Nvidia

Sanja Fidler, NVIDIA's vice president of AI research, and her AI lab team in Toronto released the LATTE3D model and demonstrated it on site at the GTC2024 Spring Forum. The technology can efficiently generate 3D models from text prompts. When paired with NVIDIA's RTX A6000 GPU, it can generate 3D models almost instantly (in less than 1 second). The 3D shapes can be enlarged and rotated at will, with no blind spots at 360 degrees. "A year ago, it took an AI model 1 hour to generate 3D visual effects of this quality, and the current...

According to media reports, NVIDIA recently announced that it will work with Indosat Ooredoo Hutchison, Indonesia's second largest mobile telecommunications company, to build an artificial intelligence development center in Solo with a total investment of up to US$200 million. This move marks NVIDIA's further expansion of its business in Indonesia and even the entire Southeast Asia. NVIDIA's layout in Southeast Asia is not limited to Indonesia. In January this year, Singapore Telecommunications also announced a partnership with NVIDIA to deploy artificial intelligence functions in its data centers throughout Southeast Asia. This cooperation will further promote the application of artificial intelligence technology in Southeast Asia...

The US continues to escalate its export restrictions on China, and Nvidia's AI chips H20 and RTX 4090D, which are specifically for the Chinese market, may be subject to new restrictions. The new rules require that the export of chips that exceed certain performance (comprehensive floating point computing power exceeding 70 TFLOPS) must apply for a license, which may lead to restrictions on the export of Nvidia's new chips to China. (Core Intelligence)

NVIDIA launches NeMo early access program, providing Curator, Customizer and Evaluator microservices to simplify generative AI development; NeMo platform supports data collation, custom fine-tuning and performance evaluation, aiming to help developers build enterprise-level custom generative AI; provides GPU-accelerated data collation, efficient fine-tuning technology and automatic evaluation tools to accelerate the listing of enterprise solutions. (AIGC Open Community)

Nvidia CEO Jensen Huang released the Blackwell GPU and the complete suite of the Blackwell platform in San Jose. He launched multiple enterprise-level generative AI microservices. He announced that TSMC and Synopsys will launch the breakthrough lithography computing platform cuLitho and put it into production. In the field of humanoid robots, Nvidia released the humanoid robot basic model Project GR00T and the new humanoid robot computer Jetson Thor, which significantly upgraded the Isaac robot platform and promoted the development of embodied intelligence. In addition, Jensen Huang…

According to Israeli news outlet Calcalist, Nvidia is in advanced talks to acquire an Israeli AI startup, Run.ai. The five-year-old startup has raised nearly $120 million. The company's technology helps application developers use Nvidia's AI server chips more efficiently. Calcalist reported that the transaction price could range from hundreds of millions of dollars to $1 billion.

Nvidia CEO Huang Renxun announced the launch of a new generation of GPU Blackwell at GTC. The first Blackwell chip is called GB200 and will be available later this year. Blackwell has 208 billion transistors and uses TSMC's 4nm process. The previous generation GPU "Hopper" H100 uses a 4nm process and integrates 80 billion transistors. Huang Renxun said: "Hopper is great, but we need a bigger GPU. Blackwell is not a chip, it is the name of a platform." Nvidia said that the base...

Nvidia held the GTC 2024 conference early this morning. At the conference, Nvidia CEO Jensen Huang introduced the new GPU Blackwell B200 and the "super chip" GB200. It is reported that the new B200 GPU consists of 208 billion transistors, while the GB200 contains two B200 GPUs and a Grace CPU. In the benchmark test of the GPT-3 large language model with 175 billion parameters, the performance of GB200 is 7 times that of its predecessor H100…

Nvidia's GTC conference is expected to showcase its most powerful AI GPU to date, the B100 Blackwell. This new processor uses a 3-nanometer process and is expected to significantly improve performance. CEO Huang Renxun's keynote speech will focus on the revolutionary moment of artificial intelligence and share the latest progress of Nvidia's accelerated computing platform. The conference will also discuss Nvidia's sovereign AI plan and will be participated by many industry leaders. (Tencent Technology)

NVIDIA will launch a new Generative AI professional certification to help developers demonstrate their technical prowess in this important field. It will be officially released at the GTC conference held from March 18 to 21, and relevant training courses will be provided on site and online. The new professional certification program includes two entry-level (Associate-level) Generative AI certifications, focusing on the mastery of large language models and multimodal workflow skills. Related course address: https://www.nvidia.com/en-us/learn/cert…

Bloomberg analyst Beth Kindig recently released a forecast report, believing that Nvidia will continue to dominate the AI field, and estimated that its revenue in 2026 will exceed US$130 billion (currently about RMB 934.7 billion), which is five times that of fiscal year 2023 (US$26 billion). Nvidia has an absolute dominant position in AI GPUs. Its current mainstream product is the H100 AI GPU. It is estimated that the H200 AI GPU and the next-generation Blackwell B100 AI GPU will be released this year. Market research institutions such as Omdia are also very optimistic about Nvidia’s...

According to NVIDIA's official microblog, NVIDIA will officially launch Generative AI Professional Certification at the GTC conference, which will enable developers, professionals and other relevant personnel to prove and demonstrate their skills and professional strength in generative AI. The newly launched professional certification program currently includes two entry-level (associate-level) Generative AI certifications, which focus on the mastery of large language models and multimodal workflow skills. The GTC2024 conference will be held in San Jose, California, USA from March 18 to 21. This is the first time that the GTC conference has returned to an offline event in 5 years. (…

Due to the decline in demand for AI chips, Nvidia's GPU shortage is improving, and more than half a dozen companies have said they have reduced their orders for Nvidia AI chips. At the same time, early buyers of large quantities of GPUs have begun to sell some of their inventory. Nvidia CEO Jensen Huang said that the supply problem is improving, but GPUs will still face a shortage of supply until 2025. Rival Intel CEO Henry Kissinger predicted that many companies will invest more cautiously in AI chips this year. Databricks co-founder and CEO Ari Golds predicts that GPU prices will drop significantly next year. (Titanium Media)

On March 8, local time, Nvidia fell more than 5% during the day, and rose 5% in the morning, the largest single-day drop since May 2023. As of the close of March 8, local time, Nvidia reported $875.28, down 5.55%, with a market value of $2.2 trillion. Its market value evaporated by $128.5 billion (about RMB 923.4 billion) overnight. Nvidia's stock price has soared this year as market demand for its chips for AI computing will remain strong. Even after the decline, the stock still rose in 2024…

On March 7, as Nvidia's stock price soared 79% this year, more directors of the company are choosing to cash out. Regulatory documents submitted on Wednesday showed that Nvidia director Tench Coxe sold 200,000 shares at a price of $850.03 to $852.50 per share on March 5, cashing out a total of about $170 million. Nvidia rose 3.18% on Wednesday, closing at $887, setting a new record high, rising for the fifth consecutive day, and its current market value is $2.2 trillion.

On March 4, server manufacturer Dell revealed Nvidia's upcoming artificial intelligence (AI) GPU, codenamed Blackwell. These chips consume up to 1000W of power, an increase of 40% over the previous generation of chips. Dell COO Jeff Clarke said that Dell will also provide the flagship PowerEdge XE9680 rack server to coincide with the launch of the B200 server GPU. In addition, the Blackwell architecture B100 and B200 will not be launched until 2025 at the earliest, which is different from the previously expected launch...

According to Chinanews.com, on Monday (4th), the US Eastern Time, the continued upward momentum of US stocks paused, and the three major stock indexes fell collectively. As of the close, the Dow fell 0.25%, the Nasdaq fell 0.41%, and the S&P 500 fell 0.12%. On the news, Atlanta Fed President Bostic expects the Fed to cut interest rates for the first time in the third quarter, and then keep it on hold to assess the impact of the policy shift on the economy. In another commentary published on the Atlanta Federal Reserve Bank website, Bostic said he was worried that businesses were too active and that the large amount of new demand released after the rate cut could exacerbate price pressures. On the disk,…

On March 3, Nvidia is reportedly developing its own PC game handheld and considering expanding to launch graphics chips with exclusive IP (intellectual property). Game handhelds are currently developing rapidly, and it is very reasonable for Nvidia to layout in this field. Current game handhelds have one thing in common, that is, they use AMD's APU, and Intel's competing products will be launched later in 2024. Nvidia hopes to expand the scope of application of chips and may even cooperate with Intel to create a high-end game handheld: "This can circumvent many limitations of the x86 architecture, and can set up appropriate heat sinks for the CPU and GPU...

Nvidia CEO Jensen Huang said on Friday that general artificial intelligence (AGI) could be available in as little as five years, depending on some definitions, Reuters reported. Huang answered a question at an economic forum hosted by Stanford University about how long it would take to achieve one of Silicon Valley's long-standing goals - creating computers that can think like humans. He said the answer depends largely on how that goal is defined. If it is defined as the ability to pass a human test, he said, artificial general intelligence (AGI) will be here soon. As of now, artificial intelligence can be achieved through...

On March 3, at an economic forum held at Stanford University, Nvidia CEO Huang Renxun said in response to the question "How long will it take to create a computer that can think like a human?" that the answer depends largely on how the goal is defined. If the definition is to be able to pass human tests, then general artificial intelligence (AGI) will come soon. "If I give artificial intelligence... all the tests you can imagine and put it in front of the computer science industry," Huang Renxun said, "I guess in five years, we will get good scores on every test." But Huang Renxun...

Original link