The NVIDIA (NVDA) bulls continue to defend their favorite actions in the world in front of the reports of the flea manufacturer next week.
Despite Deepseek’s challenge to think about AI costs, some analysts remain optimistic on Nvidia. However, this comes with a few notes of prudence – in particular, about a first quarter perspective which does not meet the high expectations of the market.
“The short -term dynamics are fluid (quarter from January to July quarter) … while in the long term (October quarter and forward), our work suggest that it seems frankly spectacular,” wrote the analyst of Loop Capital Ananda Baruah in a customer note.
Baruah reiterated a price target of $ 175 on Nvidia, which presupposes 25% of the increase in current levels.
Addition of Baruah ” [to] … NVIDIA GPU reaching 10 million to 12 million while hyperscalers seek to increase their percentage of non-CPU calculation to 50% more in the coming years (compared to around 10% currently). Do not forget … For Nvidia, the story is accelerated calculation + ge ai, which means that it is faced with two calculation possibilities of 1.0 billion of dollars in front of this in the coming years, each of which is just at the beginning. “”
NVIDIA shares have increased less than 1% in the negotiations prior to $ 140 each.
While Nvidia’s actions have rallied to around 23% compared to levels of more than February, the feeling of the fundamental principles of the company has become more mixed.
Podcast: How Bill Gates considers Nvidia
Evercore analyst, Mark Lipacis, wrote in a recent note that there are three reasons for the more prudent tone: 1) the decline in deep AI demand in the total, 2) the cycles of Calculation of change of Deepseek’s in -depth AI far from the NVIDIA and ASICS GPU [custom chips]and 3) Blackwell chip delay.
Deepseek, based in China, surprised the markets at the end of January after revealing RI, its AI model which gives a chatgpt-esque performance at a cheaper price. RI cost $ 5.6 million to build a basic model, compared to hundreds of millions of dollars incurred in American companies such as Optai and Anthropic.
The fears have instantly set up that American companies spend too much in AI infrastructure, which includes Nvidia chips.
“The conventional wisdom of last year was that the formation of incredible models was going to be possible only for a handful of companies,” said Snowflake (snow) CEO Sridhar Ramaswamy on the opening podcast from Yahoo Finance. “What Deepseek has done in recent weeks is breaking this belief by saying that they can form a model for $ 6 million.”
