Ginsen Huang, CEO of NVIDIA (NVDA), seized the fourth profit call of the company on Wednesday, looking to reaffirm the location of the chips giant in the artificial intelligence trade – Wall Street’s calm about the future technology growth.
The shares of NVIDIA were more than 7 % over the year in which the printing is being printed on Wednesday, as investors and analysts raised questions about the continued spending on artificial intelligence from the Big Tech.
Fears: The rise of Deepseek models of artificial intelligence means that developers did not need to use expensive chips such as Blackweell from NVIDIA- and that the custom chips developed by NVIDIA customers such as Amazon (Amazon) and Google (Googw, Googl) can threaten the company’s long-term health.
Huang stopped making opening notes during the NVIDIA invitation. Instead, he answered analysts’ questions all over and closed with comments that explain how models such as Deepseek require more power than previous models.
When Deepseek first appeared in the R-1 model in January, Amnesty International’s shares were sent to a tail.
This is because the company says it has developed the program, which is competing with the Openai platform, using the NVIDIA H20 chips. These processors are much less powerful than AI Titan’s Blackwele chips, prompting investors to ask whether NVIDIA is facing an existential crisis. After all, if companies are able to create Amnesty International platforms using more affordable chips, then why do you need to spend billions on high -end processors in NVIDIA?
As of 11:31:56 AM EST. The market is open.
Nvda AVGO Mrvl
However, Huang explained that since the Deepseek model, and others liked them, they made better responses when using the most powerful artificial intelligence chips, NVIDIA will continue to benefit from their use.
“Whenever the model believes the more intelligent answer. Models like Openai, Grok 3 and Deepseek-R1 are thinking models that apply the scaling time of reasoning.” “The 100x thinking models can consume more. Future thinking models can consume more account.”
Huang also explained that models such as Deepseek lead the demand for reasoning, and the process of operating artificial intelligence applications. Training of artificial intelligence models requires a huge amount of strength and performance. But when the inference becomes the main use of artificial intelligence systems, Wall Street investors have wondered whether NVIDIA customers would choose cheaper and less powerful chips.
However, Huang claims that Deepseek models, and those like them, show that inference will require a lot of strength in itself.
In addition to Deepseek processing, Huang also hit the ASICS effect on the industry – and what NVIDIA might mean. ASICS, ASICS integrated circuits, specially designed slices, as it may guess, specific applications. Google used a tensioner processing unit, ASIC, to train its Gueini AI platform.
Discover more from
Subscribe to get the latest posts sent to your email.