GPU

He will be able to continue selling his H20 GPU for AI in China, although he has cost him a 1 million dinner per diner

by

The GPU for applications of artificial intelligence (AI) H20 is the salvation of Nvidia in China. Since the sanctions package officialized by the administration of Joe Biden entered into force November 16, 2023 This is THE ONLY SOLUTION FOR IA That Jensen Huang’s company can sell to its Chinese clients. And, in addition, during the last months it is being a real success. This chip on paper is much less capable than the most sophisticated GPUs that Nvidia currently sells. In fact, this is the reason why the US Department of Commerce has allowed its sale in China in recent months. Its limitations invited us to initially assume that their reception in this Asian country would be warm, but It has not been so at all. According to Business Intelligence semiconductorsince this chip reached the Chinese market in mid -2024 its sales have grown 50% quarter to quarter, which positions it as the most successful Nvidia product today. However, sales of The H100 GPUwhich is more powerful, “only” grows 25% quarter after quarter. In spite of everything since the end of 2024, the H20 Chip is undergoing the scrutiny of the US Department of Commerce. Jensen Huang has added one of his most unlikely successes Gina Raimondo, the Secretary of Commerce during the mandate of Joe Biden, He made this warning to Nvidia In December 2023: “If redesign a chip so that it can be used for AI, we will control it the next day.” The direct allusion to the Jensen Huang company is evident. The return to the US government of Donald Trump and his collaborators far from appeasing the panorama promised to end once and for all for the sale of the H20 GPU in China. The Nvidia business in China has resentful during the last two years as a result of the sanctions imposed by the US For Nvidia this restriction would be a very hard blow. His business in this Asian country has rented during the last two years as a result of the sanctions imposed by the US, and stop selling the GPU that during the last months He has sustained this company in China I would make a difficult wound to heal. These are the circumstances in which Jensen Huang has met with Donald Trumpand has done so with a firm purpose: to ensure that the government allows Nvidia to continue selling its H20 chip in China. The Trade Department, which under Trump’s mandate is being led by Howard Lutnick, intended According to several filters to prevent this week that this GPU continue to arrive in the country of Xi Jinping. However, against all forecast the US administration has suspended, at least temporarily, its export prohibition of this chip. This conclusion is surprising, but there are even more circumstances in which Jensen Huang has achieved his goal. And it is that the general director of Nvidia has approached this negotiation directly with Donald Trump during a dinner at the restaurant of the Mar-A-Lago tourist complex housed in Palm Beach (Florida), which is owned by the latter. It has transcended that Huang and the other diners They have paid a million dollars each for attending this dinner. But Huang has compensated. Nvidia can continue selling its H20 GPU in China. At least for the moment. Although, yes, he has pledged to invest more money in data centers for the US. Image | Nvidia More information | Npr In Xataka | The Nvidia pulse and US administration becomes more virulent. The B20 GPUs for danger

The B300 GPU is the new Nvidia beast for Ia. And we already know what prepares for 2026 and 2027

by

Jensen Huang, the co -founder and general director of Nvidia, has not let out the opportunity to publicize the next GPU for artificial intelligence (AI) that have put their engineers ready in the framework of the GTC 2025 (GPU Technology Conference). The spectacular thing this electrical engineer has presented is The DGX B300 platform. This hardware is the most powerful Nvidia for generative, although according to this company it is also its most efficient proposal from an energy point of view. The Blackwell Ultra GPUs work on the B300 platform elbow with a 2.3 TB Map of HBM3E memory, delivering according to NVIDIA 72 Pflops in training processes with precision FP8 and nothing less than 144 pflops in inference tasks with precision FP4. These figures are a real monstrosity. In fact, the B300 platform is 11 times faster in inference and 4 times in training than its predecessor, the B200. This is the hardware with which Nvidia wants to maintain her leadership If we look at the consumer figures announced by NVIDIA we will see that apparently the energy efficiency of B200 and B300 platforms is similar. The first consumes approximately 14.3 KW maximumand the second one 14 kW. However, there is something that we should not overlook: the GPUs of both solutions have been implemented on the Blackwell microarchitecture, but they are not the same. The Blackwell Ultra chips of the B300 platform are more powerful than the Blackwell to dry infrastructure B200. The B300 platform integrates 50% more memory, allowing you to deal with larger AI models In addition, the B300 platform integrates 50% more memory, which in theory allows this hardware to deal with larger and more parameters. This proposal will reach the first data centers During the second semester of 2025. In any case, Nvidia has not only spoken of her current hardware in this edition of her conference dedicated to AI; He has also anticipated what his engineers are working for 2026 and 2027. The microarchitecture that will replace Blackwell is known as Rubin, and, as expected, it will be even more powerful than his predecessor. An interesting detail is that Rubin will be compatible with Blackwell at the infrastructure level, which will allow Nvidia customers to combine both solutions. In any case, Rubin will deliver 1.2 EXAFLOPS in training processes with precision FP8 compared to 0.36 EXAFLOPS of the B300 platform. It will arrive during the second half of 2026. And during the second semester of 2027 Nvidia will launch Rubin Ultra, a review that according to this company will reach 5 exaflops In training tasks with FP8 precision, so your performance in this scenario will be almost four times greater than Rubin’s. A last interesting note: Rubin will use HBM4 memory, while Rubin Ultra will have HBM4E. Image | Nvidia More information | Nvidia In Xataka | AI is already our best ally to solve the mathematical problems that seem impossible

The Singapore government has revealed which companies are involved in the delivery of GPU from Nvidia A Deepseek

by

Depseek continues to be the artificial intelligence (AI) of the moment five weeks after its irruption. And is that the debate about the hardware used by this Chinese company to train your AI model Keep on the table. High-flyer, your parent company, is A quantitative coverage fund specialized in trading algorithmic. This simply means that this company uses advanced mathematical models and computational algorithms to address investment decisions with the greatest possible success guarantees. Deepseek was born as a high-flyer secondary project to take advantage of its computer resources and “put one foot” in the AI ​​industry. Its creators say that in the training of their model they have used only 2,048 chips H800 of Nvidia. However, Some analysts defend that, in reality, its infrastructure brings together 50,000 GPU H100 bought through intermediaries. This is the problem. High-flyer could legally buy the H800 chips to the entry into force of US sanctions of November 16, 2023, but the H100 GPUs should not be in their possession. Singapore is the entrance door to China of the chips for the most advanced Nvidia The US government has suspected for many months that Chinese companies and research centers dedicated to AI acquire the most advanced NVIDIA GPUs through Singapore and Malaysian intermediary companies. This possibility is no longer just a hypothesis. And it is that the Singapore government has confirmed that it has identified those responsible for diverting to China, and presumably towards the Deepseek parent company, servers that contain the high -performance GPUs produced by NVIDIA. The US now has the opportunity to tighten its fence a little more about China This information was revealed last week by the television channel Channel News Asia, and today the Minister of Internal Affairs and Justice of Singapore, K. Shanmugam, He has confirmed it. Interestingly, it has not specified what the GPUs that incorporate these machines, but it has made public a very important fact: The name of the companies They have manufactured the servers. And they are two very important Nvidia customers: Dell Technologies and Super Micro Computer. If it is finally confirmed that Depseek, or any other Chinese company that is dedicated to AI, is getting the Nvidia avant -garde GPUs acquiring servers of these companies in Singapore or Malaysia, USA will have the opportunity to tighten your fence a little more. However, this would not demonstrate the guilt of Dell and Super Micro, although its indirect involvement in the traffic of the Nvidia chips. This circumstance would put on the table the need to control with more precision where their servers will stop, something that, on the other hand, It is not easy. Whatever the Singapore government claims to be willing to collaborate with his American counterpart to end the illegal vanguard GPU traffic. Image | Nvidia More information | Reuters In Xataka | We can forget an AI without hallucinations for now. The general director of Nvidia explains why

Openai is finishing designing its own GPU for Ia. And we already know what agreement has arrived with TSMC

by

Sam Altman and the rest of OpenAi’s directive dome are determined to stop using GPUs in the medium term artificial intelligence (AI) of Nvidia. We know it with certainty since January 2024. On that date Altman began a journey that pursued find investors with the necessary muscle To help your company Develop your own chip for AI. And, apparently, he had a good reason to do it that goes beyond reducing his dependence on Nvidia hardware. Just a few weeks earlier, in December 2023, Pat Gelsinger, the former general director of Intel, declared that the AI ​​industry is determined to leave behind CUDA (Compute Unified Device Architecture). This technology brings together the compiler and development tools used by programmers to develop their software for NVIDIA GPUs, and replace it with another option in the projects that are already underway it is a problem. “The entire industry is determined to eliminate market CUDA (…) We see it as a shallow and small pit, so we are motivated to propose a broader set of technologies both to address training and innovation or science of data “, Gelsinger defended During the event “Ai Everywhere” held in New York. In addition, he assured that Google and OpenAi are two of the companies with a great specific weight in the AI ​​industry that they want to leave CUDA behind. TSMC is the ideal ally for Openai We do not say it. It is evident that Sam Altman believes it if we stick to the steps he has taken during the last months. A little over a year ago he began his conversations with TSMC, which is the largest semiconductor manufacturer on the planet with A market share close to 60%. Altman needed to explore the possibility that this Taiwanese company manufactured its GPU for ia. After all, TSMC produces the chips designed by NVIDIA or AMD, among other companies, for this scenario of use. TSMC will manufacture the chips for ia designed by Openai in its 3 nm node The negotiation that they had already culminated successfully. TSMC will produce the GPUs for AI designed by OpenAi. But this is not the only thing we know. According to SCMP These chips will be manufactured in the 3 NM node of TSMC, which is currently Its most advanced integration technology (In 2025 it will begin producing large -scale integrated circuits In the 2 Nm node). And, in addition, Openai has already started the final stage of design of its own GPU for AI, according to Reuters. This last information fits with the date on which the company led by Sam Altman began presumably. The two media that I just mentioned argue that during the next months OpenAI will send the preliminary design of its GPU to TSMC with the purpose of starting the first validation and production tests. This project phase is known in English as tape-out. At the moment neither OpenAi or TSMC have made official statements about the advances of their collaboration, but all the information in which we have just inquired is consistent enough to give it for good. After all, according to these sources, the Taiwanese chips manufacturer will begin large -scale production of the OpenAI GPU for 2026. Image | TSMC More information | SCMP | Reuters In Xataka | Some researchers claim to have created an AI as good as those of Openai and Deepseek for $ 50. And the data is real

Log In

Forgot password?

Forgot password?

Enter your account data and we will send you a link to reset your password.

Your password reset link appears to be invalid or expired.

Log in

Privacy Policy

Add to Collection

No Collections

Here you'll find all collections you've created before.