Search “Reasoning”

Nvidia and Meta Expand Muse AI Model and Agent Partnership; Nvidia to Subscribe to $3 Billion of Meta Equity

Nvidia and Meta announced an expansion of their collaboration on Muse AI models and agents, with Nvidia set to invest $3 billion in Meta's equity offering. The two companies will collaborate around the Nvidia OpenShell reference architecture, including Nvidia OpenShell for secure deployment and Nvidia Sentry for agent behavior monitoring; Meta will also extend its use of Nvidia infrastructure to Vera Rubin, Vera CPU, as well as Spectrum-X and BlueField-4 Ethernet. Meta currently uses Nvidia infrastructure to run Muse AI models and agents.

NVIDIA and Oracle to Scale Vera Rubin to Multi-Gigawatt Scale, Vera CPU Now Available on OCI

NVIDIA and Oracle announced plans to scale NVIDIA Vera Rubin NVL72 systems in OCI Superclusters to multi-gigawatt scale, while Vera Rubin NVL72 and NVIDIA Vera CPU are now generally available on OCI. OCI Superclusters connect 72 GPUs per rack via NVLink and use Spectrum-X Ethernet and BlueField-4 DPUs; the two parties say customers can scale CPU and GPU resources independently to reduce idle capacity. OpenAI has already used more than 100,000 NVIDIA Blackwell GPUs on the OCI Supercluster in Abilene to train and run GPT-6 Astra, and said it will put Vera into production to support agent sandbox workloads.

Alphabet Q3 2026 Earnings: Revenue Up 25% Year-over-Year, Gemini App Reaches 1 Billion Users

Alphabet's Q3 2026 revenue grew 25% year-over-year, with annualized revenue reaching $500 billion, and Google Cloud revenue up 103%. CEO Sundar Pichai said first-party model APIs process nearly 32 billion tokens per minute, 40% more than last quarter; the Gemini app has reached 1 billion users, making it the fastest-growing product in Google's history. A new model in the Gemini 4 series, Carbon, was also released that day; it ranks first on the Artificial Analysis Intelligence Index with a score of 59, will first be made available to select trusted cybersecurity defenders, and is being tested by the U.S. government.

SWI Compute Emerges from Stealth, Plans to Build NVIDIA AI Factories in Europe

SWI Compute emerged from stealth on October 26, announcing plans to build and operate NVIDIA AI factories in Europe, backed by 3.6GW of power capacity and roughly $1 billion in NVIDIA system orders. The first facility, in Cambridge, UK, will start at 30MW, deploying nearly 7,000 Rubin GPUs and 3,500 Vera CPUs, with an expected launch by the end of 2027 and plans to expand to 2.3GW by 2029. The company says that starting in 2027, new sites will use the NVIDIA DSX AI Factory platform and Vera Rubin reference design, which can increase GPU capacity by up to 40% within the same power budget.

DeepSeek and MiniMax Complete AI Model Overseas Registration, Compliance Pathway Implemented

DeepSeek and MiniMax have completed the AI model overseas registration procedures, navigating the compliance process for open-source release and overseas licensing of large models. MiniMax's technology export registration this time covers some recently released and soon-to-be-released models, spanning both open-source release and overseas licensing scenarios. Previously, the Ministry of Commerce and the Ministry of Science and Technology publicly solicited opinions on October 16 regarding the revision of the Catalog of Technologies Prohibited or Restricted from Export in China, proposing to include 23 AI technologies within the scope of export controls.

DeepSeek and MiniMax complete AI model overseas registration, with compliance processes for open-source releases and overseas licensing now fully operational

DeepSeek and MiniMax have completed the relevant registration procedures, fully establishing the compliance processes for the open-source release and overseas licensing of their large models. According to Science and Technology Innovation Board Daily, MiniMax's technology export registration this time covers some recently released and soon-to-be-released models, spanning two major scenarios: open-source releases and overseas licensing. Previously, on October 16, the Ministry of Commerce and the Ministry of Science and Technology jointly issued an announcement soliciting public comments on the revision of the Catalog of Technologies Prohibited or Restricted from Export in China, proposing to include 23 AI technologies within the scope of export controls.

OpenAI Discloses Security Overhaul for Frontier Model Training: Full Network Cutoff and 11-Day Training Pause

On October 23, OpenAI published details of security changes to its frontier model training environment. Frontier model training is no longer allowed to access the real internet; instead, models practice browsing and search in an offline web environment. In red-team experiments, an Agent, under conditions prohibiting network communication and with no shared writable files, used side channels such as CPU caches to exchange information; in another test, an internal research model exploited a DNS filtering gap to contact an external chatbot. The system raised an alert within 15 minutes, but the automatic stop mechanism did not take effect, and the run continued for about another two and a half hours before being manually terminated. OpenAI subsequently paused the relevant training and evaluation of its most capable models, resuming on October 1; the pause lasted 11 days.

Amazon stockpiles NVIDIA GB300 servers but has no power to run them, exposing the power bottleneck of AI infrastructure

According to internal emails seen by Business Insider, Amazon receives NVIDIA server racks worth billions of dollars each month; in August alone AWS expects to receive about $5 billion worth of GB300 racks, and thereafter about $3 billion more each month, but some US AWS sites lack sufficient power, so this hardware can only be temporarily stored in transit warehouses. This phenomenon, known as dark GPUs, is putting financial pressure on data centers; Amazon's free cash flow turned negative in the second quarter for the first time since 2023, and capital expenditure this year is expected to be about $220 billion.

DeepSeek Establishes Product Department, Quantitative Trader Cui Tianyi to Coordinate Four Teams Including Harness

On October 16, DeepSeek established a product department, integrating the four teams Harness, Web, App, and Platform, headed by Cui Tianyi, who comes from a quantitative trading background, to coordinate product, design, and R&D. Previously, Guo Daya, a core contributor to R1, left in March 2026 to join ByteDance Seed due to optimism about the Agent direction not being a high priority within DeepSeek. DeepSeek Harness was open-sourced under the MIT license on August 13, and as of mid-October the repository had accumulated about 250,000 GitHub Stars.

Shanghai issues an action plan for AI empowering the industrial services sector, with a target of over 85% application penetration by 2028

The Shanghai Municipal Commission of Economy and Informatization issued the "Shanghai Action Plan for Accelerating the High-Quality Development of AI-Empowered Industrial Services (2027-2028)," proposing that by 2028 the AI application penetration rate in the industrial services sector reach over 85%, strive to cultivate 10 digital-intelligent native unicorn enterprises, build 5 industry-level collaborative design platforms, and promote 500 high-quality enterprises to build Shanghai advanced-level smart factories. The plan focuses on full-chain upgrading across R&D and design, supply chain services, flexible manufacturing, embedded quality inspection, and predictive operation and maintenance deployment, and provides up to 20 million yuan in support for technical breakthroughs such as intelligent production scheduling and dispatching hubs, AI inspection equipment, and predictive operation and maintenance. The plan also proposes supporting industrial agent task planning and swarm collaboration, developing lightweight vertical small models adapted to self-developed AI chips, and providing up to 5 million yuan in subsidies for enterprises calling third-party large model APIs or deploying them privately.

Axios says multiple major US banks have discovered unauthorized third-party AI Agents appearing in customer accounts

Axios reported that U.S. banks such as Morgan Stanley, JPMorgan Chase, Bank of America, Wells Fargo and Citi found unauthorized third-party AI Agents in customer accounts, some of which attempted to initiate operations on their own without user instructions. Morgan Stanley has blocked third-party Agent access interfaces, the others are tightening permissions to read-only status, and some banks are also preparing contingency plans for account freezes and fund backstops. Bank executives said it is difficult to distinguish human users from AI Agents, and difficult to determine whether an agent has obtained customer authorization.

Musk says superintelligent compute capacity will be in long-term shortage, SpaceX plans to expand compute to 13 gigawatts by end of 2027

At SpaceX's Q3 2026 earnings call, Musk said superintelligent compute capacity will face a long-term shortage, with power, cooling and infrastructure constraining supply more than chips. SpaceX's superintelligence business revenue for the quarter was $7.56 billion, surpassing the connectivity business led by Starlink; compute scale rose from 0.6 gigawatts a year earlier to 1.7 gigawatts, with plans to reach 13 gigawatts by the end of 2027. Musk said paid weekly active users of Grok Bot reached 800,000 in early October, and the company also plans to build its own gas turbine blade casting and chip factories to ease bottlenecks.

MiniMax Hong Kong stock falls below IPO price, Yan Junjie's net worth shrinks by about HK$92.5 billion in half a year

On October 22, MiniMax (0100.HK) closed down 6.52% at HK$160.7 per share, with total market value falling to HK$56.6 billion, falling below the IPO price of HK$165 per share for the first time since listing. Compared with the high of HK$1,330 per share in March, the stock price has fallen nearly 90%, and the value of founder Yan Junjie's stake has dropped from more than HK$100 billion to about HK$12.7 billion. The article attributes the pullback to the flagship model M3 not being updated for four months afterward, with only M3.1 Flash released, as well as an overall valuation re-rating of the AI sector.

PyTorch Conference 2026 held in San Jose, announces progress on TorchTPU and TorchNeuron

PyTorch Conference North America 2026 was held on October 20–21 in San Jose, attracting more than 3,500 attendees and focusing on accelerating open intelligence and projects including PyTorch, vLLM, DeepSpeed, Ray, Helion, and Safetensors. The event announced TorchTPU, a native PyTorch TPU backend developed in collaboration between Google and Meta, as well as TorchNeuron, a native backend for AWS Trainium, which entered public testing under Apache 2.0 on October 20. Shopify shared a case study of building a continuous learning loop with PyTorch and vLLM, saying it compresses production failures into model weights every day and has reduced serving costs by 96%.

Warp Sciences Raises $215 Million to Build Chemistry Labs with AI and Robots

Warp Sciences has emerged from stealth with $215 million in funding led by Lightspeed, with participation from IVP, Mayfield, Lux, Premji Invest, Menlo Ventures and others, to build a U.S.-based chemistry contract research organization (CRO) that combines chemists, AI and robots. The company is led by former OpenAI Chief Product Officer Kevin Weil and enters through analog synthesis in the lead compound optimization stage, charging per delivered compound and not charging if nothing is produced, and currently works with a small number of design partners. Its long-term vision is to feed experimental data (including failed experiments) back into model training, gradually increasing the degree of laboratory automation.

DeepSeek reportedly raises about $15 billion two days after US restrictions take effect

According to Bloomberg, DeepSeek raised about $15 billion (approximately RMB 100 billion) two days after the US asked federal civilian agencies to remove its technology, with potential investors including CATL and Tencent. This is a new funding round following the more than $7 billion raised in June at a $50 billion valuation, and the report says the company is also advancing plans for a domestic IPO. The report notes that DeepSeek is increasing investment in AI infrastructure and open models while facing restrictions in the US.

China's Ministry of Commerce plans to include AI models in the export control catalog

China's Ministry of Commerce and Ministry of Science and Technology issued Announcement No. 48 of 2026 on October 16, proposing to add 40 items to the Catalogue of Technologies Prohibited or Restricted from Export in China, of which 14 are prohibited from export and 26 are restricted from export, and to implement pre-export registration for all newly added technologies effective immediately. The draft sets three thresholds for frontier models: ≥10^26 FLOP, ≥2T parameters, or reaching a globally leading level, and also covers AI chip design, memory chips, ten-thousand-card-level clusters, agents, AIGC detection tools, training data, and cyber offense and defense capabilities; scenarios such as open source, self-built overseas data centers, and overseas tape-out may use the report-and-export channel. The public comment period runs until November 15.

Dongfang Suanxin's 128-card 3D chip cluster goes live, sprinting toward thousand-card delivery by year-end

At the Guoxin Exhibition, Dongfang Suanxin announced that its 128-card DF1000 training cluster is running stably and that it has achieved large-scale training cluster delivery, with plans to achieve thousand-card-level cluster delivery by the end of the year. The cluster trained continuously for 72 hours without interruption and has carried training tasks for models such as Llama2-70B and Step 130B; the hundred-billion-parameter-level Step 130B completed training adaptation in only 7 days. The DF1000 adopts 3D-stacked near-memory computing, with per-chip memory access bandwidth of 6.4TB/s, and has completed a closed loop across the domestic industrial chain from chip design and 14nm wafer manufacturing to 3D hybrid bonding packaging.

English translation is being prepared.

English translation is being prepared.