(Reuters) -Nvidia beat quarterly sales expectations as customers stockpiled its AI chips before fresh U.S. curbs on China exports took effect, but the same restrictions will slice off $8 billion in sales from the company's current quarter, forcing the company to offer a forecast below Wall Street estimates on Wednesday.
Shares of the world's most valuable semiconductor firm still rose 5% in extended trading as investors digested news that the hit in the current fiscal second quarter was not as bad as feared, and Nvidia talked up demand for its new Blackwell chips from customers including Microsoft. The stock is relatively flat so far this year, compared with 2024 when the shares nearly tripled in value. Nvidia now faces trade restrictions on what it can sell, and the AI data center market is also maturing.
Washington's years-long efforts to thwart Beijing's access to top-of-the-line U.S. technology have resulted in stricter restrictions on the export of Nvidia's AI chips - stifling the company's access to one of the largest markets for semiconductors.
Midway through a conference call with analysts, CEO Jensen Huang made impassioned remarks about U.S.-China policy, saying that Nvidia was at risk of being cut off from China's massive AI developer base and arguing that China's chip industry was sophisticated and closing in on the United States' dominance. But he praised U.S. President Donald Trump's recent move to rescind a so-called AI diffusion rule that would have regulated global flows of U.S. AI chips.
"President Trump wants America to win. And he also realizes that we're not the only country in the race," Huang said.
Huang told analysts that Nvidia's Hopper chips could no longer be modified for the Chinese market but did not comment on its Blackwell chips. Reuters has reported that Nvidia is preparing a Blackwell variant for the Chinese market.
Though unlikely to make up for the loss in Chinese revenue, a spate of new deals that Nvidia signed earlier this month in the Middle East could offer fresh avenues of growth - including the first phases of a 10-square-mile data center site in the United Arab Emirates that could eventually use 5 gigawatts' worth of AI infrastructure. The company has also announced similar deals in Saudi Arabia and Taiwan.
"We have a line of sight to projects requiring tens of gigawatts of Nvidia AI infrastructure in the not-too-distant future," Nvidia Chief Financial Officer Colette Kress said on the conference call.
But in the shorter term, restrictions on China exports will hurt. Kress said data center revenue in that country declined.
U.S. restrictions on the sale of Nvidia's H20 chips to China, the only AI processors it could legally export to the country, prompted Nvidia to disclose in April that it expected a $5.5 billion charge, while Huang had in May pegged the revenue impact related to the restrictions at about $15 billion. On Wednesday, Nvidia said the actual first-quarter charge due to the H20 restrictions was $1 billion less than expected because it was able to reuse some materials. It said it lost $2.5 billion in H20 sales in the first quarter and expected to miss $8 billion in the second quarter.
However, Nvidia also said the H20 brought in $4.6 billion in sales in the first quarter and that China accounted for 12.5% of overall revenue in the first quarter.
Gil Luria, an analyst with D.A. Davidson, said the overall impact of the H20 restrictions was less than feared.
"There was a removal of some China revenue from the July quarter guides, but there was also China revenue that was pulled into the first quarter. Chinese buyers were stocking up on H20 ahead of the restrictions, which is what propped up the April quarter," Luria said.
Though major cloud companies such as Microsoft and Alphabet have stood their ground on the billions they have earmarked this year for spending on expanding infrastructure for AI data centers, worries about such spending persist amid rapidly changing global trade policies.
On an adjusted basis, Nvidia earned 81 cents per share in the first quarter. Analyst estimates varied widely as Wall Street tried to assess the impact of restrictions on some of Nvidia's chip sales to China.
Excluding the charges, first-quarter adjusted earnings per share would have been 96 cents.
According to data compiled by LSEG, the estimate for the company's adjusted quarterly earnings was 93 cents per share, with 17 analysts providing estimates after April 15 when Nvidia said H20 shipments would require additional licenses.
Nvidia's data center segment revenue was $39.1 billion in the first quarter, compared with analyst estimates of $39.3 billion, according to LSEG data.
The company said it has $29.8 billion in commitments to have its products manufactured, an increase from the year before but down quarter-over-quarter.
Nvidia, a bellwether of the artificial-intelligence market, expects revenue of $45 billion, plus or minus 2%, in the second quarter, compared with analysts' average estimate of $45.90 billion, according to data compiled by LSEG.
The forecast includes a loss in H20 revenue of about $8 billion due to the recent export limitations.
"The broader concern is that trade tensions and potential tariff impacts on data center expansion could create headwinds for AI chip demand in upcoming quarters," said Emarketer analyst Jacob Bourne. "This doesn't signal an end to Nvidia's dominance, but highlights that sustaining it will require navigating an increasingly complex landscape of geopolitical, competitive, and economic challenges."
(Reporting by Arsheeya Bajwa in Bengaluru and Stephen Nellis in San FranciscoWriting by Sayantani GhoshEditing by Shounak Dasgupta and Matthew Lewis)
