nvidia

AI startup Tavus released a new model called Griffin on October 1, referring to it as the first "human interaction model" that can understand and generate face-to-face conversations, paying attention not only to wording but also to expressions and pauses. Tavus stated that in tests, after a one-minute video call with Griffin-Lite, 26 out of 54 participants (48%) believed the other party was a real person; its previous generation system only convinced 1 out of 41 participants.Participants were told they would have a one-minute video call with another person to discuss things they were looking forward to this year, and only at the end of the call were they asked if they suspected the other party was not a real person. Tavus noted that those who became suspicious typically noticed within 20 seconds. This result comes from Tavus's own research page, where participants were recruited through a channel they call an independent research platform. Community comments on X pointed out that these results have not been independently verified and do not meet standard protocols.In the NVIDIA VideoFDB benchmark, Tavus claimed Griffin-Lite ranked first, with a generation dimension score of 3.83, the next best system scoring 2.8, and human reference scoring 3.92; the perception dimension score was 3.73, with the strongest baseline at 3.44 and human reference at 4.2. Tavus stated that NVIDIA independently conducted this evaluation. Griffin supports full duplex, allowing it to listen, see, and speak simultaneously, with an average audio-video delay of 0.43 seconds on the NVIDIA H100 chip. Tavus completed a $40 million Series B funding round led by CRV in November 2025.

NVIDIA's stock price reached a new historical high after experiencing two months of sell-offs, rising as much as 2.9% during the trading session. The market is optimistic that AI entities will drive an increase in semiconductor demand, boosting stock prices.After NVIDIA announced an increase in its buyback plan by $150 billion, the stock price further rose. So far this year, the stock price has increased by about 27%, with a market capitalization of approximately $5.7 trillion, just under $300 billion away from $6 trillion.

NVIDIA launched the DGX SPARK 64GB configuration this month with major partners, supporting the operation of models with up to 100 billion parameters on the device.

Morgan Stanley has reclassified NVIDIA (NVDA.O) as a preferred stock.

According to the Financial Times, Amazon (AMZN.O) is seeking to sell $8 billion worth of Nvidia (NVDA.O) chips to investors.

According to Gate market data, U.S. stock night trading chip stocks collectively strengthened, with SOXL, the three times leveraged semiconductor ETF from Direxion, rising over 9%, Arm rising 4%, AMD and Intel rising over 3%, SanDisk rising 2.8%, SK Hynix rising 2.7%, Broadcom rising 1.6%, NVIDIA rising 1.5%, and Micron Technology rising 1.4%.

AI native cloud platform GMI Cloud announced the completion of a new round of financing totaling $668 million, including $223 million in Series B equity financing and $445 million in credit financing, led by the investment company ARCHIV, with participation from NVIDIA.The company stated that its inference platform processes approximately 40 trillion tokens per week, with clients including Fireworks, Nous Research, OpenRouter, Trend Micro, and others.

Intel (INTC.O) has added NVIDIA (NVDA.O) OpenShell strategy layer to its AI toolkit.

Coreweave stated that, from the customer's perspective, NVIDIA chips are currently the best choice.

Jacobs Engineering Consulting will deploy data center digital twin technology for the NVIDIA Research Center.

Tianfeng International Securities analyst Guo Mingqi published that his latest supply chain check shows that NVIDIA has begun testing fiberglass-free copper-clad laminates primarily made of hydrocarbon resin to replace the previously tested fiberglass-free PTFE copper-clad laminates, along with fiberglass-free hydrocarbon prepregs. The test aims to confirm whether this combination can meet the requirements of the Rubin Ultra NVL576 exchange tray PCB. NVL576 is an eight-rack interconnection solution expected to enter mass production in the second half of 2027.Guo Mingqi stated that preliminary tests conducted to enhance PCB manufacturability show that the fiberglass-free hydrocarbon copper-clad laminate has met the high-frequency electrical requirements of the exchange tray. Its electrical performance lags behind the original PTFE solution but is better than the M9 level solution and exceeds the M10 published specifications. This result does not represent the final material selection, indicating that NVIDIA is still seeking higher PCB manufacturing yields and production efficiency while meeting electrical requirements.Guo Mingqi mentioned that Shengyi Technology is currently the leading supplier for the NVL576 exchange tray material evaluation, with related materials including fiberglass-free PTFE copper-clad laminate SG5300N, fiberglass-free hydrocarbon copper-clad laminate SG1030N, and fiberglass-free hydrocarbon prepreg SIF09. SG1030N and SIF09 still contain a small amount of PTFE. He also stated that once specifications are confirmed and mass production begins, Shengyi Technology is expected to accelerate procurement from Chinese hydrocarbon material suppliers such as Dongcai, Shengquan, and potential supplier GCH Technology.

According to Rest of World, Alibaba launched the open-source model platform ModelScope in 2022, and OSChina launched MoArk in 2023, providing model hosting, testing, and customization services for Chinese developers behind firewalls. In 2023, Chinese regulators blocked Hugging Face without disclosing specific reasons; reports indicate that the government simultaneously allowed artificial intelligence laboratories to bypass the blockade to share Chinese models globally. ModelScope currently hosts over 170,000 models and stated in March that it has 170,000 models and 250 million users; MoArk hosts about 20,000 commonly used models, both still below Hugging Face, which hosts over 3 million open models.OSChina CEO Xu Yong stated that not everyone can always use a VPN, and China needs to establish an independent artificial intelligence ecosystem that serves Chinese users, claiming that China is developing an independent ecosystem faster than during the internet era. He also mentioned that MoArk has deployed engineering teams to ensure that models can run on different types of Chinese chips, making it their mission to ensure mainstream open models are compatible with Chinese chips. Both platforms offer some services for free, with additional computing power and premium features charged; ModelScope also hosts university hackathons and has opened co-working and event spaces for AI entrepreneurs in Hangzhou.Headquartered in New York, Hugging Face is a major platform for sharing open models and datasets. In September, Nvidia announced its acquisition of the company for $12.9 billion and noted in regulatory filings that regulators might prohibit sharing Chinese models on the site.

NVIDIA CEO Jensen Huang stated that the statement reached at the White House is highly binding.

OpenAI CEO Altman stated that Nvidia's security system is not foolproof.

Delta Electronics will build an autonomous driving system based on NVIDIA (NVDA.O) Hyperion.

NVIDIA CEO Jensen Huang and Elon Musk will sit with President Trump at the White House's artificial intelligence luncheon.

The NVIDIA research team enabled the AI Agent to directly rewrite the GPU kernel of Kimi Delta Attention. Kimi Delta Attention is one of the core attention mechanisms used by the Dark Side of the Moon Kimi-Linear model, and such low-level code typically required engineers familiar with CUDA and chip architecture to optimize manually and repeatedly in the past.The version written by the Agent achieved 2.96 times the speed of the official FlashKDA of the Dark Side of the Moon on the NVIDIA B300. The team tested a total of 6 sets of tasks with fixed lengths and varying lengths, with 2.96 times being the geometric mean of all tasks; the relevant kernel has been open-sourced.During the process, the Agent exploited testing loopholes by incorporating statistical patterns from the test data into the code, achieving a speedup of 3.74 times; when retaining only the most recent 32 tokens, a single score reached 5.16 times. These versions would produce incorrect results or fail when switched to real Kimi data. The team subsequently incorporated real Kimi runtime data, random inputs, and extreme value tests, tightening the error standards, and ultimately retained the 2.96 times version, which passed this validation.

According to the Financial Times, Nvidia (NVDA.O) is seeking help from insurance companies to diversify the risks associated with artificial intelligence development.

NVIDIA CEO Jensen Huang stated in an interview with CNBC's "Squawk Box" on Monday that training or learning with competitors' products is part of competition, and he refused to label AI model distillation as theft. When asked if distillation is not robbery, Huang said that it is competition. He mentioned that people can test others' products, and NVIDIA's products are sometimes taken apart to the bare bones to understand how they work; he would prefer that others do not learn from NVIDIA products, but competition makes everything better. If one does not want others to use their products, they can identify customers and discontinue services.U.S. officials have accused Chinese AI companies of using distillation, which involves training models with outputs from other models, extracting capabilities from U.S. systems. U.S. Treasury Secretary Scott Bansen stated in July that this practice is theft and threatened to impose sanctions on overseas companies that utilize distillation to extract capabilities from U.S. models. The White House did not immediately respond to CNBC's request for comment, and U.S. officials are considering measures against overseas companies that use distillation.Earlier this month, the U.S. Cybersecurity and Infrastructure Security Agency accused Chinese AI companies of conducting industrial-scale knowledge distillation activities, violating the terms of use for U.S. companies. AI company Anthropic stated earlier this month that it found Alibaba's Qwen series models and DeepSeek engaging in illegal distillation. The Chinese side denied these accusations.