Some investors are getting nervous about how a downward turn in the economy could affect Nvidia’s capital spending

Nvidia CEO Jensen Huang made bullish comments about the AI opportunity at GTC this week.

Nvidia Corp.’s shares have not gotten a big anticipated lift during the company’s huge developer conference, despite the relatively bullish comments by Chief Executive Jensen Huang.

On Thursday, Nvidia shares rose about 1% after a zigzag trading pattern since Huang’s keynote on Tuesday at the company’s GTC. The conference, which ends Friday, has about 25,000 attendees and has taken over the area around the convention center in downtown San Jose.

Wall Street is currently of two points of view. There were no real surprises in Huang’s keynote, but his overall tenor about AI and how it is pervading every industry was seen as a positive. In addition, the company outlined a road map for the next three years, with major new products planned every year — an ambitious target for the semiconductor industry.

“Investors’ attitude toward the speech appeared to be somewhat lethargic,” Benchmark analyst Cody Acree said in a note to clients. “Overall, we thought the keynote was about as expected, with Jensen once again providing a masterclass overview of AI.” Acree reiterated his buy rating, adding that he did not hear anything that concerned him.

At the same time, some investors are getting nervous about how a downward turn in the economy could affect capital spending and the current plans by cloud service providers to continue to build out massive data centers to run AI models and applications. Fears of the lower-cost DeepSeek of China still linger, with recent data from the startup lab.

Several analysts reiterated buy ratings after Nvidia’s analyst meeting on Wednesday.

“While general sentiment around the AI trade has soured a bit in recent weeks, we saw nothing to suggest that the demand environment right now is anything but robust, and the company’s position appears to be only growing stronger with time,” Bernstein Research analyst Stacy Rasgon said in a note to clients. Rasgon has an outperform rating on Nvidia’s stock.

Vivek Arya of BofA reiterated his buy rating on Nvidia, noting that concerns among investors about a potential lower market size after DeepSeek’s lower cost computing model was unveiled were “overstated, as reasoning models require 100 times more compute,” he wrote, referring to Huang’s comments.

That said, there still are concerns among investors about the DeepSeek impact and in general about spending and tariffs. Huang told both analysts and press that he did not expect a significant impact from tariffs and was very enthusiastic about DeepSeek.

Dan Morgan, a senior portfolio manager at Synovus Trust Company, told MarketWatch in an email that investors have been concerned about “economic macro concerns, export control concerns, tariffs,  DeepSeek scares and concerns about the sustainability of scaling out frontier models,” but he believes Nvidia’s very strong product cycle should offset those concerns.

DA Davidson analyst Gil Luria, though, was not quite as sanguine.

“We remain a little more skeptical on NVIDIA’s positioning of market sizing as they lean increasingly more on reasoning model tokenomics to support the bull narrative,” he said, referring to the concept of token economics, in which AI cloud services are measured by the number of tokens consumed and generated, with pricing plans based on token input and output. A token refers to the tiny bits of data that come from breaking down larger chunks of information. AI models process tokens; the faster they are processed, the faster the models learn and respond.

Luria maintained his neutral rating but lowered his price target on Nvidia to $125 from $135.

“As we pointed out from DeepSeek’s OpenSourceWeek, they are serving roughly one third of OpenAI’s volume on a fraction of the infrastructure, at just 2224 H800s,” Luria wrote, referring to Nvidia’s older Hopper architecture. “While very few labs are extracting that kind of efficiency out of their systems, we’d note that DeepSeek[’s] open sourcing their algorithms and repositories leaves the door open for considerable efficiency gains from other frontier labs.”