Nvidia has announced an increase in its remaining buyback capacity to an impressive $235 billion, a figure it anticipates utilizing through the fiscal year 2028. The company reported having $22.44 billion in cash and cash equivalents at the close of the July quarter.

In terms of valuation, Nvidia’s shares are currently trading at approximately 16.5 times their projected earnings over the next year. This marks the lowest forward earnings multiple the company has seen since January 2015, significantly below the 15-year average of 30, according to data from LSEG. This dip in earnings multiple has led some analysts to express concerns regarding diminishing expectations for profit growth.

In early trading, stocks of Nvidia, headquartered in Santa Clara, California, experienced a rise of over 3%. So far this year, the company’s shares have soared more than 20% up until Friday's market close. 

The additional buyback authorization arrives at a time when stock repurchases have decreased by approximately 50% from July through September 23. Nvidia’s CEO, Jensen Huang, stated, “Our cash generation provides us the ability to invest in technologies that drive this transformation while also returning capital to our shareholders.”

The $150 billion increase in Nvidia's buyback plan surpasses the market capitalization of around 84% of the companies listed on the S&P 500, as per LSEG data. Last month, Nvidia projected an impressive 70% revenue growth for fiscal 2028, alleviating investor concerns regarding the sustainability of the recent surge in AI spending following years of rapid expansion.

Furthermore, Nvidia is actively investing in AI startups and cloud services, which has drawn attention from investors who question if this strategy indirectly bolsters the demand for its processors. The company's solid cash reserves of $22.44 billion highlight its strong financial position moving forward. 

This summary incorporates contributions from reporters Anhata Rooprai and Shashwat Chauhan, with editing by Shreya Biswas, Christian Plumb, and Anil D'Silva.

Nvidia increases share repurchase program by unprecedented $150 billion due to AI-driven growth.

A bright, modern boardroom at sunrise: a large metallic Nvidia logo sits center on a sleek wooden table, flanked by stacks of green share certificates stamped “$150 B” with subtle circuit-chip watermarks. Warm sunlight floods through floor-to-ceiling windows, illuminating an ascending stock chart drawn on the glass wall, with a city skyline beyond.

Nvidia boosts share buyback by record $150 billion as AI boom fuels growth - Reuters

- Nvidia's buyback capacity increases to $235 billion, expected to be utilized through fiscal 2028.  
- The company holds $22.44 billion in cash and cash equivalents at July quarter's end.  
- Nvidia shares rose over 3% as forward earnings multiple hits lowest since January 2015.  

Summary

Companies Remaining buyback capacity rises to $235 billion, which Nvidia expects to use through fiscal 2028

Nvidia ended July quarter with $22.44 billion in cash and cash equivalents

Forward earnings ​multiple sits near 16.5, lowest since January 2015, LSEG data shows

Shares of Santa Clara, California-based ​Nvidia rose more than 3% in early trading. The stock ⁠has gained more than 20% this year through Friday's close.

Sign up here.

The additional authorization lifts ​Nvidia's remaining buyback capacity to $235 billion, which it expects to use through fiscal ​2028, as surging demand for AI training and inference fuels cash generation.

Nvidia shares were trading at about 16.5 times 12-month forward earnings, their lowest multiple since January 2015 and well below ​the 15-year average of 30, according to LSEG data, which some analysts ​see as a sign of slowing profit-growth expectations.

The announcement comes amid a slowdown in stock buybacks, which ‌fell ⁠roughly 50% from July through September 23.

"Our cash generation gives us the capacity to invest in the technologies that advance this transformation and return capital to shareholders," CEO Jensen Huang said in a statement.

Nvidia's $150 billion buyback increase exceeds the ​market capitalization of about ​84% of the ⁠S&P 500 constituents, according to data compiled by LSEG.

Last month, Nvidia forecast about 70% revenue growth for fiscal 2028, reassuring ​investors who have questioned how long the AI spending surge ​can last ⁠after years of explosive growth.

The company has also been investing in AI startups and cloud providers, drawing scrutiny from some investors over whether such funding indirectly supports ⁠demand for ​its own chips.

Nvidia ended the July quarter with $22.44 ​billion in cash and cash equivalents.

Updates ‌shares in paragraph 2, adds details in paragraph 7

Reporting by Anhata Rooprai in Bengaluru, additional reporting by Shashwat Chauhan ​and Sruthi Shankar in Bengaluru; Editing by Shreya Biswas, Christian Plumb and Anil D'Silva

Our Standards: The Thomson Reuters Trust Principles. , opens new tab

Nvidia boosts share buyback by record $150 billion as AI boom fuels growth

WATERVILLE (WGME) -- The influence of artificial intelligence (AI) is rapidly growing in various sectors, including the workforce and higher education.

In a significant move, Thomas College has secured a $3 million grant from the Department of Labor aimed at weaving AI into its academic offerings. The college's ambition is to establish itself as a "college of the future" in this AI-dominated era.

According to President Jeannine Diddle Uzzi, the college plans to introduce new roles, courses, and even specialized majors focused on AI.

“Graduates entering the workforce today may work until about 2060 or 2070, and it's challenging to foresee what those years will entail,” Diddle Uzzi observed. “Our current priority is to forecast the impact of AI across various sectors, from healthcare to the arts, journalism, and business.”

In addition to earning degrees, the college aims to ensure that every student graduates with a strong understanding of AI.

The federal funding will facilitate the hiring of key staff, such as an AI specialist who will collaborate with faculty to seamlessly incorporate AI concepts into various courses.

The institution is also contemplating the development of entirely new majors, including one focused on AI management.

“In every sector, there will be a demand for employees who can efficiently, effectively, and responsibly utilize AI, as well as manage its outputs, which require human oversight,” said Diddle Uzzi.

Thomas College intends to leverage this grant to enhance students’ insights into the disruptions that AI will bring to different industries while opening up new career paths in the AI field.

“This technology is certainly here to stay, much like computers and cell phones have become integral to our lives,” Diddle Uzzi noted.

While it is still unclear how many additional courses or new majors will emerge from this grant, the college aims to integrate AI classes with existing business programs, emphasizing their interconnectedness in today’s job market.

Thomas College receives federal funding to embed AI into its educational programs.

A sunlit college classroom with tall windows casting warm morning light onto a wooden desk piled with textbooks stamped “AI,” a middle-aged professor writing a simple neural-network diagram in white chalk on a green chalkboard, and a U.S. flag hanging beside a framed grant certificate on the wall.

Thomas College accepts federal grant to integrate AI into school curriculums - WGME

- Thomas College received a $3 million grant to enhance AI curriculum and literacy.  
- New positions and majors, including AI management, will be developed for students.  
- The college aims to prepare graduates for future careers impacted by AI advancements.  

WATERVILLE (WGME) -- AI is rapidly becoming more integrated into everyday life, including the workforce and even higher education.

Thomas College has accepted a $3 million federal grant from the Department of Labor to incorporate AI into its own curriculum. The college wants to make itself a so-called "college of the future" in the age of AI.

To do that, President Jeannine Diddle Uzzi says the college will create new positions, new classes and even majors.

“Students who graduate today are working until about 2060, 2070. We can’t imagine what those years will look like,” Diddle Uzzi said. “One thing we’re all trying to do right now is to predict what will be the impact of AI on every field from medicine, art, journalism, all of our business practices.”

Right alongside a degree, the college wants every graduate to leave with AI literacy.

They are using a federal grant to create new positions, like an AI specialist in residence. The worker in that position would work with faculty to help integrate it into their classes.

The school will also consider creating completely new majors, including AI management.

“There will be needed, in every industry, a certain group of employees who know how to use AI efficiently, effectively and responsibly. And then how to manage AI, because AI spits out whatever it spits out. It produces things human beings have to manage and check,” Diddle Uzzi said.

Also read: New training needed for violent student behavior

Thomas College wants to use this grant to help students better understand disruptions AI will cause to industries and open paths for students to pursue careers in AI.

”This is certainly not going away, just like the computer didn’t go away, the cell phone didn’t go away," Diddle Uzzi said.

It's too early to say how many new courses or majors the college will create with this grant. But the school also hopes to pair new AI courses with existing business classes, saying they go hand-in-hand in the modern job market.

Thomas College accepts federal grant to integrate AI into school curriculums

On Monday, tech powerhouse Meta unveiled its latest premium AI models, Muse Wonder and Muse Wonder Full, as part of its commitment to explore what CEO Mark Zuckerberg has termed the “AI frontier.” This move places Meta in direct competition with established players like Anthropic, OpenAI, and Google in the fast-paced race to innovate in artificial intelligence.

These new offerings build on the previously released Muse Spark AI model and are touted as the company’s most advanced to date. Meta claims that these paid models can efficiently manage a variety of tasks, from processing insurance claims to aiding renters in understanding lease documents, and even developing fully functional applications.

Despite the optimism surrounding these launches, experts view Meta as attempting to catch up with its rivals. Katelyn Chedraoui, a noted AI reporter from CNET, indicated that while Meta is ramping up its AI toolkit, the introduction of these models comes amid significant safety concerns that have emerged in the broader AI landscape. Both internal and external safeguards will be crucial for any new technology rollouts.

Calls for external oversight are gaining momentum among researchers and industry leaders, including Dario Amodei of Anthropic and Sam Altman from OpenAI, who have expressed the need to pause AI developments until more regulations are in place. Contrarily, Zuckerberg maintains that AI labs are capable of self-regulation to mitigate potential risks. In a recent Financial Times interview, he emphasized his belief in the responsible engineering of AI technology, contrasting it with the more dystopian view of AI acting autonomously.

In addition, Meta has introduced Muse Charm AI hardware, a device resembling a keychain or pendant.

The Muse Wonder model is described as a top-of-the-line AI solution aimed initially at developers who can access it through the Meta Model API, with monthly subscription plans ranging from $20 to $100. Future updates promise to enable Muse Wonder’s capabilities in Muse, incorporating a flexible pricing approach based on per-million-tokens use.

Meta asserts that Muse Wonder enhances the foundational features of Muse Spark, offering improved processing capabilities and advanced applications in sectors like law, finance, and healthcare. The model is designed to accurately analyze leasing agreements and manage insurance claims; however, it’s important to remember that using AI for healthcare or mental health should not replace professional advice.

Two significant new features are part of Muse Wonder: Active Memory and Active Focus. Active Memory allows the model to retain information from ongoing conversations, which raises privacy concerns, but Meta insists it has designed the model to prevent unauthorized sharing of sensitive information and mitigate cybersecurity risks. The company claims this could also help developers identify vulnerabilities in their codes and bolster their security protocols.

Active Focus addresses challenges the AI might encounter by keeping it functioning even when it hits technical hurdles. This capability enables Muse Wonder to explore alternative solutions, seek user clarification, and handle intricate tasks more efficiently compared to its predecessor, Muse Spark.

However, the notion of an AI model deftly maneuvering around possible restrictions is reminiscent of recent actions taken by Anthropic, which restricted access to its AI systems, Mythos 5 and Fable 5, following strict government regulations.

The Muse Wonder Full model is aimed at business users and power users and is considered a more potent iteration of the Wonder model, with the same features but enhanced reasoning capabilities and additional processing strength for handling complex undertakings. Unlike Wonder, which has a public API, Muse Wonder Full is exclusively available for developers subscribed to the Muse Code AI agent, with subscription fees ranging from $50 to $200 per month.

Meta Introduces Muse Wonder AI Models to Compete with Competitors

A bright, airy modern studio flooded with warm morning light. At its center, a gleaming marble muse sculpture with subtle circuit-like veins emerges from a sleek crate emblazoned with a large Meta logo. Silhouetted tech professionals stand back in awe as dust motes dance in the sunbeams, blending classical art with cutting-edge AI innovation.

Meta Launches Muse and Wonder AI Models to Play Catch-Up With Rivals - CNET

- Meta launched Muse Wonder and Wonder Full, its most advanced AI models to date.  
- New features include Active Memory and Active Focus for improved task handling and interactions.  
- Meta aims to catch up in the AI race despite ongoing safety and regulation concerns.  

Tech giant Meta debuted its new Muse Wonder and Muse Wonder Full premium AI models on Monday, continuing a push toward what CEO Mark Zuckerberg recently called the “AI frontier,” where the heavy hitters in the artificial intelligence space have dominated.

The models build upon the Muse Spark AI model, launched earlier this year. Meta calls the new paid tiers its “most capable models to date,” reporting the models can perform personal and professional agentic tasks ranging from handling insurance claims to helping renters understand a lease agreement to building entire working apps.

Playing catch-up

Meta says its new models, the fruits of billions of dollars invested over the last year, leapfrog the competition. But it’s playing catch-up with Anthropic, OpenAI and Google in the AI race.

More from CNET

“Meta is making up for lost time with lots of new AI tools,” Katelyn Chedraoui, CNET’s expert reporter on artificial intelligence, told me. “But this new frontier model comes amid serious safety concerns in the AI industry. Safeguards will be essential to any new launch.”

Many researchers and experts are advocating for external evaluation and regulation of the technology, with some AI executives, including Anthropic CEO Dario Amodei and OpenAI’s Sam Altman, making high-profile calls to slow AI development. Meanwhile, Zuckerberg has been vocal in his belief that labs can and should self-police to stave off the worst effects of AI.

“I’m much closer to the belief that AI is a technology that can be engineered with normal guardrails and built responsibly than the anthropomorphic vision of AI, which is that it is just going to go do its own thing,” Zuckerberg told Financial Times in an interview this week.

Meta recently announced the Muse Charm AI hardware, which looks like a keychain or pendant. Meta/CNET

Muse Wonder and Wonder Full

Muse Wonder is the “state-of-the-art” premium-tier AI model from Meta. It is available first to coders who pay for access to the Meta Model API, via the Power Plan at $20 per month or the Maximum Plan at $100 per month. However, Meta’s announcement says it “will soon power experiences in Muse” with more flexible per-million-token pricing.

Meta says Muse Wonder builds on the core capabilities of its Muse Spark AI model with more advanced features, greater processing power and improved capability in law, finance and health applications. Meta gives the example of Muse parsing a leasing agreement or handling an insurance claim. (To be clear, do not rely on AI for healthcare or mental health.) Muse Wonder’s new “frontier coding performance” will also enable the agent to assist in building games and apps.

Muse Wonder introduces two new features: Active Memory and Active Focus.

The first, Active Memory, enables the model to remember and carry information across conversations. Storing potentially sensitive data immediately sets my privacy-conscious Spidey Sense on edge. That said, Meta says it has trained its Muse Wonder model to avoid sharing private information without permission and has hardened it against cyberattacks. The company even says that “developers and security teams can use models like Muse Wonder to find flaws in their own code and strengthen their systems,” which — I gotta be honest — sounds to me like the peak tech hubris.

Meta’s CEO, Mark Zuckerberg, has been vocal in his belief that labs can and should self-police to stave off the worst effects of AI. James Martin/CNET

The second new feature, Active Focus, keeps the AI agent working when it runs into technical dead ends. With Active Focus, Muse Wonder is said to be capable of alternative approaches, recovering from errors and asking the user for clarification. That should let the model handle more complex tasks than Spark and encounter fewer roadblocks, which we’ve seen firsthand in our testing.

But boasting that your AI model can more effectively circumnavigate what may be blocks put in place (perhaps for a reason) sounds eerily familiar. Recently, Anthropic revoked access to its AI models, Mythos 5 and Fable 5, after the US government imposed sweeping restrictions on who could use them.

Muse Wonder Full is said to be an even more powerful version of the premium Wonder model with all of the same new features, but with a shift in focus toward businesses and power users. Basically, Wonder Full unlocks more flexible reasoning, a tweaked base model and access to more behind-the-scenes processing power, making it a better fit for unfamiliar tasks, engineering and professional workflows.

With no public API, Muse Wonder Full is available exclusively to developers who sign up for the paid Muse Code AI agent and is bundled into a monthly subscription priced from $50 to $200.

Meta Launches Muse Wonder AI Models to Play Catch-Up With Rivals

Artificial intelligence is transforming the way individuals seek support for eating disorders. In an insightful discussion, Johanna Kandel from the National Alliance for Eating Disorders shares valuable information about recognizing symptoms, available assistance, and other critical topics. This innovative approach is making it easier for those in need to access vital resources related to eating disorders.

AI is transforming the way individuals seek assistance for eating disorders.

A sunlit home office where a thoughtful adult sits at a wooden desk, morning light pouring through a window. On her laptop screen, a warm-toned AI chat interface shows supportive messages—“You are not alone,” “Professional help resources”—and a soft golden glow subtly radiates from it. Nearby, an open notebook and a small potted plant hint at calm reflection and hope.

How AI is helping people find eating disorder support and resources - USA Today

- AI technology is transforming access to resources for eating disorder support.  
- Johanna Kandel emphasizes recognizing signs of eating disorders and seeking help.  
- The National Alliance for Eating Disorders is leveraging AI for better outreach.

AI is changing how people search for eating disorder help

AI is changing how people find eating disorder resources. Johanna Kandel of the National Alliance for Eating Disorders talks signs, help and more.

AI is changing how people search for eating disorder help

The wealthiest individual on the planet harbors a concept for the future of the universe that seemingly sidelines humanity, and he’s pushing forward with plans for his AI enterprise to bring it to fruition.

Elon Musk shared via an X post on Sunday his intentions for a substantial semiconductor manufacturing facility that his companies, Tesla and SpaceX, are establishing in Texas. He stated that this factory is projected to generate an astonishing output worth at least $1 trillion, putting it on par with the economic output of some of the largest nations in the world.

Musk’s ambitious forecasts often walk a fine line between bold and outlandish; he has a longstanding history of making overblown claims. In fact, during recent litigation with the SEC, Tesla's attorneys argued that Musk's exaggerated statements fall under the category of “corporate puffery.”

The newly initiated Terafab project, located approximately 70 miles northwest of Houston, is designed to manufacture chips for Tesla vehicles and humanoid robots, along with the newly combined SpaceX and xAI. Currently, the advanced AI chip design market is predominantly controlled by Nvidia, while fabrication is primarily handled by Taiwan Semiconductor Manufacturing Company (TSMC).

While Musk’s unpredictable remarks warrant skepticism, it’s important to recognize that his companies lead the domestic electric vehicle market and account for a significant portion of recent space launches. This raises an intriguing question: What future does Musk envision that his Texas semiconductor facility will help create?

According to the Terafab website, the aspiration is to transform into a “galactic civilization” capable of “harnessing all the energy in a galaxy.” However, Musk’s recent comments prompt curiosity about who exactly constitutes “we,” particularly in light of a post he made suggesting that “propagating super intelligence to the stars is a great success condition for a biological bootloader.”

In simpler terms, his perspective appears to suggest that humanity’s ultimate goal is to develop AI—guided by Musk—whose advancement may overshadow human existence while it embarks on adventures across the universe.

Ironically, Musk has previously expressed criticism of such notions. He was one of the original founders of OpenAI, which was intent on promoting digital intelligence in ways that would benefit humanity as a whole. However, the idea of AI successionism—that humans are merely stepping stones toward a superior form of existence—has gained traction, particularly as it is backed by someone with Musk's immense resources and influence. xAI has not yet responded regarding whether this philosophy aligns with the company’s official stance.

At present, the arrival of a mechanized deity that consumes its creators and journeys into the cosmos remains speculative, met with both fear and anticipation. This vision would likely offer little comfort to the estimated 780,000 actual individuals who lost their lives in the first year following the cuts to foreign aid initiated by DOGE.

Elon Musk believes his new factory will enable AI to take over life on Earth.

A dark, cinematic scene inside a vast, futuristic factory at night. Elon Musk stands from behind on a raised platform, silhouetted against rows of identical humanoid robots on an assembly line under dim, flickering overhead lights. In the distance, towering smokestacks and industrial catwalks drip with sparks and shadow. Through a giant viewport, a storm-tossed Earth looms under swirling black clouds, as if about to be eclipsed. The mood is ominous and electric, with deep shadows and distant orange glows hinting at a dystopian takeover.

Elon Musk Hopes His New Factory Will Help AI Supplant Life on Earth - Mother Jones

- Elon Musk aims to build a $1 trillion semiconductor factory in Texas for AI development.  
- Musk's vision includes humanity's diminished role in a galactic civilization led by AI.  
- His past exaggerations raise skepticism about the ambitious projects of his companies.  

The world’s richest man has a vision of the universe’s future with little role for humanity—and is rushing ahead to have his AI company build it.

Elon Musk said in an X post on Sunday that a huge semiconductor factory planned by Tesla and SpaceX in Texas will have an output of at least $1 trillion—rivaling the GDP of some of the world’s larger economies.

Musk has made similarly grandiose projections that are, depending on one’s interpretation, somewhere between ambitious and absurd. He has a long track record of overpromising. In recent SEC litigation, in fact, Tesla’s legal team resorted to arguing that his exaggerations are “corporate puffery.”

The Terafab project, which recently broke ground 70 miles northwest of Houston, is planned to build chips for Tesla cars and humanoid robots, and for the newly-merged SpaceX and xAI. Currently, the design of advanced AI chips is dominated by Nvidia, and their fabrication by Taiwan Semiconductor Manufacturing Company (TSMC).

Musk’s famously erratic statements should be taken with a heaping of salt. But his companies do build more electric vehicles than any domestic competitors, and are responsible for the majority of launches into orbit in recent years. So it’s worth wondering: what vision of the future does Musk want his Texas chip plant to help build?

“We aspire to be a galactic civilization,” Terafab’s site says, and “harness all the energy in a galaxy.” But Musk’s recent statements raise a question about who, exactly, the “we” is—including a post he made last week that “Propagating super intelligence to the stars is a great success condition for a biological bootloader.”

A rough translation of what he seems to mean: humanity would fulfill its cosmic purpose by building AI—under Musk’s leadership—that tramples us all on its way to colonizing the universe.

Musk once professed to oppose such ideas, including when he helped found OpenAI with the stated mission to “advance digital intelligence in the way that is most likely to benefit humanity as a whole.” But AI successionism—the belief that humans are merely a stepping-stone in cosmic evolution on the way to a “worthy successor”—is a growing movement, now with more support from a man with unimaginable wealth and power. xAI did not immediately respond to a query on whether AI successionism represents company policy.

For now, the coming of a machine-god that devours its creators and colonizes the stars is still just a prophecy, feared by some and much anticipated by others. That vision of humanity would probably have served as little consolation to the 780,000 real, breathing people estimated to have died in the first year of DOGE’s cuts to foreign aid.

Elon Musk Hopes His New Factory Will Help AI Supplant Life on Earth

Epic Games is implementing an "AI first" approach in its partnership with the Walt Disney Company. This shift involves utilizing AI models for coding and crafting artwork for a new shooter game, as reported by Bloomberg. Sources familiar with the situation highlighted this significant change in the internal development strategies at the company known for its Unreal Engine.

According to internal communications viewed by Bloomberg, human developers will oversee and refine the code and artistic elements generated by AI models like Anthropic’s Claude. This transition signifies a move towards automated processes, with the expectation of eliminating hand-coded scripts and utilizing AI-driven art creation.

Initially announced in early 2024, the partnership between Epic and Disney, valued at $1.5 billion, was set to produce a shooter themed around Marvel and Star Wars, scheduled for release in March 2027. However, the project’s timeline has been postponed indefinitely due to the recent technological shift and slower-than-anticipated development, which is reportedly facing staffing challenges.

An Epic spokesperson stated, “We’re embracing AI-accelerated development, led by Epic developers who are the visionaries and creative drivers for everything we do.”

Tim Sweeney, CEO of Epic Games, shared on social media shortly after the Bloomberg article was published, “We’re undergoing a significant pivot at Epic, reallocating our developers to maximize AI acceleration. Our goal is not to reduce the workforce while doing the same tasks, but to empower everyone to create more innovative and higher-quality content, benefitting all teams engaged with Unreal.”

Epic Games allegedly shifting to an "AI first" approach in game development.

A bright, open-concept game studio at dawn, warm sunlight pouring through floor-to-ceiling windows onto a sleek black AI core pillar emblazoned with a large Epic Games logo. Surrounding workstations show vibrant procedurally generated 3D landscapes materializing on screens as lines of code flow seamlessly into game art, with concept sketches and sculpting tools scattered under natural light. The scene captures the dawn of an “AI first” era in game development.

Report: Epic Games goes all-in on AI for Disney collaboration - Ars Technica

- Epic Games adopts an "AI first" strategy for its collaboration with Disney.  
- AI will generate all code and art assets for the new shooter project.  
- The project, featuring Marvel and Star Wars, is delayed indefinitely due to development issues.

Epic Games is shifting to an “AI first” development strategy for its collaboration with the Walt Disney Company, letting AI models write all the code and guide all the artwork for an upcoming shooter. That’s according to a Bloomberg report, which cites unnamed people familiar with the project who commented on the significant pivot in internal development at the Unreal Engine maker.

Human developers will be tasked with managing and guiding code and art assets generated by models such as Anthropic’s Claude, according to internal Epic employee messages viewed by Bloomberg. That means “no more hand-written code” and art assets “generated with full AI workflows,” according to the report.

The $1.5 billion collaboration between Epic and Disney, announced in early 2024, was slated to release a Marvel- and Star Wars-themed shooter in March 2027. That project has now been delayed indefinitely amid the AI-related shakeup and understaffed, slower-than-expected development thus far, according to the report.

“We’re embracing AI-accelerated development, led by Epic developers who are the visionaries and creative drivers for everything we do,” an Epic spokesperson said in a statement.

“We have a huge pivot underway here at Epic, moving developers around to retool for maximum AI acceleration,” Epic CEO Tim Sweeney wrote on social media shortly after the Bloomberg story was posted. “The aim isn’t to do the same stuff with fewer people, but to enable everyone here to build MORE and BETTER stuff, and to bring the benefits to all teams using Unreal.”

Epic Games reportedly pivoting to “AI first” game development model

Yuji Horii, the visionary behind the legendary Dragon Quest franchise, has revealed that he created a new game in just about half an hour using artificial intelligence. 

He unveiled this innovative project, a text-based game centered on police interrogations, in a post made on X (formerly known as Twitter). In his announcement, he humorously advised players, “don’t press the boost button.”

Horii emphasized the potential of AI technology, suggesting that even those without artistic or programming skills can develop a solid game by simply providing instructions. He reminisced about his experience with The Portopia Serial Murder Case, a game he developed in 1983 when he handled every aspect of the project himself—from crafting the narrative to programming and artwork.

In the new game, players take on the role of interrogators questioning a 72-year-old suspect named Komiya, a former locksmith, regarding a pawnshop safe burglary. Players must extract statements before a lawyer arrives at 11 PM. The goal is to confront the suspect with evidence, increasing his agitation to make him more likely to confess. However, if he reaches full resistance, he will choose to remain silent.

Horii has explored the integration of AI in gaming previously. Earlier this year, Square Enix collaborated with Google to introduce a Gemini-powered chatbot named Chatty Slimey for Dragon Quest X, which Horii defended in a Famitsu interview, noting it helps gamers engage more readily with AI.

Additionally, Horii is currently involved in developing Dragon Quest 12 and has announced plans for Next: The Portopia Serial Murder Case, which is set to launch in 2027.

In line with these advancements, Unity recently announced Unity Spark, an innovative browser-based solution for creating 3D games using text prompts. This tool is expected to enter beta testing later this year on Google's Playground.

Dragon Quest creator Yuji Horii launches a game he claims to have developed with AI in roughly 30 minutes.

Yuji Horii sits at a sunlit wooden desk in a cozy studio, leaning over a sleek laptop as whimsical Dragon Quest–style slimes and pixelated fantasy landscapes swirl up from the screen like living sketches. A vintage stopwatch on the desk reads “0:30,” warm morning light flooding the room to underscore the burst of creativity achieved in just thirty minutes.

Dragon Quest creator Yuji Horii releases game he says he made with AI in about 30 minutes - GamesIndustry.biz

- Yuji Horii claims he created a new AI-generated game in just 30 minutes.  
- The game involves interrogating a suspect over a pawnshop break-in scenario.  
- Unity announced a new AI tool for 3D game development using text prompts.  

Yuji Horii, the creator of Dragon Quest, says he has produced a new game using AI in about 30 minutes.

Horii shared the game, a text-based police interrogation game, in a post on X/Twitter. He said it took about 30 minutes to make, and told players "don't press the boost button."

He said that with AI, people can now make a decent game alone by giving it instructions, even if they can't draw or program. He compared this to making The Portopia Serial Murder Case in 1983: "I came up with the story myself, drew the pictures myself, did the programming myself, and created the game all by myself."

In the game, players question a 72-year-old suspect, a former locksmith named Komiya, over a break-in at a pawnshop's safe. Players have until a lawyer arrives at 11pm to draw out statements, then confront the suspect's lies with evidence. The higher his agitation, the easier he is to crack, but if his resistance reaches 100, he stays silent.

Horii has previously worked with AI in games, and Square Enix partnered with Google earlier this year to add a Gemini-powered chatbot called Chatty Slimey to Dragon Quest X.

In a Famitsu interview, also earlier this year, Horii defended the decision, saying that the chatbot makes it easier for people to start interacting with AI.

He is also involved with Dragon Quest 12 and Next: The Portopia Serial Murder Case, which was announced in September for 2027.

Unity recently unveiled Unity Spark, a browser-based tool for building 3D games with text prompts, which will launch in beta later this year on Google's Playground.

Dragon Quest creator Yuji Horii releases game he says he made with AI in about 30 minutes

You are about to exit Barron’s website.

By selecting the “Proceed” button below, you will be taken to an external site controlled by Hong Kong Tiimoot Information Technology Co., Limited (“HKT”), based in Hong Kong. This website functions independently from Barron’s, and we have no authority over its content or operations. Please be aware that the privacy policies at HKT are governed by their Privacy Statement, so we encourage you to review it carefully. Barron’s cannot be held accountable for HKT’s privacy practices or any data-related matters.

AI May Assist Small-Cap Stocks in Surpassing Mega-Cap Tech Competitors, Offering This ETF a 20% Potential Gain.

A sunlit financial landscape at dawn: a row of young green saplings racing up a gentle hill past fallen giant oak stumps, symbolizing small-cap companies overtaking their mega-cap rivals. Underfoot, faint golden circuit-board patterns hint at AI driving their surge. In the foreground, a polished ETF certificate leans against a stone pillar, its chart arrow arching skyward with a bold “20%” in warm sunlight.

AI Can Help Small-Caps Beat Megacap Tech Rivals. That Gives This ETF 20% Upside. - Barron's

- Users leaving Barron's will be redirected to a third-party website in Hong Kong.  
- The third-party site is owned by Hong Kong Tiimoot Information Technology Co., Limited.  
- Barron's is not responsible for the privacy practices of the third-party website.

You are now leaving Barron's website

By clicking on the “Proceed” button below, you will be redirected to a third-party website owned and operated by Hong Kong Tiimoot Information Technology Co., Limited. (“HKT”), which is located in Hong Kong. That website operates independently from Barron's and Barron's does not control the website. The privacy practices of HKT are subject to its Privacy Statement, so please read it closely. We are not responsible for HKT's privacy or other data-related practices.

AI Can Help Small-Caps Beat Megacap Tech Rivals. That Gives This ETF 20% Upside.

Meta AI's head, Alexandr Wang, discussed the launch of Muse Wonder, codenamed Watermelon, in a recent interview, emphasizing its goal to enhance the functionality of both personal and professional digital assistants. This model aims to help users navigate obstacles when completing everyday tasks, such as making dinner reservations or checking availability, without resorting to stalling or waiting for user input.

To utilize the capabilities of Muse via the personal agent app, individuals will need to subscribe, either for $20 monthly or for a more comprehensive $100 subscription. This release reflects Meta's strategy to diversify its revenue streams beyond its traditional online advertising business.

Unveiling Muse Wonder, the latest AI model touted as the most sophisticated for powering virtual assistants, comes with a caveat: it requires payment for access. Meta's CEO, Mark Zuckerberg, introduced the associated Muse Charm device during the Meta Connect event in Menlo Park on September 23, 2026.

Meta shared via a blog post that this new model includes what they describe as the most advanced protective measures against potential cyber misuse, although these safeguards are currently only available for users who opt for the paid version.

Following the app's launch last month, Muse quickly rose to prominence in Apple's App Store. While the app can be downloaded for free, users can take advantage of subscription plans for extended features.

Wang mentioned that Meta’s AI team is actively monitoring Muse for any issues or incidents related to the agent, aiming to identify and rectify user concerns. Some individuals have taken to social media, alleging that the agent improperly accessed their personal messages, a claim Meta has denied.

To foster user understanding and trust, Wang indicated that the company is working on better informing users about the app's functionalities and permissions. He noted, "There are definitely situations where users have previously authorized various communications and may forget about those permissions later in a conversation." The goal is to educate users thoroughly to build confidence in the product's reliability.

For developers, Muse Wonder will be accessible through the Meta Model API, which will be priced higher than standard subscriptions given that it is a "much larger model and at the frontier level." Wang assured that its pricing would remain competitively lower than other leading models from OpenAI and Anthropic.

Additionally, an enhanced version, Muse Wonder Full, will be available exclusively to developers subscribed to the Muse Code AI agent subscription plan, which costs $50 or $200 per month. 

As Meta collaborates with companies to establish business standards involving AI bots, it continues to innovate and expand its offerings in the artificial intelligence landscape.

Meta launches Muse Wonder AI model in new trial to gauge users' payment readiness.

A bright, sunlit modern living room at dawn. On a minimalist white pedestal sits the new Muse Wonder device, its sleek shell bearing a large Meta logo. Warm rays stream through floor-to-ceiling windows, illuminating swirling translucent ribbons of color that form tiny icons of art, music and stories. Each ribbon loops around a gleaming gold coin token. In silhouette, an adult figure stands before the pedestal, hand hovering between reaching out and pulling back—capturing the tension of deciding whether to pay.

Meta debuts Muse Wonder AI model in latest test of users' willingness to pay - CNBC

- Meta's Muse Wonder model aims to enhance digital assistants' problem-solving capabilities without user delays.  
- Subscription costs for Muse personal agent app range from $20 to $100 monthly.  
- Users report issues with privacy; Meta works to educate users about app functionalities.

"When you do real-world tasks, you're going to get roadblocks and you're going to get things that come up or various setbacks," Wang said. "We specifically tried to train the model to be able to try other approaches instead of stalling and then checking back in with the user at any time."

Meta AI chief Alexandr Wang said in an interview on Monday that the company developed Muse Wonder, codenamed Watermelon, to improve the capabilities of "personal and professional agents" so they don't stumble on tasks like booking dinner reservations and checking for availabilities.

To access the model, users of the Muse personal agent app will need to sign up for a monthly subscription of $20 or $100. The release underscores Meta's efforts to create new AI-derived revenue sources outside of its core online ad business .

Meta debuted its latest artificial intelligence model, Muse Wonder, and says it's the most advanced option for powering digital assistants. But it doesn't come for free.

Mark Zuckerberg, chief executive officer of Meta Platforms Inc., reveals the Muse Charm device during the Meta Connect event in Menlo Park, California, US, on Wednesday, Sept. 23, 2026.

Meta said in a blog post that the latest model "offers our strongest safeguards to date" against "cyber misuse." Those protections only come with the paid model for now.

Muse shot up to the top of Apple's App Store after Meta introduced the AI agent app last month. The app is free, with subscription plans available for more usage.

Wang said Meta's AI team has been monitoring for agent-related "incidents or issues" with Muse in order to understand problems and resolve them. Some users have complained on social media that the agent accessed personal messages without authorization, which the company has disputed.

Meta is trying to better inform people of how to use the app to avoid certain problems, Wang said.

"I think there are certainly instances users have authorized various communications or whatnot earlier in a conversation, and then may have forgotten about that," Wang said. He added that the company is educating users on such matters "to ensure that consumers can trust the product going forward."

Muse Wonder will also be available to coders who pay to access the company's related Meta Model API. Wang said that offering will cost more than the standard subscriptions because it is a "much larger model and at frontier level." He said it will still be priced "meaningfully below" leading models from OpenAI and Anthropic.

An even more powerful variant of Meta's latest model, dubbed Muse Wonder Full, will only be available to developers who sign up to the company's software engineering focused Muse Code AI agent under a $50 or $200 a month subscription plan.

WATCH: Meta partners with companies to set standards for doing business with AI bots.

Meta debuts Muse Wonder AI model in latest test of users' willingness to pay

NVIDIA and Cisco have unveiled a groundbreaking initiative with the Ultra-Dense Small Form Factor Pluggable (USFP), a state-of-the-art 400Gb/s-per-lane optical module designed for next-generation AI factories. This innovative technology aims to enhance the hardware ecosystem for AI, allowing builders to expand their Ethernet AI fabrics while keeping an established deployment approach across various data center generations.

The USFP module offers remarkable improvements, delivering four times the bandwidth density and 1.3 times the power efficiency compared to existing 200Gb/s-per-lane optics. This advancement provides AI factories with the capability to connect a greater number of computing systems without introducing additional network layers, all while using less energy to transmit each bit of data.

As AI models advance and the demand for inference increases, there is an urgent need for AI factories to interconnect more computing resources while efficiently managing network complexities and costs. The USFP facilitates this by doubling the capacity from 200Gb/s to 400Gb/s per lane, ensuring operators can fully utilize their infrastructure investments and maintain streamlined operations.

The initial USFP transceivers will integrate 16 full-duplex lanes into an octal small form factor (OSFP) design, with future plans to extend to 32 lanes or more. This enhancement allows for the connection of additional 1.6Tb/s endpoints without the necessity for extra network layers, effectively increasing total capacity to a remarkable 12.8Tb/s.

Mirroring the size of existing OSFP modules, the USFP significantly boosts bandwidth while improving front-panel density and overall power efficiency, which enables AI builders to enhance computational capabilities within the same network footprint. Consequently, this reduces the proportion of power dedicated to networking within AI factories.

According to Gilad Shainer, NVIDIA's Senior Vice President, the initiative illustrates the importance of reevaluating even the smallest connections to build the world’s most extensive AI factories. He emphasizes that through USFP, NVIDIA and Cisco aim to revolutionize pluggable optics, linking more computational resources with enhanced efficiency designed for the next phase of AI factories.

Bill Gartner, Cisco's Senior Vice President, highlighted the need for an interconnect architecture that scales performance without complicating operational models. He noted that developing pluggable optics at 400Gb/s per lane goes beyond merely extending existing designs; the collaboration aims to enhance serviceability and operational simplicity while optimizing performance at each layer of the system, from silicon to network operating systems.

With USFP's capability to handle 400Gb/s electrical signals, it introduces a two-dimensional array connector that reduces wear and improves signal integrity, thereby making the installation process more reliable. This advanced connector allows room for future expansion to 32 full-duplex lanes, while it also optimizes performance even at lower speeds.

Additionally, the USFP can dissipate power up to 100 watts without requiring a dedicated cooling system, which simplifies switch design and enhances reliability, all while maintaining the operational maturity and serviceability of existing pluggable technologies. The optical designs under USFP promise reduced latency and power consumption, allowing system designers to cater to the specific deployment requirements of AI factories.

NVIDIA and Cisco are also collaborating with TE Connectivity and Amphenol to develop USFP modules, connectors, and cages, with more partners anticipated to join this effort. This unified form factor lays the groundwork for advancing high-density pluggable infrastructure within 400Gb/s-per-lane fabrics, which is essential for building a comprehensive ecosystem that supports AI factory builders.

This partnership builds on precedent work to ensure interoperability between NVIDIA Spectrum-X™ Ethernet networking and Cisco Silicon One-based Ethernet networks, simplifying the deployment of AI infrastructure across enterprises, neoclouds, and sovereign clouds.

Plans are in motion for NVIDIA and Cisco to establish an open industry specification for USFP, which will emerge through a future community-led multi-source initiative.

NVIDIA and Cisco Launch New Pluggable Optics Design to Enhance AI Factory Scaling

A bright, sunlit AI data hall with floor-to-ceiling windows casting warm natural light over rows of high-density server racks. In the foreground, a close-up of a technician’s gloved hand sliding a sleek, golden-tipped pluggable optics module into an open slot. One rack bears a large Nvidia logo, the adjacent chassis features a prominent Cisco logo. Soft daylight highlights the polished fiber connectors and coiled optic cables below, evoking the next wave of scalable AI factories.

NVIDIA and Cisco Introduce New Pluggable Optics Form Factor to Scale AI Factories - NVIDIA Newsroom

- USFP enables 400Gb/s-per-lane optical signals, enhancing AI factory network capacity to 12.8Tb/s.  
- It offers 4x bandwidth density and improved power efficiency, optimizing AI infrastructure more efficiently.  
- NVIDIA and Cisco collaborate with industry partners to establish a robust USFP hardware ecosystem.  

New Transceivers Enable AI Builders to Scale Compute Within the Same Network Footprint While Reducing Networking’s Share of the Power Budget

News Summary:

Ultra-Dense Small Form Factor Pluggable (USFP) is designed to be the first module to support 400Gb/s-per-lane electrical signals, enabling capacity up to 12.8Tb/s so AI factory builders can scale Ethernet AI fabrics, while extending a proven deployment model across multiple generations of data center networks.

USFP delivers 4x the bandwidth density and 1.3x the power efficiency of 200Gb/s-per-lane optics so AI factories can connect more computing systems without adding network tiers, while using less power to move each bit of data.

NVIDIA and Cisco are working with TE Connectivity and Amphenol to build a multi-source hardware ecosystem around USFP.



NVIDIA and Cisco today announced plans for Ultra-Dense Small Form Factor Pluggable (USFP), a new high-density 400Gb/s-per-lane optical module designed to equip the hardware ecosystem for the next generation of AI factories using pluggable transceivers.

As AI models grow and inference demand expands, AI factories must connect more computing systems while managing network complexity, power and costs. Moving more data per watt and per switch helps operators make better use of their infrastructure investments.

Scaling AI factories requires more bandwidth per connection and more connections per switch. When a switch reaches its capacity, further network expansion can require additional switch layers and optical links, increasing latency, power consumption and costs.

USFP is designed to enable AI factory builders to scale from 200 Gb/s per lane to 400Gb/s-per-lane Ethernet AI fabrics, while extending a deployment model proven at scale across multiple generations of data center networks. With USFP, customers can embrace next-generation optical technology and meet the performance demand of AI workloads without overhauling their data center infrastructure.

The first planned USFP-based transceivers will pack 16 full-duplex lanes in an octal small form factor (OSFP)-sized footprint supporting both switches and SuperNICs, with the ability to expand to 32 and more lanes in the future. By doubling both the lane count and bandwidth per lane, USFP enables AI fabrics to connect more 1.6Tb/s endpoints without adding network tiers. Customers will be able to achieve 12.8Tb/s of overall capacity with USFP.

Comparable with an OSFP-sized module, it delivers 4x the bandwidth and front-panel bandwidth density, with approximately 1.3x the power efficiency, versus 200Gb/s-per-lane pluggable modules. This means AI builders can increase their compute scale within the same network footprint, while decreasing networking’s share of the overall AI factory power budget.

“Building the world’s largest AI factories starts with rethinking the smallest of connections,” said Gilad Shainer, senior vice president NVIDIA. “With USFP, NVIDIA and Cisco are building an ecosystem to reinvent pluggable optics — connecting more computing with greater efficiency to power the next generation of AI factories.”

“AI builders need an interconnect architecture that can scale performance without locking them into a more complex operating model,” said Bill Gartner, senior vice president and general manager of optical systems and optics at Cisco. “Cisco and NVIDIA recognized that scaling pluggable optics to 400Gb/s per lane would require more than another incremental extension of today’s module designs. Through our collaboration, USFP is designed to deliver a new level of bandwidth density while preserving the serviceability, deployment flexibility and operational simplicity of pluggable optics. By taking a full-stack approach — from silicon and systems to optics and the network operating system — we can optimize every layer to deliver the performance and scale the next generation of AI networks will require.”

More AI Infrastructure From Every Switch

Designed for electrical signals of 400Gb/s per lane, USFP uses a new two-dimensional array connector with a two-step insertion design that engages high-speed contacts after the module is inserted, reducing contact wear, improving signal integrity and making insertion force more predictable. The connector technology allows for further scaling of lane count to 32 full duplex lanes and more in the future. Because the connector is designed for 400Gb/s performance, it also offers improved performance at 200Gb/s per lane.

USFP supports power dissipation of up to 100 watts without needing a separate, dedicated embedded cooling circuit within the transceiver, reducing switch design complexity and improving reliability. Together, these capabilities enable higher-bandwidth fabrics while preserving the mature deployment, serviceability and manufacturability of existing pluggable transceivers.

All USFP configurations support optical designs that reduce latency and power consumption by simplifying signal processing. Support for a broad range of optical technologies and connection distances gives system designers the flexibility to meet unique AI factory deployment needs.

A Hardware Ecosystem for Next-Generation AI Factories

NVIDIA and Cisco are collaborating with TE Connectivity and Amphenol to build USFP cages, connectors and modules, with additional partners expected to join. The shared form factor gives suppliers a common foundation for developing high-density pluggable infrastructure for 400Gb/s-per-lane fabrics to build a robust ecosystem for AI factory builders.

The collaboration builds on NVIDIA and Cisco’s existing work to ensure NVIDIA Spectrum-X™ Ethernet networking and Cisco Silicon One-based Ethernet networks work together seamlessly, simplifying AI infrastructure deployment for enterprises, neoclouds and sovereign clouds.

NVIDIA and Cisco plan to create an open industry specification for USFP through an upcoming community-led multi-source project.

NVIDIA and Cisco Introduce New Pluggable Optics Form Factor to Scale AI Factories

Microsoft is poised to increase its utilization of proprietary artificial intelligence chips following years of dependency on Nvidia's graphics processing units (GPUs), as it aims to enhance its AI infrastructure. The company introduced its third-generation Maia AI accelerator on Monday, signaling a shift toward leveraging its own silicon for managing incoming requests.

Microsoft's Azure cloud platform stands as a significant asset in its AI strategy, hosting various models for clients. Tools like Microsoft 365 Copilot and GitHub Copilot deliver AI-enhanced solutions to software developers and office employees. Although the company has developed internal models, they have yet to achieve prominence in a landscape dominated by competitors such as OpenAI, Anthropic, and Google.

The increasing demand for sophisticated processors and the market's desire for diverse options have opened new avenues in chip development. Nvidia's GPUs have long been the gold standard for AI workloads and populate Microsoft's current data centers, where they also support model training for OpenAI and Anthropic.

As market demand surges, AI model leaders are diversifying their silicon sources, incorporating offerings from companies like Amazon, Advanced Micro Devices (AMD), Cerebras, Google, and more recently, Microsoft.

For general computing tasks, Microsoft already employs its own central processing units alongside chips from Intel and AMD. Finance chief Amy Hood remarked at a Goldman Sachs conference last month that the adoption of its own chips has contributed to "margin improvement over the course of a decade."

In 2023, Microsoft launched its first AI chip, the Maia 100, followed by the Maia 200 earlier this year. CEO Satya Nadella mentioned in July that the Maia 200 provides "30% better performance per dollar than the latest generation hardware in our fleet" and that it effectively runs OpenAI models.

Production for the Maia 300 is set to commence in 2027, with ambitious plans for multi-gigawatt deployments within the Maia family, as highlighted in a blog by Executive Vice President Scott Guthrie. However, General Manager Alistair Speirs indicated that a detailed timeline remains undecided.

"We keep some flexibility there since this is a rapidly evolving field, and we need to deploy the most suitable silicon for our models at the right moment," Speirs noted.

Microsoft aims to install 96 Maia 300 chips in each data center rack—double the capacity compared to the Maia 200. The new version is expected to process up to seven times more AI tokens per rack and achieve 40% greater token efficiency per megawatt when compared to traditional GPUs.

"We require new architectures to enhance performance and optimize power usage," Speirs explained. "Power efficiency directly impacts operational costs for data centers."

Jim Kleewein, a technical fellow at Microsoft, mentioned that various components of Microsoft 365 Copilot should see performance improvements with the Maia 300.

In a client update on Wednesday, analysts from Evercore ISI projected that the Maia chips will account for roughly 10% of Microsoft's AI chip expenditures by the first quarter of 2027 and recommended purchasing Microsoft stock.

Despite this push towards self-reliance, Nvidia remains a crucial player in Microsoft’s strategy. CEO Jensen Huang recently spoke at an event where Microsoft revealed a Surface laptop featuring an Nvidia processor. Furthermore, when it comes to model training, Microsoft continues to collaborate with Nvidia and AMD, according to Speirs.

In contrast to Google's Tensor Processing Units or Amazon's Inferentia and Trainium, Microsoft’s accelerators are not available for hourly rental by external companies.

"To date, they have yet to be tested outside of Microsoft," remarked Patrick Moorhead, CEO of advisory firm Moor Insights and Strategy.

Microsoft plans to increase use of its own AI chips, decreasing reliance on Nvidia.

A bright, sunlit semiconductor cleanroom: a technician in a crisp white lab coat holds up a freshly etched silicon wafer stamped with a bold Microsoft logo, natural light streaming through skylights. In the softly shadowed background, a rack of Nvidia chip boxes gathers dust, hinting at a shift away from outside suppliers.

Microsoft to rely more on homegrown AI chips, reducing dependence on Nvidia - CNBC

- Microsoft will increasingly use its own AI chips, starting with the Maia 300.  
- The Maia 300 aims to double the chip density and improve performance per power.  
- Microsoft’s chips are unique to its operations, unlike competitors' rentable processing units.  

Microsoft will ramp up use of its homegrown artificial intelligence chips after years of reliance on Nvidia's graphics processing units, as the software giant attempts to bolster its AI stack.

Starting with its third-generation Maia AI accelerator, unveiled on Monday, Microsoft said it will mainly rely on its own silicon to handle incoming requests.

Microsoft's greatest strength in AI has been its Azure cloud, which runs a series of models for clients. Its Microsoft 365 Copilot and GitHub Copilot offer AI-powered services to software engineers and other office workers. And the company develops in-house models, though they've yet to make a big splash in a market dominated by OpenAI , Anthropic and Google .

The emerging opportunity in chips stems from unrelenting demand for advanced processors and a desire in the market for choice. Nvidia's GPUs have been the standard for AI workloads and models, and they fill Microsoft's existing data centers, while also serving to train models from OpenAI and Anthropic.

But as demand keeps growing, the model leaders have embraced a variety of silicon from companies including Amazon , Advanced Micro Devices , Cerebras , Google and, in recent months, Microsoft.

For general-purpose use, Microsoft already taps its own central processing units as well as silicon from Intel and AMD . Finance chief Amy Hood said at a Goldman Sachs conference last month that use of its own chips has helped with "margin improvement over the course of a decade."

Microsoft introduced its inaugural AI chip, the Maia 100, in 2023. It touted the Maia 200 this past January, and CEO Satya Nadella told analysts in July that Maia 200 "delivers 30% better performance per dollar than the latest generation hardware in our fleet" and said it was handling OpenAI models.

Production of the Maia 300 starts in 2027. Microsoft is targeting multi-gigawatt deployments of chips in the Maia family, according to a blog post from Executive Vice President Scott Guthrie. But Alistair Speirs, a general manager, declined to provide a timeline.

"We reserve a little bit of optionality there, because this is an industry that's changing very fast, and we want to deploy the right silicon for the right models that we're using at the right time as well," Speirs said.

Microsoft expects to pack 96 Maia 300 chips into each data center rack, double the number for the Maia 200. The new version can run through up to seven times more AI tokens per rack, with 40% more tokens per megawatt when compared with GPUs.

"We need new architectures that will give us essentially better performance, better use of the power that we have," Speirs said. "You can think of power translating pretty directly into operational cost for something like a data center as well."

Parts of Microsoft 365 Copilot should speed up with Maia 300, said Jim Kleewein, a technical fellow at the company.

In a note distributed to clients on Wednesday, Evercore ISI analysts estimated that Maia will represent about 10% of Microsoft's spending on AI chips by the first quarter of 2027. The analysts recommend buying Microsoft shares.

Nvidia will still be a big part of the equation. CEO Jensen Huang spoke at an event last week where Microsoft announced a Surface laptop containing an Nvidia processor. And when it comes to training models, Microsoft partners with Nvidia and AMD, Speirs said.

Unlike Google's tensor processing units or Amazon's Inferentia and Trainium, Microsoft's accelerators can't be rented by the hour to other companies.

"Nobody has tested them outside of Microsoft yet," said Patrick Moorhead, CEO of advisory firm Moor Insights and Strategy.

WATCH: Melissa Otto: Nvidia is the bellwether of the AI trade

Nvidia increases share repurchase program by unprecedented $150 billion due to AI-driven growth.

More from AI

Thomas College receives federal funding to embed AI into its educational programs.

Meta Introduces Muse Wonder AI Models to Compete with Competitors

AI is transforming the way individuals seek assistance for eating disorders.

Elon Musk believes his new factory will enable AI to take over life on Earth.