Nafidia In the first quarter of the financial strong financial Profits Wednesday.
Wall Street was pleased with the continuous growth of NVIDIA, which reached 69 % during the quarter. The company’s data center department continues to rise with companies, countries and cloud services providers canceling NVIDIA graphics processing units, or GPUS, l artificial intelligence programming.
“The team continues to maintain a step from a step from 1 to 1 competitors through its silicone/hardware/software and a strong ecosystem, and the team is reviewing itself by stopping it to the release of new products and more segmentation of products over time,” said Harlan Sur, an analyst at JPMorgan:
Here are three fast food from the company Profits:
China can be a $ 50 billion market for NVIDIA, but US export controls are in its way
NVIDIA expects to sell about $ 45 billion of chips during the July quarterly, and it was revealed on Wednesday, but this is missing in sales of about $ 8 billion that the company had recorded if not to restrict exports of H20 chips without a license.
NVIDIA also said it missed sales of $ 2.5 billion in April quarterly thanks to the export restrictions on H20.
The CEO of Nvidia Jensen Huang said in calling the company’s profits that China Like a $ 50 billion market that was actually closed to NVIDIA.
He also said that export controls were misleading, and will only encourage Chinese artificial intelligence developers to use local chips, instead of making an American platform to choose artificial intelligence programs.
“The United States has relied on its policy on the assumption that China could not make artificial intelligence chips. This assumption was always doubtful, and now this has become a clear mistake,” Huang said.
He said that export controls were leading artificial intelligence talents to use chips from local Chinese competitors, such as Huawei.
“We want to prefer every developer in the American world Technology chimneysHuang told CNBC’s Jim Kramer Wednesday night.
NVIDIA said she hadn’t had a ready -made China alternative chip, but she was considering “interesting products” options that could be sold on the market.
Black -down balance in Blackwell in the company some concerns about the influence of China.
“NVIDIA puts fully digestive concerns for rest, which indicates the acceleration of business unlike the Chinese opposite winds about the growth drivers who seem solid. Everything must be improved from here,” said Morgan Stanley analyst Joseph Moore.
Cloud service providers are still the most important NVIDIA customers
Nvidia says it has many clients ranging from sovereign countries to universities to institutions that want to search for artificial intelligence.
But he emphasized again on Wednesday that cloud service providers – such as companies like Microsoft I visit, Google clouds, Oracle Cloud infrastructure and Amazon Web services – still constitute about half of the data center revenues, which have been sales of $ 39.1 billion during this quarter.
These companies tend to buy the fastest and latest NVIDIA chips, including Blackweell, which includes 70 % of NVIDIA data center sales during this quarter.
The company said that Microsoft has already published “tens of thousands” of BlackWell graphics processing units, “100 trillion symbols” in the first quarter. The symbols are a measure of artificial intelligence output.
It will be the first place to get Blackwell Ultra, which is an updated version of the slide with an additional memory and performance. Nafidia said that the shipments of these systems will begin during the current quarter.
“The general expectations and the environment in general seem very encouraging,” said Stacy Rachton from Bernstein.
“Amids a quarter of a messy, Nafidia is very well corresponding to,” he said.
Look forward: Blackwell and AI inference
Over the past few years, many NVIDIA graphics processing units have been used for a dense resource process called training, as data is processed through the artificial intelligence model to gain new capabilities.
Now, Huang talks about the possibility of taking graphics processing units in NVIDIA to serve artificial intelligence models for millions of customers, a process called inference in industry. He said in the profit call that comes in terms of new increasing demand.
“In general, we believe that the leadership of technology in NVDA is still strong, as Blackwell’s shipments that benefit from the massive growth in thinking about artificial intelligence and achieving savings size,” said Ross Simmour from Deutsche Bank.
Huang says that the latest artificial intelligence models need to create more symbols – or create more output – in order to do “think”, which improves the answers of artificial intelligence. Of course, the latest Blackwell chips from NVIDIA for this purpose.
“We are witnessing a sharp leap in the demand for conclusion,” Huang said. “Openai, Microsoft and Google are a functional leap in the symbolic generation.”
Huang compared modern artificial intelligence models to the “one shot” approach that ChatGPT used when he first appeared in 2022, and he said that the new models need “a hundred, a thousand times more” computing.
“He is essentially thinking about himself, and leads to a step -by -step problem,” Huang said. “Planning may be multiple paths to answer. Tools can be used, read pdf files, web pages, watch videos, then produce a result.”
Reward: Fears of Jensen
Huang hit a more clear tone during the invitation, with a great focus on the impact of export controls instead of usual preaching about the capabilities of the global changes of Amnesty International.
He spoke at least about the call about the restrictions of American chips, and clearly stated the amount of influence that the borders have on current and future works.
“The artificial intelligence race is not only related to the chips,” he said. “It comes to the stack that the world is working on. With the growth of this stack to include 6G and Quantum, the leadership of the American global infrastructure is at stake.”
– Christina Pancysinsvilus in CNBC contributed to this article.
Correction: Stacy, an analyst at Bernstein. A previous version made mistakes in his name errors.
https://image.cnbcfm.com/api/v1/image/107423288-17174221132024-06-02t142115z_931346104_rc2038asm6yp_rtrmadp_0_nvidia-taiwan.jpeg?v=1717422187&w=1920&h=1080
Source link