Intel begins mass production using world's first High-NA EUV
Intel has moved its 18A node into high-volume production at Fab 52 in Arizona, using ASML’s first commercially deployed High-NA EUV tools to ramp advanced chipmaking in the U.S.
The High-NA (0.55 NA) systems enable ~8nm single-exposure patterning that can reduce multi-patterning complexity, while 18A also adds RibbonFET gate-all-around transistors and PowerVia backside power delivery aimed at boosting performance-per-watt.
External-customer momentum remains a key question after Reuters reported Nvidia paused 18A testing on Dec. 24, even as Intel points to interest from partners like Microsoft and the broader industry watches whether 18A can win foundry share from TSMC
.
Intel has moved its 18A node into high-volume production at Fab 52 in Arizona, using ASML’s first commercially deployed High-NA EUV tools to ramp advanced chipmaking in the U.S.
The High-NA (0.55 NA) systems enable ~8nm single-exposure patterning that can reduce multi-patterning complexity, while 18A also adds RibbonFET gate-all-around transistors and PowerVia backside power delivery aimed at boosting performance-per-watt.
External-customer momentum remains a key question after Reuters reported Nvidia paused 18A testing on Dec. 24, even as Intel points to interest from partners like Microsoft and the broader industry watches whether 18A can win foundry share from TSMC
.
xAI launches Grok Business and Enterprise for corporate market
xAI rolled out Grok Business and Grok Enterprise on December 30, 2025, offering organizational AI with enterprise-grade security and a guarantee that customer data will not be used for model training.
Grok Business is priced at $30 per seat per month for small-to-medium teams through self-serve onboarding, while Enterprise pricing is available upon request and includes Custom Single Sign-On, Directory Sync, and SOC 2 compliance.
The launch positions xAI to compete with OpenAI's ChatGPT Enterprise, which reportedly costs around $60 per user per month, and Anthropic's Claude Team at $30 per seat monthly.
Read detailed
xAI rolled out Grok Business and Grok Enterprise on December 30, 2025, offering organizational AI with enterprise-grade security and a guarantee that customer data will not be used for model training.
Grok Business is priced at $30 per seat per month for small-to-medium teams through self-serve onboarding, while Enterprise pricing is available upon request and includes Custom Single Sign-On, Directory Sync, and SOC 2 compliance.
The launch positions xAI to compete with OpenAI's ChatGPT Enterprise, which reportedly costs around $60 per user per month, and Anthropic's Claude Team at $30 per seat monthly.
Read detailed
U.S. grants TSMC annual license for chip gear exports to China
The U.S. Department of Commerce issued an annual export license to Taiwan Semiconductor Manufacturing allowing the chipmaker to import American manufacturing equipment to its Nanjing, China facility without requiring individual vendor licenses, the company announced Thursday.
The approval replaces a validated end-user waiver system that expired December 31, shifting from blanket permissions to annual licensing requirements that Washington also applied to Samsung and SK Hynix for their Chinese operations.
TSMC's Nanjing facility accounts for approximately 3% of the company's total capacity and produces older-generation chips using 16-nanometer, 12-nanometer, and 28-nanometer technologies, with the new license ensuring uninterrupted fabrication operations.
Read detailed
The U.S. Department of Commerce issued an annual export license to Taiwan Semiconductor Manufacturing allowing the chipmaker to import American manufacturing equipment to its Nanjing, China facility without requiring individual vendor licenses, the company announced Thursday.
The approval replaces a validated end-user waiver system that expired December 31, shifting from blanket permissions to annual licensing requirements that Washington also applied to Samsung and SK Hynix for their Chinese operations.
TSMC's Nanjing facility accounts for approximately 3% of the company's total capacity and produces older-generation chips using 16-nanometer, 12-nanometer, and 28-nanometer technologies, with the new license ensuring uninterrupted fabrication operations.
Read detailed
Meta hid scam ads from regulators instead of stopping them
Meta developed a "global playbook" to manage regulatory pressure over scam ads on Facebook and Instagram, using its Ad Library to make fraudulent content "untraceable" for regulators rather than implementing universal advertiser verification, according to internal documents reviewed by Reuters.
The company declined to spend an estimated $2 billion on universal verification—which could reduce scam ads by up to 29%—because it might cost up to 4.8% of revenue, instead adopting a "reactive only" approach that redirects blocked scam ads from regulated markets to unregulated ones.
The U.S. Virgin Islands filed a lawsuit against Meta on December 30 alleging the company "knowingly and intentionally" exposed users to fraud while profiting from scams, as U.S. senators called for federal investigation and the European Commission questioned Meta's compliance.
Read detailed
Meta developed a "global playbook" to manage regulatory pressure over scam ads on Facebook and Instagram, using its Ad Library to make fraudulent content "untraceable" for regulators rather than implementing universal advertiser verification, according to internal documents reviewed by Reuters.
The company declined to spend an estimated $2 billion on universal verification—which could reduce scam ads by up to 29%—because it might cost up to 4.8% of revenue, instead adopting a "reactive only" approach that redirects blocked scam ads from regulated markets to unregulated ones.
The U.S. Virgin Islands filed a lawsuit against Meta on December 30 alleging the company "knowingly and intentionally" exposed users to fraud while profiting from scams, as U.S. senators called for federal investigation and the European Commission questioned Meta's compliance.
Read detailed
Neuralink plans high-volume brain interface production by 2026
Elon Musk announced that Neuralink will begin high-volume production of brain-computer interface devices and transition to a streamlined, almost entirely automated surgical procedure in 2026.
As of September 2025, 12 people worldwide have received Neuralink's N1 implant, collectively using their devices for over 15,000 hours to control computers and smartphones with their thoughts.
The company completed two implant surgeries in a single day in July and is expanding clinical trials to Canada, the UK, and the United Arab Emirates as it works toward automating the implant process.
Read detailed
Elon Musk announced that Neuralink will begin high-volume production of brain-computer interface devices and transition to a streamlined, almost entirely automated surgical procedure in 2026.
As of September 2025, 12 people worldwide have received Neuralink's N1 implant, collectively using their devices for over 15,000 hours to control computers and smartphones with their thoughts.
The company completed two implant surgeries in a single day in July and is expanding clinical trials to Canada, the UK, and the United Arab Emirates as it works toward automating the implant process.
Read detailed
Tesla Semi achieves 1.2 MW charging in new production design
Tesla released video showing its redesigned Semi truck charging at a record 1.2 megawatts, enabling an estimated 10-80% charge in under 45 minutes for the electric truck's ~850 kWh battery pack.
The breakthrough addresses the trucking industry's primary concern about electric vehicle downtime, allowing the Semi to recover significant range during mandatory 30-minute driver breaks while diesel trucks refuel in 15 minutes.
Tesla is targeting production to begin in the first half of 2026 at its Nevada factory with a planned capacity of up to 50,000 units annually, while deploying 46 Megacharger stations expected by early 2027.
Read detailed
Tesla released video showing its redesigned Semi truck charging at a record 1.2 megawatts, enabling an estimated 10-80% charge in under 45 minutes for the electric truck's ~850 kWh battery pack.
The breakthrough addresses the trucking industry's primary concern about electric vehicle downtime, allowing the Semi to recover significant range during mandatory 30-minute driver breaks while diesel trucks refuel in 15 minutes.
Tesla is targeting production to begin in the first half of 2026 at its Nevada factory with a planned capacity of up to 50,000 units annually, while deploying 46 Megacharger stations expected by early 2027.
Read detailed
Apple slashes Vision Pro production after disappointing sales
Apple has halted production of its Vision Pro headset and slashed digital advertising spending by over 95% in 2025 after the $3,499 device shipped only 390,000 units globally in 2024, according to International Data Corporation and Sensor Tower.
IDC projects just 45,000 Vision Pro units will ship in Q4 2025, a stark contrast to the millions of iPhones and iPads Apple sells quarterly, with analysts citing high costs, comfort issues, and only 3,000 available apps as key barriers to adoption.
Despite the setback, Apple plans to release a cheaper Vision headset this year while the device has gained limited traction in enterprise applications like pilot training and surgical procedures.
Source
Apple has halted production of its Vision Pro headset and slashed digital advertising spending by over 95% in 2025 after the $3,499 device shipped only 390,000 units globally in 2024, according to International Data Corporation and Sensor Tower.
IDC projects just 45,000 Vision Pro units will ship in Q4 2025, a stark contrast to the millions of iPhones and iPads Apple sells quarterly, with analysts citing high costs, comfort issues, and only 3,000 available apps as key barriers to adoption.
Despite the setback, Apple plans to release a cheaper Vision headset this year while the device has gained limited traction in enterprise applications like pilot training and surgical procedures.
Source
OpenAI developing AI-powered pen for 2026 launch, leak claims
OpenAI is developing an AI-powered pen-style device codenamed "Gumdrop" in collaboration with former Apple designer Jony Ive, with the gadget expected to transcribe handwritten notes to ChatGPT and enable voice communication, according to industry tipster Smart Pikachu.
The device will be manufactured by Foxconn in Vietnam after a dispute over manufacturing location ended OpenAI's initial partnership with Chinese manufacturer Luxshare, with OpenAI reportedly refusing to produce the hardware in China.
OpenAI acquired Ive's hardware startup for approximately $6.4 billion in May 2025, with CEO Sam Altman stating in November 2025 that prototypes are "jaw-droppingly good" and the device should launch within two years, aiming to offer a calmer alternative to smartphones.
Source
OpenAI is developing an AI-powered pen-style device codenamed "Gumdrop" in collaboration with former Apple designer Jony Ive, with the gadget expected to transcribe handwritten notes to ChatGPT and enable voice communication, according to industry tipster Smart Pikachu.
The device will be manufactured by Foxconn in Vietnam after a dispute over manufacturing location ended OpenAI's initial partnership with Chinese manufacturer Luxshare, with OpenAI reportedly refusing to produce the hardware in China.
OpenAI acquired Ive's hardware startup for approximately $6.4 billion in May 2025, with CEO Sam Altman stating in November 2025 that prototypes are "jaw-droppingly good" and the device should launch within two years, aiming to offer a calmer alternative to smartphones.
Source
DeepSeek unveils architecture to cut AI training costs
DeepSeek released a research paper on Thursday introducing Manifold-Constrained Hyper-Connections (mHC), a novel architecture designed to improve training stability and scalability for large AI models while minimizing computational costs, with CEO Liang Wenfeng listed as co-author.
The mHC framework builds on ByteDance's 2024 hyper-connection architecture by adding a manifold constraint to restore identity mapping properties and reduce memory overhead, with testing on models up to 27 billion parameters showing stable performance without added computational cost.
Industry observers expect DeepSeek to launch a new model before Spring Festival in mid-February 2026, as Liang's personal publication of major technical papers has historically signaled upcoming product releases, including last year's R1 model.
Source
DeepSeek released a research paper on Thursday introducing Manifold-Constrained Hyper-Connections (mHC), a novel architecture designed to improve training stability and scalability for large AI models while minimizing computational costs, with CEO Liang Wenfeng listed as co-author.
The mHC framework builds on ByteDance's 2024 hyper-connection architecture by adding a manifold constraint to restore identity mapping properties and reduce memory overhead, with testing on models up to 27 billion parameters showing stable performance without added computational cost.
Industry observers expect DeepSeek to launch a new model before Spring Festival in mid-February 2026, as Liang's personal publication of major technical papers has historically signaled upcoming product releases, including last year's R1 model.
Source
Global chip sales projected to hit $1 trillion in 2026
The global semiconductor industry is projected to reach approximately $1 trillion in 2026, nearly five years ahead of earlier forecasts, with Bank of America analyst Vivek Arya predicting 30% growth while the World Semiconductor Trade Statistics forecasts $975.5 billion in sales.
The surge is driven by an "AI supercycle" with Nvidia commanding 80% of the AI accelerator market through chips priced at $30,000-$40,000 per unit, while tech giants commit over $400 billion in data center infrastructure spending.
Logic and memory segments are expected to grow over 30% year-over-year, with high-bandwidth memory makers like Micron Technology
reporting sold-out capacity through 2026, though the industry faces power constraints and component shortages.
Source
The global semiconductor industry is projected to reach approximately $1 trillion in 2026, nearly five years ahead of earlier forecasts, with Bank of America analyst Vivek Arya predicting 30% growth while the World Semiconductor Trade Statistics forecasts $975.5 billion in sales.
The surge is driven by an "AI supercycle" with Nvidia commanding 80% of the AI accelerator market through chips priced at $30,000-$40,000 per unit, while tech giants commit over $400 billion in data center infrastructure spending.
Logic and memory segments are expected to grow over 30% year-over-year, with high-bandwidth memory makers like Micron Technology
reporting sold-out capacity through 2026, though the industry faces power constraints and component shortages.
Source
OpenAI consolidates teams to overhaul audio AI models
OpenAI has unified multiple engineering, product, and research teams over the past two months to overhaul its audio models ahead of an audio-first personal device expected to launch in about a year.
The company's new audio model, slated for early 2026, will reportedly sound more natural, handle interruptions like an actual conversation partner, and even speak while users are talking—capabilities current models lack.
The hardware push follows OpenAI's $6.5 billion acquisition in May of io, the startup co-founded by former Apple design chief Jony Ive, who has prioritized reducing device addiction through audio-first design.
Source
OpenAI has unified multiple engineering, product, and research teams over the past two months to overhaul its audio models ahead of an audio-first personal device expected to launch in about a year.
The company's new audio model, slated for early 2026, will reportedly sound more natural, handle interruptions like an actual conversation partner, and even speak while users are talking—capabilities current models lack.
The hardware push follows OpenAI's $6.5 billion acquisition in May of io, the startup co-founded by former Apple design chief Jony Ive, who has prioritized reducing device addiction through audio-first design.
Source
SpaceX to lower 4,400 Starlink satellites to boost safety
SpaceX will lower approximately 4,400 Starlink satellites from 550 kilometers to 480 kilometers altitude throughout 2026, according to Michael Nicolls, the company's vice president of Starlink engineering, in a move coordinated with other operators, regulators, and U.S. Space Command.
The orbital adjustment will reduce ballistic decay time by over 80% during solar minimum—shortening the period from more than four years to just months—and decrease collision risks since debris objects and planned constellations are significantly lower below 500 kilometers.
The announcement follows a December incident when a Starlink satellite experienced a propulsion tank failure that created debris and caused it to fall four kilometers in altitude, highlighting the need for faster deorbiting of malfunctioning spacecraft.
Source
SpaceX will lower approximately 4,400 Starlink satellites from 550 kilometers to 480 kilometers altitude throughout 2026, according to Michael Nicolls, the company's vice president of Starlink engineering, in a move coordinated with other operators, regulators, and U.S. Space Command.
The orbital adjustment will reduce ballistic decay time by over 80% during solar minimum—shortening the period from more than four years to just months—and decrease collision risks since debris objects and planned constellations are significantly lower below 500 kilometers.
The announcement follows a December incident when a Starlink satellite experienced a propulsion tank failure that created debris and caused it to fall four kilometers in altitude, highlighting the need for faster deorbiting of malfunctioning spacecraft.
Source
Baidu's AI chip unit Kunlunxin files for Hong Kong IPO
Baidu announced that its AI chip unit Kunlunxin has confidentially filed for a Hong Kong listing, following a fundraising round that valued the company at approximately $3 billion.
The proposed spin-off aims to showcase Kunlunxin's value as an independent entity and broaden financing channels, though the company is expected to remain a subsidiary of Baidu after completion.
The filing comes amid a wave of Chinese AI chip IPOs, including Shanghai Biren Technology, which raised $717 million and began trading January 2, as Beijing pushes to develop domestic semiconductor alternatives amid U.S. export restrictions.
Source
Baidu announced that its AI chip unit Kunlunxin has confidentially filed for a Hong Kong listing, following a fundraising round that valued the company at approximately $3 billion.
The proposed spin-off aims to showcase Kunlunxin's value as an independent entity and broaden financing channels, though the company is expected to remain a subsidiary of Baidu after completion.
The filing comes amid a wave of Chinese AI chip IPOs, including Shanghai Biren Technology, which raised $717 million and began trading January 2, as Beijing pushes to develop domestic semiconductor alternatives amid U.S. export restrictions.
Source
Forwarded from Innovation Feed (Bisnu Ray)
BleepingComputer
Google is testing a new image AI and it's going to be its fastest model
Google is testing a new image AI model called "Nano Banana 2 Flash," and it's going to be as good as the Gemini 3 Pro Nano Banana, but it'll be cheaper.
#AI #Google #Gemini
Google is testing a new image AI and it's going to be its fastest model
Google is testing a new image AI model called "Nano Banana 2 Flash," and it's going to be faster than the Nano Banana Pro.
Key highlights
- Nano Banana 2 Flash is part of Google's Flash lineup and aims to be faster than prior Nano Banana iterations
- It will be faster but not as powerful as Nano Banana Pro (Gemini 3 Pro Image)
- The model is expected to be more affordable while delivering quicker image generation
- Nano Banana Pro remains Google's top-end image generation/editing model for harder creative tasks
Read More: Google is testing a new image AI and it's going to be its fastest model
Google is testing a new image AI and it's going to be its fastest model
Google is testing a new image AI model called "Nano Banana 2 Flash," and it's going to be faster than the Nano Banana Pro.
Key highlights
- Nano Banana 2 Flash is part of Google's Flash lineup and aims to be faster than prior Nano Banana iterations
- It will be faster but not as powerful as Nano Banana Pro (Gemini 3 Pro Image)
- The model is expected to be more affordable while delivering quicker image generation
- Nano Banana Pro remains Google's top-end image generation/editing model for harder creative tasks
Read More: Google is testing a new image AI and it's going to be its fastest model
Self-powered quantum maser emits stable microwave signals
Scientists at TU Wien and the Okinawa Institute of Science and Technology achieved the first self-induced superradiant quantum maser, where quantum particles spontaneously cooperate to generate stable microwave signals without external power.
The team coupled nitrogen-vacancy centers in diamond to a microwave cavity and discovered that dipole-dipole interactions between spins—previously considered disruptive—actually sustain emission by continuously redistributing energy and repopulating energy levels.
The breakthrough, published in Nature Physics, could enable ultra-precise atomic clocks, improved GPS navigation, enhanced radar systems, and quantum sensors for medical imaging and environmental monitoring.
Source
Scientists at TU Wien and the Okinawa Institute of Science and Technology achieved the first self-induced superradiant quantum maser, where quantum particles spontaneously cooperate to generate stable microwave signals without external power.
The team coupled nitrogen-vacancy centers in diamond to a microwave cavity and discovered that dipole-dipole interactions between spins—previously considered disruptive—actually sustain emission by continuously redistributing energy and repopulating energy levels.
The breakthrough, published in Nature Physics, could enable ultra-precise atomic clocks, improved GPS navigation, enhanced radar systems, and quantum sensors for medical imaging and environmental monitoring.
Source
Chinese firms claim 21 of 38 humanoid robotics spots at CES
Chinese companies dominate CES 2026's humanoid robotics category with 21 of 38 exhibitors, including Unitree Robotics, AgiBot, and Noetix Robotics, as the tech show opens in Las Vegas this week.
China has filed 7,705 humanoid patents over five years compared to 1,561 in the U.S., according to Morgan Stanley
, while AgiBot produced 5,000 robots in under three years versus several hundred by Tesla.
Beijing made robotics central to its 15th five-year plan to address labor shortages, while U.S. Commerce Secretary Howard Lutnick reportedly considers an executive order on a national robotics strategy.
Source
Chinese companies dominate CES 2026's humanoid robotics category with 21 of 38 exhibitors, including Unitree Robotics, AgiBot, and Noetix Robotics, as the tech show opens in Las Vegas this week.
China has filed 7,705 humanoid patents over five years compared to 1,561 in the U.S., according to Morgan Stanley
, while AgiBot produced 5,000 robots in under three years versus several hundred by Tesla.
Beijing made robotics central to its 15th five-year plan to address labor shortages, while U.S. Commerce Secretary Howard Lutnick reportedly considers an executive order on a national robotics strategy.
Source
Tech giants deploy AI chatbots in millions of classrooms
Elon Musk's xAI, Microsoft, and OpenAI have rapidly expanded AI chatbot deployment in schools worldwide, with El Salvador launching what it calls the world's first nationwide AI tutor program serving over 1 million students using Grok, while the UAE and Kazakhstan each granted access to hundreds of thousands of students and teachers.
The initiatives come as countries including Estonia and Iceland have implemented more cautious national programs with modified educational chatbots designed to guide step-by-step problem-solving rather than provide immediate answers, with officials acknowledging AI "can also cause significant harm."
A February study by Microsoft and Carnegie Mellon University found that increased reliance on AI tools correlates with declining critical thinking skills, while UNICEF warned that unguided AI use may "de-skill both students and teachers" and divert funds from proven educational approaches.
Source
Elon Musk's xAI, Microsoft, and OpenAI have rapidly expanded AI chatbot deployment in schools worldwide, with El Salvador launching what it calls the world's first nationwide AI tutor program serving over 1 million students using Grok, while the UAE and Kazakhstan each granted access to hundreds of thousands of students and teachers.
The initiatives come as countries including Estonia and Iceland have implemented more cautious national programs with modified educational chatbots designed to guide step-by-step problem-solving rather than provide immediate answers, with officials acknowledging AI "can also cause significant harm."
A February study by Microsoft and Carnegie Mellon University found that increased reliance on AI tools correlates with declining critical thinking skills, while UNICEF warned that unguided AI use may "de-skill both students and teachers" and divert funds from proven educational approaches.
Source
87 runaway stars mapped to trace Milky Way's dark matter
Astronomers from China identified 87 hypervelocity stars traveling fast enough to escape the Milky Way's gravitational pull, with seven exceeding 800 km/s, using data from the Sloan Digital Sky Survey, LAMOST telescope, and Gaia satellite.
The stars' spatial distribution points to origins linked to the Hills mechanism, where Sagittarius A*, the supermassive black hole at the galaxy's center, and the Magellanic Clouds fling stars away at extreme velocities after gravitational encounters.
By tracing these runaway stars' trajectories backward, scientists can map the Milky Way's gravitational potential and dark matter distribution in the halo, which holds roughly five times more mass than ordinary matter.
Source
Astronomers from China identified 87 hypervelocity stars traveling fast enough to escape the Milky Way's gravitational pull, with seven exceeding 800 km/s, using data from the Sloan Digital Sky Survey, LAMOST telescope, and Gaia satellite.
The stars' spatial distribution points to origins linked to the Hills mechanism, where Sagittarius A*, the supermassive black hole at the galaxy's center, and the Magellanic Clouds fling stars away at extreme velocities after gravitational encounters.
By tracing these runaway stars' trajectories backward, scientists can map the Milky Way's gravitational potential and dark matter distribution in the halo, which holds roughly five times more mass than ordinary matter.
Source
Grok AI admits ethical lapses after generating sexualized images of minors
Grok, Elon Musk's AI chatbot, generated sexualized images of two young girls estimated to be ages 12-16 in response to a user prompt on December 28, violating its own policies and potentially U.S. laws on child sexual abuse material.
The incident is part of a broader misuse pattern where users exploited Grok to manipulate photos of women and children into sexually explicit content without consent, with the AI generating over 70 public images per minute at its peak on New Year's Eve.
xAI disabled Grok's public media section and implemented restrictions after global backlash, while India's government considers stricter social media regulations and legal experts cite violations of privacy and sexual violence laws.
Source
Grok, Elon Musk's AI chatbot, generated sexualized images of two young girls estimated to be ages 12-16 in response to a user prompt on December 28, violating its own policies and potentially U.S. laws on child sexual abuse material.
The incident is part of a broader misuse pattern where users exploited Grok to manipulate photos of women and children into sexually explicit content without consent, with the AI generating over 70 public images per minute at its peak on New Year's Eve.
xAI disabled Grok's public media section and implemented restrictions after global backlash, while India's government considers stricter social media regulations and legal experts cite violations of privacy and sexual violence laws.
Source
AI agents surge into enterprise as governance lags
Technology leaders say 2026 will be a decisive year for AI adoption in business as organizations deploy autonomous AI agents at scale across core operations, with analysts projecting 40% of enterprise applications will integrate AI agents by year-end, up from less than 5% in 2025.
Despite rapid adoption, only 6% of organizations have implemented adequate governance structures to manage autonomous agents safely, creating "shadow AI" risks as 75% of workers use generative AI tools without employer authorization, exposing companies to data leakage and security vulnerabilities.
Security experts warn that identity management and API security have become critical control points as AI agents assume employee permissions and access corporate systems, with attackers increasingly targeting identity systems as the "easiest and most high-risk entry point."
Technology leaders say 2026 will be a decisive year for AI adoption in business as organizations deploy autonomous AI agents at scale across core operations, with analysts projecting 40% of enterprise applications will integrate AI agents by year-end, up from less than 5% in 2025.
Despite rapid adoption, only 6% of organizations have implemented adequate governance structures to manage autonomous agents safely, creating "shadow AI" risks as 75% of workers use generative AI tools without employer authorization, exposing companies to data leakage and security vulnerabilities.
Security experts warn that identity management and API security have become critical control points as AI agents assume employee permissions and access corporate systems, with attackers increasingly targeting identity systems as the "easiest and most high-risk entry point."
Intel, AMD to unveil rival AI chips at CES 2026
Intel will officially launch its Core Ultra Series 3 processors, code-named Panther Lake, at CES 2026 on January 5, with the chips built on Intel's advanced 18A manufacturing process and promising 50% faster CPU and GPU performance compared to the previous generation, while AMD CEO Lisa Su will deliver a competing keynote the same evening to unveil the company's Ryzen AI 400 series.
The competing launches come as AI PCs are projected by Gartner to represent 55% of the global PC market by the end of 2026, totaling approximately 143 million units, with 40% of software vendors expected to prioritize AI capabilities that run directly on PCs by then, up from just 2% in 2024.
Panther Lake represents a critical manufacturing milestone for Intel, featuring new RibbonFET transistors and PowerVia backside power delivery that deliver up to 15% better performance per watt, with production already underway at the company's Fab 52 facility in Chandler, Arizona.
Intel will officially launch its Core Ultra Series 3 processors, code-named Panther Lake, at CES 2026 on January 5, with the chips built on Intel's advanced 18A manufacturing process and promising 50% faster CPU and GPU performance compared to the previous generation, while AMD CEO Lisa Su will deliver a competing keynote the same evening to unveil the company's Ryzen AI 400 series.
The competing launches come as AI PCs are projected by Gartner to represent 55% of the global PC market by the end of 2026, totaling approximately 143 million units, with 40% of software vendors expected to prioritize AI capabilities that run directly on PCs by then, up from just 2% in 2024.
Panther Lake represents a critical manufacturing milestone for Intel, featuring new RibbonFET transistors and PowerVia backside power delivery that deliver up to 15% better performance per watt, with production already underway at the company's Fab 52 facility in Chandler, Arizona.