaster F2-425, a low-cost local cloud backup and streaming NAS, the UGREEN NASync DH4300 Plus, a low-powered 4-bay cloud backup, and its smaller sibling, the DH2300, a two-bay NAS from UGREEN.

On the gaming side

Learn about upcoming game releases, Xbox rumors, new hardware, software updates, freebies, deals, discounts, and more.

Microsoft has updated the Xbox app with some important changes, such as faster download speeds, Advanced Shader Delivery on the ROG Xbox Ally and Ally X, and more.

Electronic Arts launched a new battle royale mode for Battlefield 6. The free-to-play mode is now out, allowing up to 100 players on the biggest arena ever made for the game. Battlefield REDSEC is now available for free on PC, Xbox, and PlayStation.

Nvidia announced new games for the GeForce NOW streaming service. The latest additions include ARC Raiders, The Outer Worlds 2, Wreckreation, Outbreak Island, Keeper, Dark Moon, and more. Keep in mind that you must own these games to play them on GeForce NOW.

The Epic Games Store is giving away two games this week: Bendy and the Ink Machine and Five Nights at Freddy's: Into the Pit. You can grab these spooky titles until next Thursday. As usual, for more deals and freebies, check out this week's Weekend PC Game Deals article.

Other gaming news includes the following:

• 33 Immortals' biggest update ever launched with new weapons, objectives, and more.

• Looks like the Silent Hill 2 remake is finally coming to Xbox.

Source

Hope you enjoyed this news post. Feedback welcome.

Posted Monday 3 November 2025 at 4:00 am AEST (my time).

News posts... 2023: 5,800+ | 2024: 5,700+ | 2025 (till end of October): 5,009

RIP Matrix

Those figures might not raise eyebrows in Silicon Valley, but they’re starting to make Wall Street nervous. 

The issue came to a head this week as Meta reported quarterly earnings, which showed the company’s operating expenses jumping $7 billion year-over-year and nearly $20 billion in capital expense. It was the result of intense spending on AI talent and infrastructure, which has yet to bring in meaningful revenue for the company. When analysts pressed for more specifics, Mark Zuckerberg made it clear the spending was just getting started. 

“The right thing to do is to try to accelerate this to make sure that we have the compute that we need, both for the AI research and new things that we’re doing, and to try to get to a different state on our compute stance on the core business,” Zuckerberg told analysts on the call. “Our view is that when we get the new models that we’re building in MSL in there and get like truly frontier models with novel capabilities that you don’t have in other places, then I think that this is just a massive latent opportunity.” 

If his goal was to reassure investors, it didn’t work. By the end of the call, Meta’s share price had plummeted in value. Two days later, the rout has only deepened. The Meta’s stock dropped 12% by the closing bell on Friday, representing more than $200 billion in lost market cap. 

It’s dangerous to read too much into stock prices, and in strict financial terms, Meta’s quarterly earnings weren’t that bad. ($20 billion in quarterly profit is nothing to complain about.) But this was the first quarter in which Meta’s aggressive AI spending on both talent and infrastructure had a visible impact on the company’s bottom line. Even more alarming was that, aside from a lot of enormous data centers and well-compensated AI researchers, it wasn’t clear what the money actually bought.  

Analysts pressed Zuckerberg on why he was spending so much on AI, and when they could expect to see revenue from the growing spending. But the call came at an odd spot in Meta’s planning, with no clear budget for projected spending and no available product that could anchor a revenue forecast. As a result, Zuckerberg was left with only general claims about the promise of AI.  

“There are going to be all kinds of new products around different content formats, and we’re starting to see that,” he asid during the call. “And then there are the business versions of all these too, like business A … the other part is how more intelligent models are just going to improve the core business and improve the recommendations that we make across the Family of Apps and improve the recommendations in advertising.” 

Meta isn’t the only company spending billions of dollars on AI infrastructure, so it’s worth teasing out why this same spending isn’t spooking investors at Google or Nvidia, both of which had a great quarter. OpenAI is the biggest offender, spending the same amount with far less financial cushion than Meta.  

There really are concerns that we’re creating a bubble, and if we are, Meta’s core business will let it ride things out better than most. 

But if you ask Sam Altman why he’s spending hundreds of billions of dollars on compute, he’ll tell you he’s operating one of the fastest growing consumer services in human history — and one bringing in $20 billion a year in revenue. We can argue about how sustainable the growth rate is (that’s a separate blog post), but there really is a fast-growing product at the bottom of all the OpenAI hype. A fast-growing ARR figure goes a long way to answer questions. 

Meta doesn’t have a product like that, and it’s not clear where it’s going to come from.  

The company’s most powerful AI product is the Meta AI assistant, which Zuckerberg noted on the call has more than a billion active users. But those numbers are surely juiced by the three billion active users on Facebook and Instagram, and it’s hard to see the current version of Meta AI as a competitor to ChatGPT. There’s also the Vibes video generator, which really did boost daily active users, but has limited business impact beyond that.  

The most ambitious project is the Vanguard smart glasses released earlier this month. However, the glasses feel more like an extension of Meta’s Reality Labs work than a real attempt to harness the power of LLMs.  

Put simply, these are promising experiments, not fully formed products. 

It’s telling then that when he was pressed on infrastructure spending, Zuckerberg’s response wasn’t to point to the recent launches, but to focus on the next generation. 

Zuckerberg stressed, while emphasizing the pending impact of the Superintelligence Lab’s new models, that he was very excited about new products.  

“It’s not just Meta AI as an assistant,” he said. “We expect to build novel models and novel products, and I’m excited to share more when we have it.”  

But this was an earnings call, not a product launch, so all he could say was that there would be more to share “in the coming months.” 

As the market response showed that answer is wearing thin.  

To be fair, it’s only been four months since Zuckerberg restructured his company’s AI team, and the new Superintelligence team hasn’t had time to launch an earthshaking AI product yet. But as the company spends billions of dollars to stay competitive in AI, there’s still no clear indication of what role Zuckerberg wants to play in the new industry.  

Will Meta AI use the company’s detailed store of personal data to grow into a ChatGPT competitor? Is Vibes the first step in a consumer entertainment play, building off Meta’s targeted ad system? Or maybe Zuckerberg’s references to “business AI” are hints at a more detailed enterprise play? 

So far, it’s anyone’s guess. Whatever the answer, the pressure is on Meta to find it — and soon. 

Source

Dubai: The UAE has claimed the top global ranking for artificial intelligence adoption in the workplace, achieving the highest rate of AI utilisation among the working population in 2025, surpassing some of the world’s most advanced digital economies, according to a new report from Microsoft’s AI for Good Lab.

The report, which surveyed AI diffusion across more than 30 countries, found that 59.4 per cent of the UAE’s working-age population uses AI tools daily, such as Microsoft Copilot, ChatGPT, and Midjourney, the highest proportion worldwide. Singapore ranked second with 58.6 per cent, followed by Norway and Ireland.

Microsoft’s findings place the UAE in the elite category of global “Leaders of Adoption,” a designation reserved for nations that not only deploy artificial intelligence at scale but also shape its development, regulation, and application across the economy and society. 

The group includes only a handful of countries, among them Singapore and Norway, recognised for building comprehensive ecosystems to support AI advancement, covering legislative frameworks, skilled talent, and advanced technological platforms.

According to Microsoft, the UAE’s ascent to the top spot reflects a decade-long national effort guided by the UAE National Strategy for Artificial Intelligence 2031, which laid the groundwork for integrating AI into public services, business operations, and education.

Key drivers of this achievement include:

•     Progressive digital legislation that enables technology adoption and fosters innovation.
•     Expanding cloud infrastructure and attracting global investment in data centers.
•     Founding Mohamed bin Zayed University of Artificial Intelligence (MBZUAI), the world’s first graduate-level AI university.
•     Support for startups and entrepreneurs in emerging technology sectors.
•     Government-led training programs designed to upskill national talent and build advanced digital capabilities.

The report also notes that the UAE has become one of the fastest-evolving markets globally for AI integration, particularly across government services, finance, energy, transportation, healthcare, and education, sectors that have seen significant efficiency gains through AI adoption.

While the UAE’s achievement demonstrates its rapid digital transformation, Microsoft’s data reveals a striking regional imbalance. 

Within the Middle East, AI adoption rates remain lower elsewhere: Qatar (35.7 per cent), Saudi Arabia (23.7 per cent), Kuwait (17.7 per cent), and Egypt (12.5 per cent). 

In South Asia and Sub-Saharan Africa, adoption rates fall below 10 per cent, hindered by limited connectivity, computing resources, and linguistic diversity.

Language barriers also play a major role, the report suggests, as countries dominated by low-resource languages lag in AI diffusion. 

Nearly four billion people globally still lack access to the infrastructure, from broadband to reliable electricity, necessary to use AI tools effectively.

To support regional progress, Microsoft has expanded its AI for Good Lab to Abu Dhabi, marking the company’s first such facility in the Middle East. 

The lab will collaborate with nonprofits and research partners on issues such as sustainability, public health, and disaster response. 

The move builds on Microsoft’s existing partnerships with the UAE government and Abu Dhabi-based G42 to accelerate cloud and AI infrastructure across the Emirates.

The Microsoft report situates the UAE at the forefront of a historic global transformation. More than 1.2 billion people worldwide are now using AI, making it the fastest-adopted technology in human history. Yet few nations have embedded it as deeply into their economies as the UAE, where AI has become both a tool of efficiency and a cornerstone of future growth.

With an AI market valued at USD 3.47 billion in 2023 and projected to grow at an annual rate of 43.9 per cent through 2030, the Emirates’ model shows how strategic vision, infrastructure investment, and human capital development can turn technology into a national advantage.

Source

Microsoft CEO Satya Nadella said during an interview alongside OpenAI CEO Sam Altman that the problem in the AI industry is not an excess supply of compute, but rather a lack of power to accommodate all those GPUs. He said this on YouTube in response to Brad Gerstner, the host of Bg2 Pod, when asked whether Nadella and Altman agreed with Nvidia CEO Jensen Huang, who said there is no chance of a compute glut in the next two to three years.

“Well, I mean, I think the cycles of demand and supply in this particular case, you can’t really predict, right? The point is: what’s the secular trend? The secular trend is what Sam (the OpenAI CEO) said, which is, at the end of the day, because quite frankly, the biggest issue we are now having is not a compute glut, but it’s power — it’s sort of the ability to get the builds done fast enough close to power,” Satya said in the podcast. “So, if you can’t do that, you may actually have a bunch of chips sitting in inventory that I can’t plug in. In fact, that is my problem today. It’s not a supply issue of chips; it’s actually the fact that I don’t have warm shells to plug into.”

AI’s power consumption has been a topic many experts have discussed since last year. This came to the forefront as soon as Nvidia fixed the GPU shortage, and many tech companies are now investing in research in small modular nuclear reactors to help scale their power sources as they build increasingly large data centers.

This has already caused consumer energy bills to skyrocket, showing how the AI infrastructure being built out is negatively affecting the average American. OpenAI has even called on the federal government to build 100 gigawatts of power generation annually, saying that it’s a strategic asset in the U.S.’s push for supremacy in its AI race with China. This comes after some experts said Beijing is miles ahead in electricity supply due to its massive investments in hydropower and nuclear power.

Aside from the lack of power, they also discussed the possibility of more advanced consumer hardware hitting the market. “Someday, we will make a[n] incredible consumer device that can run a GPT-5 or GPT-6-capable model completely locally at a low power draw — and this is like so hard to wrap my head around,” Altman said. Gerstner then commented, “That will be incredible, and that’s the type of thing that scares some of the people who are building, obviously, these large, centralized compute stacks.”

This highlights another risk that companies must bear as they bet billions of dollars on massive AI data centers. While you would still need the infrastructure to train new models, the data center demand that many estimate will come from the widespread use of AI might not materialize if semiconductor advancements enable us to run them locally.

This could hasten the popping of the AI bubble, which some experts like Pat Gelsinger say is still several years away. But if and when that happens, we will be in for a shock as even non-tech companies would be hit by this collapse, exposing nearly $20 trillion in market cap.

Source

The foundations of the internet were set 56 years ago. However, like many other great moments in history, the story of the first computer-to-computer message sent over a distributed packet-switched network, began with an underwhelming fizz, rather than a bang. Moreover, it involved a computer crash and a ‘typo’ – how thoroughly modern.

If you’ve read or watched anything about the history of the internet, you will have probably heard of ARPANET (the Advanced Research Projects Agency Network). At around 10.30pm on October 29, 1969, two scientists set out to test communications between two of the four systems ARPANET systems, 350 miles apart.

Charley Kline from UCLA intended to send the first message across this embryonic distributed network. He’d settled on just five characters, typing out ‘LOGIN’ for the momentous occasion. However, Bill Duvall at Stanford Research Institute (SRI) only received ‘LO’ before his computer system crashed… Great start.

Kline didn’t realize the magnitude of this achievement at the time. Instead, he and Duvall, instantly went into troubleshooting mode to fix the issue that caused the crash.

Buffer overflow

It turned out that the Interface Message Processors used as a kind of ‘modem’ to connect the ARPANET machines were not ready for the throughput demanded. The IMPs were designed for data transfers of around 10 characters per second, but the ‘LOGIN’ message had been transmitted at 5,000 characters per second, Duvall noted in an interview with the BBC last year.

Ironing out this major issue took the scientists just an extra hour to complete. According to the first-hand testimony, after changing the communication buffer size and rebuilding the system, this distributed packet-switched network could send and receive messages without issues, as intended.

At the time, ARPANET was just a four system network. In addition to the UCLA and SRI nodes, noted above, there were systems at the UC Santa Barbara, and the University of Utah, forming this primordial internet.

Two decades before the WWW

Of course, ARPANET had its roots in military research, designed to sustain communications in the event of a nuclear attack. It would be decades before the internet would become a consumer-facing utility.

For example, between 1969 and the founding of the first consumer ISPs, we would see key innovations like Email (1971), the TCP/IP protocol development and standardization (1973-83), the DNS system (1983), the rollout of NSFNET in the mid to late 80s, and the invention of the World Wide Web (HTML, HTTP, URLs) around 1990.

In 2025, the internet feels like an economic and societal essential of modern business and life. And all this started with a stuttering ‘LO.’

Source

Paul's Hardware (1.53M subscribers)

November 2, 2025

Video length: 8m 41s

0:00 Welcome to Paul’s Tech News - November 2, 2025

1:32 NVIDIA Hits $5 Trillion 

3:27 Memory and Storage Prices Continue to Go Up Because of AI 

TECH BRIEFS 

4:50 AMD Zen6 Ryzen “Medusa” CPUs Confirmed 

5:35 AMD confirms that its older RDNA 1 and RDNA 2 GPUs are now in “maintenance mode” 

6:29 ASUS’ dual-mode QHD 540Hz and HD 720Hz monitor listed in China at $1,100 

7:28 Make Your GPU Sing - OCCT Coil Whine Detector Plays Songs

Source

Hope you enjoyed this news post. Feedback welcome.

Posted Sunday 2 November 2025 at 5:15 pm AEST (my time).

News posts... 2023: 5,800+ | 2024: 5,700+ | 2025 (till end of October): 5,009

RIP Matrix

An ingenious developer has figured out how to run Windows 95 in a browser window, no installation required. Originally, to run Windows 95, you needed a computer with a minimum of 4 MB of RAM. You also needed a 386DX processor or higher, and 50-55 MB of free hard drive space. To be fair, recommended requirements for optimal performance were higher at 8 MB of RAM, a 486 processor, and a bigger hard drive. Today, things are a little different. 

Windows 11 requirements are more substantial as outlined in everything you need to know about the newest Windows OS. As for Windows 95 on browser, it's not just the core OS. Using the tool, you can install packages spanning a variety of software, from Zork 1, 2, and 3, to Norton Utilities 2.0. They're certainly dated compared to today's standards, but it's a great way for those of us who remember Windows 95 to look back; and a great way for those too young to remember to explore old-school computing applications.

What can you do in the browser version of Windows 95?

The tool (or window, rather) takes a minute to load when you first visit the page, so wait a few minutes until you see Windows 95 has finished booting. From there, you can interact with anything in the operating system as if it were a mini computer. Click Full Screen in the top right of the smaller window to make it take up the entire monitor view for your primary monitor. At the bottom, just below the window, you'll see a space to choose disks to load, simulating the insertion of floppy disks into the machine.

If you want to use one of those disks, or one of those programs, select it from the dropdown and then click Load. For example, selecting "Oregon Trail" will load that disk. To explore the loaded disk or run programs, double click My Computer > File (at the top) then Explore. In the left hand pane of the window that appears, select the disk you loaded. If you didn't change, it should be drive A: Then, double-click the .EXE file (executable) for the program you want to run. 

Ultimately, you can play games, install and use various programs, and explore Windows 95 as if it was freshly installed on a legacy computer. The experience is still quite buggy, and some of the features and programs are not going to work, unfortunately. But it's definitely an eye-opening and educational experience, especially if you've never used Windows 95 before. Always remember the most important part: have fun! With the end of Windows 10 drawing near, the latest versions may eventually run in emulators like this. Maybe, maybe not. It is cool to think about, though.

Source

I've been having a blast with The Outer Worlds 2, and it certainly feels more stable than a lot of high-profile video game releases these days — but I'd never go so far as to claim it's perfect.

I've encountered my fair share of bugs and quirks in Obsidian Entertainment's latest masterpiece, and one in particular annoyed me. Well, I suppose The Outer Worlds 2 team is listening in on me, because they just released a new patch that fixes that exact bug.

If you've been locked outside Otto's saferoom and believed you may just be an idiot, fret no more. It was a bug, and the latest update for The Outer Worlds 2 fixes it... Oh, and I guess there are quite a few other crucial patches in this release, too.

Finally, I can get past this dumb lock

The 2019-released The Outer Worlds was Obsidian’s first first-person RPG to come out since Fallout: New Vegas in 2010. The relatively short sci-fi RPG went for a humor-filled story set in a megacorporation-owned star system. Despite its smaller scope and rather meh combat, I enjoyed my time with the interesting companions, fun quests, and ample amount of satire. Now, with Microsoft by its side, Obsidian is back with a sequel that’s promising everything to be bigger and better, and with several dozen hours now under my belt, I can definitely say that it has delivered on that. As for whether that has translated into a better game, you’ll have to read on to find out.

New system, same universe

Gone are the days of randomly stumbling into being a captain of a space vessel, a la the original RPG. The sequel is set up in a new corner of the same hyper-consumer galaxy, but this time as an official agent of the Earth Directorate, a faction that is supposed to be the voice of Earth in human colonies in distant systems. Arcadia is the new playground, and a fantastic intro sequence that serves as a tutorial, quickly gets me hooked into this mildly absurd universe.

Despite the multitude of face, hair, body, and other customization options, the appearance part of character creation goes by fast. Third-person perspective is an option this time, but loot-hunting and first-person shooting are more my thing. The real meat of any RPG is in the aspects we can choose to shape the protagonist, and The Outer Worlds 2 does not disappoint here. Backgrounds, Traits, and Skills all come together to make my own prospective hero that could probably save the day by saying the right things if he’s lucky.

Usually in RPGs, if I choose to go with a character build that’s not focused on beating speech and hack checks, I feel like I lose out on a lot of content and ‘good endings’ that are just out of reach. The Outer Worlds 2 is where I felt that same stab, even while playing a character that could talk his way out of anything.

I came across jammed doors that needed strength or engineering, peculiar mixing machines requiring medicine or science knowledge, and plenty of questlines that wanted bizarre skill checks to have a more delicate ending. For the most part, there were still multitudes of ways to get inside locked buildings or find solutions to quests, but I think this is one of those rare RPGs where I could role-play as somebody or a nobody and still feel like the game didn’t forget about catering to my playstyle.

A certain catastrophe or two later, the wider world of the RPG quickly opens up, revealing the massive explorable areas that each planet receives in place of having a single connected map. It’s a good approach that fits in well with the space sci-fi nature of the title. I should say that I couldn’t connect with this zone system when Avowed used it, making the transitions jarring, but here, it feels more natural.

Setting the wildly important main questline aside for later, I dived headfirst into this system’s history and current happenings. War has decimated most of the livable areas here. Auntie's Choice is one of the biggest factions I quickly run into, a megacorporation that lives and dies by its consumerist rules and is easily the funniest of the bunch. Almost everyone, from its foot soldiers to its leaders, is obsessed with increasing profits, listening to advertisements for the latest food-adjacent snacks, and making up cutthroat schemes to rise through the ranks.

On the other end of this boxing ring is the Order of the Ascendant faction. Possibly one of the most unique groups I’ve faced in an RPG, these highly religious folks are attempting to find a grand equation that can solve the universe and get rid of its random possibilities, ultimately making the future predictable via math.

If you’re familiar with The Hitchhiker's Guide to the Galaxy’s Ultimate Question or Psychohistory from the Foundation, you should see where this is going. The ongoing war between the two factions is a major part of the RPG, one where I could help either side or not give a care about it. There are also multiple other groups involved in the solar system that are playing their own games while vying for power or simply making profits. To keep spoilers to a minimum, I won’t go into detail on these or the wider plotline.

However, not every person I meet in these conflicts is crazy or a warmonger. Through work, smooth talk, or straight-up blackmail, I was able to make individuals see how their company or belief system is failing to bring about peace. At the same time, I could also go full maniac and wipe out entire sides completely, a la Megaton, or have everyone hate me equally instead. There’s not much that the game doesn’t let you do.

Obsidian’s writing shines the most when there’s humor involved, making even random quests and interactions with minor NPCs memorable. I could be bargaining for a top hat as the reward, whilst the fate of an entire town hangs in the balance, and at the next moment, convincing a plumber to help perform surgery at a nearby refugee center. I have lost countless hours scouring the world just to talk to more people.

Immersive and Reactive

Thanks to having multiple planetary bodies and massive complexes, there are drastic changes between biomes and locales. There are black market smugglers working out of a dead husk of a space station, soldiers fighting out of bio-weapon poisoned trenches, gorgeous cave systems crawling with alien crabs, stretches of desert wastelands that have been sucked dry of resources, and so much more to explore.

Even 30 hours into the game, I was running into new companions and vital new gameplay abilities. This might be because I didn’t realize just how much time I was spending exploring every nook and cranny or talking to the dozens of NPCs I came across. Even still, the quality of writing and quests did not dip throughout all these hours. I cleaned up war crimes, bribed officials, helped with prison breaks, and solved countless mysteries while having a delightful time. All of this may sound standard for an RPG of this size, but it’s the Flaws system, returning from the original game, that adds some spice to the gameplay loop.

When the game detects I am doing something often or falling prey to some gameplay mechanic a little too much, it presents a choice: accept it as a flaw with negative and positive side effects or ignore it completely and keep going as usual.

As an example, since I like to sneak around a lot to assassinate or break into places, I was offered the Bad Knees Flaw. This involves making a horrid, and accurate, crunching noise when crouching, alerting people nearby, but also makes sneaking around much faster than usual. Another favorite involved my habit of reloading too much, which gave me increased magazine capacity but lower damage if I finished firing a magazine. Of course, it’s all optional, but who wouldn’t want to live on the edge a little?

As the space adventure continues, companions can slowly fill up the rooms of my small space vessel. Some are individuals who don’t have anywhere else to go, some have their own goals in life and want to use my resources to get there, and a few are just outright crazy. It’s a fun little mix, and I found myself switching around who I was taking into the wild often to match situations or factions I’d be facing.

They interject often into conversations with their opinions and sometimes propose new solutions that might be against the values of the original quest-giver, or even my own. My favorite companions, a Judge Dredd-type executioner and an undercover superspy old lady, were favorites because they were simply fun and immersive to be around. The duo would often banter in the background and comment on what’s happening with ongoing quests.

Speaking of immersion, one aspect that modern Fallout games utilize to great effect is the in-game radio. Trekking across giant wastelands on foot, ducking into random structures to find loot and looking through every drawer for an hour can be fine with background music, but with a radio, that suddenly became an hour of immersive gameplay that just slips by without a worry.

The Outer Worlds 2 radio doesn’t have a DJ or classic songs of that caliber, but it comes close. Custom-made jingles for various products of this universe, musical numbers, and announcements regarding current war efforts are just a few programs that run on the three primary stations.

Jingles about the bottom line being the most important and advertisements for food-adjacent snacks from a certain corporation had me stop combat just to listen in. Soothing gospels about how many digits are in Pi and the love letter to the Pythagoras theorem had me in stitches when I paid attention to the lyrics.

All this combines to make The Outer Worlds 2 an extremely immersive universe to melt into. However, I do recommend playing it on a lower difficulty, and that’s to keep the bullet sponginess to a minimum.

Combat and stealth as a whole is deeper than I thought it would be, with bullet time, shields, body dissolving to hide tracks, various status effects, and wild weapon types available. But even wild animals I came across can eat up magazine after magazine of fiery bullets and get stomped on by two companions, just to shake it all off.

It’s an RPG, so I understand that having more powerful weapons is the answer, but it seemed like each area had foes that were just a tad too powerful for the latest and greatest weapons I’d be looting, crafting, or receiving as rewards. I am hoping Obsidian can tweak this a little.

Performance and Visuals

The Outer Worlds 2 is a beautiful RPG, and it’s clear that Obsidian has paid attention to even the smallest details. Every piece of machinery, propaganda poster, rubble, and office desk has a huge amount of detail close up, and these are the things that not many would even bother to look at.
The massive number of available weapons has the same quality, with the added bonus of intricate animation work to make them feel like actual sci-fi killing machines. Obsidian has piled on almost every effect it can think of for these weapons, so prepare for a fireworks show.

The animations as a whole are a joy to look at. Things like the manual computer hacking, lock picking with shaky fingers, and engineering doors our character opens up to crank and twist random things in, all have a humorous but deliberate feel to them, as if my character is saying with a posh accent, “Why of course, plugging a giant shaft and revving it is exactly how I hack computers.”

Voices, sound effects, and music have also been top-notch. NPCs love talking to each other about the latest happenings in their lives or the actual game world, and numerous times I’ve waited around in areas just to see what hilarious nonsense they come up with. I already mentioned the radio being a highlight, but the regular music that plays in the background has been smooth to listen to as well. The new orchestral score mixes with the series’ main theme to great effect.

The Outer Worlds 2 is an Unreal Engine 5 game, so it does have that look to the environment and NPCs. It’s not a bad look, just something familiar. Thankfully, stutters were nonexistent during my time with the game, as the rather extensive shader compilation session took care of that with the first start handily. It’s a small price to pay for having a smooth gameplay experience.

However, if there is one thing I am disappointed by in Outer Worlds 2, it’s the performance. I have an AMD Radeon RX 9070 XT 16GB paired with an eight-core Ryzen 7 3700X and 32GB of RAM, with the game running at 1440p resolution. Just to hit 60 FPS, I had to utilize the Balanced upscaling option on FSR 4, coupled with a mix of High and Very High options. I’m essentially putting up with some minor shimmery edges, thanks to the upscaling, instead of having blurry shadows, less vegetation, and crowd density. The ray tracing option is something I would steer way clear of until either there are massive optimization updates or graphics cards from a few years in the future.

Now we come to the RPG’s bugs. Sure, there’s the standard jank you can expect from large games such as this. I’ve seen NPCs walking half in the ground, bodies stuck in the air, and occasional disappearing loot.

However, I also encountered a weird bug that broke things heavily. This involved losing the ability to save my game completely. Neither quick save nor manual save would function, with only the load option available. By the time I realized this was happening, I had already spent over an hour exploring and finishing quests, so even though it fixed itself when I reloaded the last save, it could have been a much larger loss of progress and my time. Thankfully, the bug hasn’t resurfaced since then.

Conclusion

Coupled with great RPG design and solid writing, The Outer Worlds 2 is one of the best RPGs I have played in the past few years. The decisions I have made have led my character through twists and turns that have ended entire towns and given power to the craziest of corporations. Like in iconic classics of the past, the game does not pull any punches when it comes to choices, where sometimes, there’s just no easy way out or a scenario where everything works out.

The less-than-ideal performance on PCs running modern hardware is something I keep seeing in AAA games lately, and The Outer Worlds 2 is not breaking that tradition. Thankfully, except for the aforementioned save-related bug, my experience has been relatively smooth with no stutters to speak of. One other negative is the bullet sponginess of enemies, which can drag down the cool feeling of coming across special weapons only to realize they need two magazines just to kill one random enemy.

Immersive companions with actual opinions, the hilarious radio stations, and the returning Flaws system are all brilliant and deeply immersive. This might be the most New Vegas-feeling RPG since New Vegas. As long as your system can handle it, or if you’re on a console, I highly recommend any RPG fan to check out The Outer Worlds 2.

The Outer Worlds 2 is available now on PC via Steam and Microsoft Store, as well as Xbox Series X|S and PlayStation 5, with a $69.99 price tag. It is also a part of Xbox Game Pass Ultimate and PC Game Pass subscriptions.

This review was conducted on the PC version of the game provided by Microsoft.

Source

Hope you enjoyed this news post. Feedback welcome.

Posted Saturday 1 November 2025 at 5:51 pm AEST (my time).

News posts... 2023: 5,800+ | 2024: 5,700+ | 2025 (till end of October): 5,009

RIP Matrix

While enterprises continue to respond to the end of life of Windows 10 and the wide scale adoption of Windows 11, there are signs that some sectors are finding the transformation difficult.

A report in Digital Health News revealed that while a UK health authority has managed to upgrade most of its machines to the newer operating system, it is unable to complete the process, as a certain number of devices do not support Windows 11. 

According to the report, Rotherham Health Trust was faced with a hefty bill to render equipment usable, with one supplier quoting £25,000 (about $32,900) to upgrade a three-year-old device.

The Rotherham incident is not an isolated one, according to a post on independent community website Windows Forum, which noted, “a handful of clinical suppliers refused or failed to certify their software for Windows current OS.”

These medical systems operate in environments which make their upgrades more difficult, said Timon Watson, director analyst with Gartner. “These devices will operate on isolated networks with no access to the outside world and [are] managed with dedicated tools and processes. Any changes to these environments require extensive testing and validation, which may involve external parties and take months to complete.”

He added that the healthcare industry has specific requirements that influence any change. “Computers connected to machinery or medical devices are often purchased with and are considered as part of that equipment by the vendor,” he said. “The vendor will retain responsibility for maintaining the computer, meaning any software changes or OS updates may require professional services, which can cost thousands of dollars.”

Vendors in other industrial sectors have also discovered compatibility issues. Rockwell, for example, put a note on its website saying that its ASEM 6300P and 6300B industrial computers are generally not compatible with Windows 11 due to their older processors. There are sure to be other industrial vendors who will have the same issues.

There are other signs that in the wider world, organizations are not upgrading their systems with any great haste. The maritime sector is one that is resisting the change; according to research from Marlink, nearly half of all vessels are yet to adopt the new operating system, putting shipping fleets in danger of cyber attacks in the future.

However, according to Chris Silva, a workplace security specialist with Gartner, it could be some time before there is any impact. “While end of life of an operating system can have implications for the security posture of the OS itself, that impact is not immediate.”

In the meantime, he said, there are plenty of third party vendors who will offer support to secure older and discontinued operating systems that continue to be essential to organizations.

Microsoft has also considered the needs of those companies that have found it difficult to adopt Windows 11 in every aspect of their infrastructure. It has introduced Windows 10 IoT edition, offering support up until 2032, which will meet the needs of many industrial customers.

For those companies that have not gone down this route, there will be decisions to be made. Health authorities faced with the lack of support for Windows 11 on their current equipment may decide that these devices may have outlived their usefulness.

“Organizations in the future may not upgrade their existing equipment at all,” said Watson. “This means new equipment would need to be purchased, which, using an MRI scanner or X-ray equipment as an example, could cost hundreds of thousands or even millions of dollars.”

Source