←Search results

By relevance · Total 2327 results Clear search

Friday

  1. Phoenix Technology

    US tech giants' $750 billion AI investment faces challenges as core business growth slows

    According to Bloomberg, the advertising, consumer and retail businesses that underpin US tech giants' combined $750 billion in AI investment this year have shown signs of slowing growth. Alphabet's search advertising revenue posted its lowest growth rate in five quarters, Amazon's advertising business grew 11% year over year, the slowest since it began disclosing the figure separately in 2022, and its online stores grew only 2%. Google Chief Business Officer Philipp Schindler acknowledged that search queries under AI modes are monetized at a lower level, and Alphabet's stock suffered its largest drop in history, with nearly $500 billion in market value wiped out.

    According to Bloomberg, the advertising, consumer and retail businesses that underpin US tech giants' combined $750 billion in AI investment this year have shown signs of slowing growth. Alphabet's search advertising revenue posted its lowest growth rate in five quarters, Amazon's advertising business grew 11% year over year, the slowest since it began disclosing the figure separately in 2022, and its online stores grew only 2%. Google Chief Business Officer Philipp Schindler acknowledged that s

    Matched fieldsBody

  2. Leikeji

    AI glasses diverge in 2026: what should you look for when buying smart glasses during Double 11?

    In the second quarter of 2026, China's smart glasses shipments fell 8% year-on-year, the first negative growth, but AI camera glasses grew 81.4% year-on-year and AR/XR glasses grew 123%, while audio-only glasses cooled due to homogenization. The article provides buying criteria for three scenarios—AI glasses with displays, AI camera glasses, and AR movie-viewing glasses—and lists Double 11 prices including RayNeo iO (34 grams, 2,499 yuan, 1,911.74 yuan after national subsidy), RayNeo V4, Huawei AI Glasses, Qianwen N1 Pro, and the RayNeo GT series.

    In the second quarter of 2026, China's smart glasses shipments fell 8% year-on-year, the first negative growth, but AI camera glasses grew 81.4% year-on-year and AR/XR glasses grew 123%, while audio-only glasses cooled due to homogenization. The article provides buying criteria for three scenarios—AI glasses with displays, AI camera glasses, and AR movie-viewing glasses—and lists Double 11 prices including RayNeo iO (34 grams, 2,499 yuan, 1,911.74 yuan after national subsidy), RayNeo V4, Huawei

    Matched fieldsBody

  3. CNETSelected

    Mistral Releases Flagship Model Mistral Large 4 and Opens Weight Downloads

    Mistral AI has released its new flagship frontier model, Mistral Large 4, codenamed Le Chonk, making it available for download after ending its public preview. The model supports text and image processing, with a focus on strengthening defensive cybersecurity, agentic capabilities, and geospatial and grounding capabilities, and is said to achieve state-of-the-art results on agentic benchmarks such as SWE-bench and Terminal Bench. The model was trained on more than 160 languages and is available via Mistral Studio, the Mistral API, OpenRouter, Hugging Face open weights, Vibe Code, and Nvidia-hosted endpoints.

    Mistral AI has released its new flagship frontier model, Mistral Large 4, codenamed Le Chonk, making it available for download after ending its public preview. The model supports text and image processing, with a focus on strengthening defensive cybersecurity, agentic capabilities, and geospatial and grounding capabilities, and is said to achieve state-of-the-art results on agentic benchmarks such as SWE-bench and Terminal Bench. The model was trained on more than 160 languages and is available

    Matched fieldsBody

Wednesday

  1. The Rundown AI

    Google DeepMind releases the Genie 4 world model, which can generate sound effects and converse with characters in the scene

    Google DeepMind has released Genie 4, a new-generation world model that can turn prompts into explorable, realistic scenes in real time, and for the first time the Genie model itself generates background sounds and music, while characters in the scene can also hold real-time conversations. Users move via the keyboard and use text prompts to modify content in the scene instantly, with the scene remaining coherent across multiple inputs. In the demo, a simulated Apptronik Apollo robot plays checkers, with each move planned by Gemini 4 Argon, pointing to Genie's future use in robot training.

    Google DeepMind has released Genie 4, a new-generation world model that can turn prompts into explorable, realistic scenes in real time, and for the first time the Genie model itself generates background sounds and music, while characters in the scene can also hold real-time conversations. Users move via the keyboard and use text prompts to modify content in the scene instantly, with the scene remaining coherent across multiple inputs. In the demo, a simulated Apptronik Apollo robot plays checke

    Matched fieldsBody

  2. QbitAI

    Physical Intelligence in Talks for at Least $1 Billion in New Funding, Valuation Could Reach $20 Billion

    According to Bloomberg, U.S. robotics AI company Physical Intelligence is in talks for a new funding round of at least $1 billion, with existing investors Avenir and Lux Capital considering leading the round, and a post-money valuation expected at about $20 billion. The company was founded in 2024 and was valued at $5.6 billion in its November funding round last year; it currently has no commercial revenue and has not mass-produced robot products. Its technical approach is to build on VLM by introducing action generation capabilities to train VLA models, and it has iterated through π0, π0.5, and π0.7; this month it launched a beta version of a robot foundation model API, offering model fine-tuning and inference services to external developers.

    According to Bloomberg, U.S. robotics AI company Physical Intelligence is in talks for a new funding round of at least $1 billion, with existing investors Avenir and Lux Capital considering leading the round, and a post-money valuation expected at about $20 billion. The company was founded in 2024 and was valued at $5.6 billion in its November funding round last year; it currently has no commercial revenue and has not mass-produced robot products. Its technical approach is to build on VLM by int

    Matched fieldsBody

  3. TechSpot

    Gartner Survey: Only 3% of Organizations Achieve Cost Reduction Through AI Productivity, While 85% Still Plan to Increase Investment

    Data presented by Gartner at the HR Symposium in Orlando shows that only 3% of organizations have met their cost-reduction goals through productivity gains, and 97% of organizations cannot achieve cost savings through productivity alone, yet 77% of board members still expect AI to deliver productivity and cost benefits. Nearly 40% of companies have already exceeded their AI budgets, and 85% of companies still plan to continue increasing investment next year; Gartner recommends that enterprises first reach consensus on where AI creates value, and among the 297 chief HR officers it surveyed, 78% said leadership has no consensus on this.

    Data presented by Gartner at the HR Symposium in Orlando shows that only 3% of organizations have met their cost-reduction goals through productivity gains, and 97% of organizations cannot achieve cost savings through productivity alone, yet 77% of board members still expect AI to deliver productivity and cost benefits. Nearly 40% of companies have already exceeded their AI budgets, and 85% of companies still plan to continue increasing investment next year; Gartner recommends that enterprises f

    Matched fieldsBody

  4. Leiphone

    Liu Rui, Director of Huawei's Xiaoyi Large Model Application Lab: AI assistants must move beyond the chat box, from passive response to proactive service

    Liu Rui, the first director of Huawei's Xiaoyi Large Model Application Lab, proposed that AI assistants must move beyond the chat box and shift from passive response to proactive service. The conversational photo-editing technology he previously failed to get approved as a project at SenseTime has already been unveiled at the 2025 Huawei Mate 80 launch event. The lab has also launched the Aura streaming video understanding model, which supports real-time video interaction and proactive service, as well as Boogu, an open-source foundational image generation model. Liu Rui believes the biggest bottleneck at present remains insufficient model capability.

    Liu Rui, the first director of Huawei's Xiaoyi Large Model Application Lab, proposed that AI assistants must move beyond the chat box and shift from passive response to proactive service. The conversational photo-editing technology he previously failed to get approved as a project at SenseTime has already been unveiled at the 2025 Huawei Mate 80 launch event. The lab has also launched the Aura streaming video understanding model, which supports real-time video interaction and proactive service,

    Matched fieldsBody

  5. TechCrunch AI

    Have AI executives become disconnected from the general public, making it hard for them to build products for ordinary people?

    In an interview, Marissa Mayer said her new app Dazzle generated birthday speech talking points for her 10-year-old daughter that were "empathetic, creative, and animal-loving," sparking criticism that Silicon Valley culture is out of touch with ordinary people. Pew research shows that 50% of U.S. adults believe AI will weaken people's ability to form meaningful relationships, while only 5% believe AI will make such connections easier. The article also mentions that Sam Altman said he cannot imagine raising a newborn without ChatGPT, and that Mark Zuckerberg predicted outsourcing emotional needs to AI therapists or AI companions will become more accepted.

    In an interview, Marissa Mayer said her new app Dazzle generated birthday speech talking points for her 10-year-old daughter that were "empathetic, creative, and animal-loving," sparking criticism that Silicon Valley culture is out of touch with ordinary people. Pew research shows that 50% of U.S. adults believe AI will weaken people's ability to form meaningful relationships, while only 5% believe AI will make such connections easier. The article also mentions that Sam Altman said he cannot ima

    Matched fieldsBody

  6. Jiwei.comSelected

    Anthropic Files IPO Prospectus; US Media Says Its Listing Will Test the Entire AI Industry

    Anthropic has officially filed its IPO prospectus and could list as early as next month; the IPO may raise $100 billion. The prospectus discloses that it plans to invest up to $807 billion over the next decade in data centers, chips, and cloud computing, but its third-quarter revenue was $16.8 billion, up 45% quarter-over-quarter, a slower growth rate than the 141% in the second quarter. The New York Times said this listing will, for the first time, give the outside world an in-depth look at the operating conditions of the key companies underpinning the AI boom, and will test the sustainability of massive investments in chips, data centers, and other areas.

    Anthropic has officially filed its IPO prospectus and could list as early as next month; the IPO may raise $100 billion. The prospectus discloses that it plans to invest up to $807 billion over the next decade in data centers, chips, and cloud computing, but its third-quarter revenue was $16.8 billion, up 45% quarter-over-quarter, a slower growth rate than the 141% in the second quarter. The New York Times said this listing will, for the first time, give the outside world an in-depth look at the

    Matched fieldsBody

  7. Solidot

    Data centers are stirring up the world: about 12,250 worldwide, with the United States leading at 53.73 GW of installed capacity

    The number of data centers worldwide has grown from about 8,000 in 2021 to about 12,250, with the United States leading at 4,767 and Virginia having the highest density. U.S. data center installed capacity reaches 53.73 GW, China 31.90 GW, and the EU 11.92 GW; total global data center electricity consumption is estimated to have exceeded that of all of France. The largest data center currently in operation is China Telecom's facility in Hohhot, Inner Mongolia, covering about 1 million square meters.

    The number of data centers worldwide has grown from about 8,000 in 2021 to about 12,250, with the United States leading at 4,767 and Virginia having the highest density. U.S. data center installed capacity reaches 53.73 GW, China 31.90 GW, and the EU 11.92 GW; total global data center electricity consumption is estimated to have exceeded that of all of France. The largest data center currently in operation is China Telecom's facility in Hohhot, Inner Mongolia, covering about 1 million square met

    Matched fieldsBody

  8. TMTPost · AGI

    Edge AI Daily Morning Brief: Anthropic Files for IPO, OpenAI Releases GPT-6.1…

    On October 26, 2026, Anthropic submitted an S-1 to the SEC, planning to list on Nasdaq under the ticker "ANT," with an IPO valuation potentially exceeding $2 trillion. OpenAI released GPT-6.1 Terra the same day, priced at $1 per million input tokens and $5 per million output tokens on the standard API, completing tasks more than three times faster than the flagship GPT-6 Astra. Google DeepMind launched a research preview of its world model Genie 4, achieving synchronized audio and video generation for the first time and doubling its consistency duration to two minutes.

    On October 26, 2026, Anthropic submitted an S-1 to the SEC, planning to list on Nasdaq under the ticker "ANT," with an IPO valuation potentially exceeding $2 trillion. OpenAI released GPT-6.1 Terra the same day, priced at $1 per million input tokens and $5 per million output tokens on the standard API, completing tasks more than three times faster than the flagship GPT-6 Astra. Google DeepMind launched a research preview of its world model Genie 4, achieving synchronized audio and video generati

    Matched fieldsBody

Thursday

  1. MIT Tech Review

    Partnership on AI research: Agent frameworks generally lack observability records

    Madhulika Srikumar and Claire Leibowicz of Partnership on AI write that current agent frameworks cannot reliably generate the key observability records needed for safe deployment. Together with colleagues from academia, civil society, and industry, the authors studied the four most widely used agent frameworks and found that signals such as persistent Agent IDs and human approval records are missing, making it difficult to determine which agent did what and whether it exceeded its authority. The article recommends using OpenTelemetry to fill in the records of "what happened," and calls on framework developers, enterprises adopting agents, and regulators to each bear their respective responsibilities.

    Madhulika Srikumar and Claire Leibowicz of Partnership on AI write that current agent frameworks cannot reliably generate the key observability records needed for safe deployment. Together with colleagues from academia, civil society, and industry, the authors studied the four most widely used agent frameworks and found that signals such as persistent Agent IDs and human approval records are missing, making it difficult to determine which agent did what and whether it exceeded its authority. The

    Matched fieldsBody

  2. t3n

    Mistral Large 4 weights go live on Hugging Face; the company says it leads on CyberGym-E2E and…

    After a three-week preview, Mistral AI has released the weights of Mistral Large 4 on Hugging Face. This multimodal model was trained on 3,800 Nvidia accelerators and has a parameter count of one trillion, and it is currently still difficult to run on a gaming PC or a single GPU. Mistral says it leads GLM-5.3 and Kimi K3 on the cybersecurity benchmark CyberGym-E2E, and is on par with GLM-5.3 on the agentic coding benchmark Terminal Bench 4.0, but there is no independent evaluation yet; preview pricing is $1.4 per million input tokens and $4.2 per million output tokens.

    After a three-week preview, Mistral AI has released the weights of Mistral Large 4 on Hugging Face. This multimodal model was trained on 3,800 Nvidia accelerators and has a parameter count of one trillion, and it is currently still difficult to run on a gaming PC or a single GPU. Mistral says it leads GLM-5.3 and Kimi K3 on the cybersecurity benchmark CyberGym-E2E, and is on par with GLM-5.3 on the agentic coding benchmark Terminal Bench 4.0, but there is no independent evaluation yet; preview p

    Matched fieldsBody

  3. QbitAI

    WeRide subsidiary JingShuo Technology completes tens of millions of USD pre-A round financing

    WeRide's subsidiary JingShuo Technology has completed a pre-A financing round of tens of millions of USD, with investors including Jiayu Capital, HKSTP (Hong Kong Science and Technology Parks), and Sirius Capital. JingShuo Technology centers on its data-model integrated platform WorldEngine and its self-developed world model GENESIS-Robotics, and already has 100+ customers and ecosystem partners, with orders in hand amounting to several billion yuan and annual revenue growth exceeding 100%. This round of financing will be used to expand the core R&D team and deepen the synergistic capabilities of the data platform, world model, and skill asset engine.

    WeRide's subsidiary JingShuo Technology has completed a pre-A financing round of tens of millions of USD, with investors including Jiayu Capital, HKSTP (Hong Kong Science and Technology Parks), and Sirius Capital. JingShuo Technology centers on its data-model integrated platform WorldEngine and its self-developed world model GENESIS-Robotics, and already has 100+ customers and ecosystem partners, with orders in hand amounting to several billion yuan and annual revenue growth exceeding 100%. This

    Matched fieldsBody

  4. AWS Machine Learning

    Distributed Training on EKS with Ray Train DLC and EFA

    AWS has released the Ray Train DLC image ray:train-ml-cuda-v1.1, which comes pre-integrated with Ray 2.58.0, PyTorch 2.13.0, CUDA 13.0.2, NCCL 2.29.7, and EFA along with the AWS OFI NCCL plugin, enabling distributed training on EKS without the need to build images yourself. The example uses PyTorch FSDP to fine-tune Qwen2.5-1.5B on 8 NVIDIA L4 GPUs across two g6.12xlarge instances, with cross-node communication going directly over EFA. The image is publicly hosted on Amazon ECR Public, so nodes can pull it without credentials.

    EKS cluster with a CPU system node running KubeRay and the Ray head, and two EFA-enabled g6.12xlarge GPU worker nodes in one Availability Zone

    AWS has released the Ray Train DLC image ray:train-ml-cuda-v1.1, which comes pre-integrated with Ray 2.58.0, PyTorch 2.13.0, CUDA 13.0.2, NCCL 2.29.7, and EFA along with the AWS OFI NCCL plugin, enabling distributed training on EKS without the need to build images yourself. The example uses PyTorch FSDP to fine-tune Qwen2.5-1.5B on 8 NVIDIA L4 GPUs across two g6.12xlarge instances, with cross-node communication going directly over EFA. The image is publicly hosted on Amazon ECR Public, so nodes

    Matched fieldsBody

  5. TechSpot

    Check Point Turns Network Firewalls into a Cloud Subscription Service

    Check Point launches Firewall-as-a-Subscription, turning network firewalls into a cloud subscription service. Customers no longer need to purchase hardware devices upfront, but must commit to at least 36 months. The service is based on its latest firewall platform R82.20 and includes an agentic console for managing firewall devices and services, Force hardware firewalls, SandBlast's AI threat protection, and instant security updates. Check Point says the move is intended to address emerging threats brought by AI agents and to lower the barrier for enterprises to upgrade with a predictable monthly fee.

    Check Point launches Firewall-as-a-Subscription, turning network firewalls into a cloud subscription service. Customers no longer need to purchase hardware devices upfront, but must commit to at least 36 months. The service is based on its latest firewall platform R82.20 and includes an agentic console for managing firewall devices and services, Force hardware firewalls, SandBlast's AI threat protection, and instant security updates. Check Point says the move is intended to address emerging thre

    Matched fieldsBody

  6. TechSpot

    Alphabet shares fell more than 10% after hours, as Google AI search monetization lags traditional search

    Alphabet's quarterly revenue was $127.6 billion, up 25% year-over-year, but earnings per share of $2.12 fell short of the expected approximately $3.01, and the stock fell more than 10% after hours. Chief Business Officer Philipp Schindler said that AI Mode's monetization capability per query is lower than traditional search, which is one of the reasons why search revenue growth slowed from 17% to 12%. Capital expenditures this quarter reached $51.4 billion, free cash flow was negative $9.1 billion, and Google Cloud revenue rose to $30.71 billion.

    Alphabet's quarterly revenue was $127.6 billion, up 25% year-over-year, but earnings per share of $2.12 fell short of the expected approximately $3.01, and the stock fell more than 10% after hours. Chief Business Officer Philipp Schindler said that AI Mode's monetization capability per query is lower than traditional search, which is one of the reasons why search revenue growth slowed from 17% to 12%. Capital expenditures this quarter reached $51.4 billion, free cash flow was negative $9.1 billi

    Matched fieldsBody

  7. Leiphone

    DeepSeek Open-Sources Inference Engine DeepInfer, Discloses That a Single Cluster Processes About 155 Trillion Tokens per Day

    DeepSeek has open-sourced its inference engine DeepInfer on GitHub—the inference engine behind DeepSeek-V4.1-Flash—which includes technologies such as MoE expert parallelism, asynchronous batch scheduling, custom kernels, and speculative decoding, and discloses that a single cluster needs to process about 155 trillion tokens per day. Its Zero-Overhead Batch Scheduler lets the CPU schedule computation one round ahead, overlapping with GPU forward passes, and handles cross-round state misalignment through total_inflight_tokens accounting, Remap mapping, and DeferredRelease deferred reclamation. What is currently available is offline inference capability; components such as online KV Cache storage and cross-replica load balancing have not yet been fully disclosed, and the official team has not separately published end-to-end speedup data for this scheduling mechanism relative to synchronous solutions.

    DeepSeek has open-sourced its inference engine DeepInfer on GitHub—the inference engine behind DeepSeek-V4.1-Flash—which includes technologies such as MoE expert parallelism, asynchronous batch scheduling, custom kernels, and speculative decoding, and discloses that a single cluster needs to process about 155 trillion tokens per day. Its Zero-Overhead Batch Scheduler lets the CPU schedule computation one round ahead, overlapping with GPU forward passes, and handles cross-round state misalignment

    Matched fieldsBody

  8. TechCrunch AI

    Google Launches Connected Apps Platform, Opening Gemini and Search Access to Developers

    Google has released the Connected Apps Platform, allowing enterprises and developers to submit their own apps for integration with Gemini and Google Search, and to choose to launch on only one or both. The platform is currently available in beta preview to US developers; after launch, it will allow tracking of traffic, latency, and error rates, while retaining brand identity and verified accounts. Gemini already has more than 50 Connected Apps, including partners such as Adobe, Experian, Localiza, and Peloton; these apps will enter Google Search in the coming weeks. The meditation app Calm will also launch on Gemini and Search.

    Google has released the Connected Apps Platform, allowing enterprises and developers to submit their own apps for integration with Gemini and Google Search, and to choose to launch on only one or both. The platform is currently available in beta preview to US developers; after launch, it will allow tracking of traffic, latency, and error rates, while retaining brand identity and verified accounts. Gemini already has more than 50 Connected Apps, including partners such as Adobe, Experian, Localiz

    Matched fieldsBody

Tuesday

  1. Zhidongxi

    Rokid launches full-color AI glasses Rokid Color and second-generation Rokid Glasses 2, starting at 4,999 yuan

    Leqi Rokid unveiled the second-generation AI glasses Rokid Glasses 2, its first full-color AI glasses Rokid Color, and the smart ring Rokid Ring at its October 27 launch event in Hangzhou. Rokid Color starts at 4,999 yuan and uses a 0.16cc Micro-LED full-color optical engine exclusively customized with JBD, supporting 16.77 million colors and a 167% sRGB color gamut, with the device weighing 55g; Rokid Glasses 2 starts at 3,699 yuan and is the first in China to feature the Qualcomm Snapdragon AR1+ platform, equipped with a dual-chip architecture combining it with the Bestechnic BES2900BP. Rokid also released the Rokid Prism connector, which allows AI glasses to invoke desktop agents such as Hermes and Codex to execute tasks.

    Leqi Rokid unveiled the second-generation AI glasses Rokid Glasses 2, its first full-color AI glasses Rokid Color, and the smart ring Rokid Ring at its October 27 launch event in Hangzhou. Rokid Color starts at 4,999 yuan and uses a 0.16cc Micro-LED full-color optical engine exclusively customized with JBD, supporting 16.77 million colors and a 167% sRGB color gamut, with the device weighing 55g; Rokid Glasses 2 starts at 3,699 yuan and is the first in China to feature the Qualcomm Snapdragon AR

    Matched fieldsBody