The U.S. has approved sales of advanced Nvidia chips to Saudi Arabia’s HUMAIN and the United Arab Emirates’ G42, authorizing the state-backed firms to buy up to 35,000 chips, worth an estimated $1 billion.

The approval of these chip exports marks a major reversal for the U.S., which had previously balked at the idea of direct exports to state-backed AI companies in the Gulf. Export controls were put into place to avoid advanced American technology making its way to China through the back door of Gulf Arab states.  

Before former President Joe Biden left office in January, he administered a final round of export restrictions on advanced AI chips, targeting companies like Nvidia, in a sweeping effort to keep that cutting-edge U.S. intellectual property out of China’s reach.

Now, President Donald Trump is moving to expand the reach of such advanced technology in order to “promote continued American AI dominance and global technological leadership,” the U.S. Commerce Department said in a statement published on Wednesday. 

The U.S. Commerce Department approved the chip exports, with the condition the state-backed AI outfits agree to “rigorous security and reporting requirements,” overseen by the Department of Commerce’s Bureau of Industry and Security.

The export approval follows Saudi Crown Prince Mohammed bin Salman’s trip to Washington this week where the Kingdom pledged to spend $1 trillion in the U.S., up from $600 billion originally committed during Trump’s Gulf tour in May.

“Even if we don’t get to that, both sides have skin in the game,” Afshin Molavi, senior fellow at the Foreign Policy Institute of the Johns Hopkins University School of Advanced International Studies, told CNBC’s Dan Murphy.

Saudi Arabia’s AI company HUMAIN, backed by its nearly $1 trillion Public Investment Fund signed a long list of partnerships with Adobe, Qualcomm, AMD, Cisco, GlobalAI, Groq, Luma, and xAI at a U.S.-Saudi Investment Forum held in Washington, D.C this week. Notably, HUMAIN will be teaming up with Elon Musk’s xAI to build a 500 megawatt data center in the Kingdom.

“What we want to do in 2026 is to build the capacity equivalent to what Saudi has built in the last 20 years, in one year,” Tareq Amin, CEO of HUMAIN, said at the summit. HUMAIN is hoping to position Saudi Arabia as the third biggest global AI hub, after the likes of the U.S. and China.

Just two weeks ago, Microsoft secured an export license for advanced chips to the UAE. Microsoft’s key partner in the UAE is G42, but the local AI company was notably absent from the Microsoft announcement, until today.

Nvidia on Wednesday reported fiscal third-quarter earnings that beat expectations, and provided a strong forecast for the current quarter.

Wall Street welcomed the report, and Nvidia stock rose after the release and during the conference call. Other stocks in the so-called artificial intelligence trade also saw a boost.

A closer look at Nvidia’s report shows that it continues to dominate the market for AI chips called GPUs, and CEO Jensen Huang sounded confident in the company’s products and bullish on the company’s outlook during a call with analysts.

Nvidia said it expects about $65 billion in sales in the current quarter, which ends in late January. That would be 65% growth on an annual basis.

Here are three key takeaways from Nvidia’s earnings:

On Wednesday’s earnings call with analysts, Huang began his comments by rejecting the premise of an “AI bubble” held by some investors who are concerned about the billions of dollars being spent on Nvidia chips and potential return on investment.

“There’s been a lot of talk about an AI bubble,” Huang said. “From our vantage point, we see something very different.”

Huang said there were three different kinds of uses for AI that are currently growing, and that all three are contributing to the boom in infrastructure investments.

He said that non-AI software, like for data processing, was increasingly being run on the company’s GPUs, that AI will create new kinds of apps, and that “agentic AI” which doesn’t need user input, will require additional computing power.

Huang said that people will soon start appreciating what’s happening underneath the surface of the AI boom, versus “the simplistic view of what’s happening to CapEx and investment.”

Bernstein analysts said in a note that Huang’s comments helped settle investor fears of a bubble after a recent pullback in AI names, saying “perhaps the AI trade is not yet dead after all.”

“More than just good numbers, we believe investors needed some hand-holding from Jensen which he provided in spades,” the analysts wrote.

Nvidia said on Wednesday that its order backlog didn’t even include a few recent announcements, like the company’s deal with Anthropic or the expansion of a deal with Saudi Arabia this week.

“The number will grow,” CFO Colette Kress said on the call, saying the company was on track to hit the forecast. “We’ll probably be taking more orders.”

“We see the opportunity to grow for quite some time,” Huang said.

Several analyst notes on Thursday drew attention to the $500 billion forecast and the addition of the recently announced deals.

Jefferies said Nvidia “answered the bell” in its earnings report and said the numbers should help steady the AI trade into the end of the year.

“We don’t expect every AI bear to be satisfied, but these results and added context from management around demand outlook should offer some near-term reprieve,” the analysts wrote.

The company eventually got the licenses this summer after Huang personally met with President Donald Trump and struck a deal to give the U.S. government 15% of China sales.

But it turns out that the sales of H20 chips during the quarter was “insignificant.” Kress told analysts that the company recorded $50 million in H20 sales during the period.

“Sizable purchase orders never materialized in the quarter due to geopolitical issues and the increasingly competitive market in China,” Kress said.

Nvidia has argued that the U.S. government should allow exports of the most advanced chips because it’s better for national security if Chinese developers get used to Nvidia technology, rather than being forced to use Chinese chips and make them better.

The H20 is old technology, but Nvidia wants to gain approval to send a version of its current-generation Blackwell chip in China.

“While we were disappointed in the current state that prevents us from shipping more competitive data center compute products to China, we are committed to continued engagement with the US and China governments and will continue to advocate for America’s ability to compete around the world,” Kress said.

Analysts at Melius said Thursday that the lack of China sales made the numbers “all the more extraordinary” and projected Nvidia would generate nearly $400 billion in free cash flow over the next nine quarters.

“Currently Nvidia isn’t delivering to China and we are not counting on this situation to get straightened out,” the firm said.

CNBC’s Sam Subin contributed to this story.

Alphabet’s Waymo on Thursday announced that it will soon begin manually driving its robotaxi vehicles in Minneapolis, Tampa and New Orleans.

The Google sister company will start operating test drives in that trio of towns with human drivers in hopes of launching its driverless robotaxi service there as soon as next year, the company said.

If Waymo does begin operating in those markets next year, that would bring the robotaxi company’s list of 2026 planned expansions to 15 cities.

On Tuesday, Waymo said it plans to start operating its vehicles with no human driver in Dallas, Houston, San Antonio, Miami and Orlando in the coming weeks, with plans to open service to the public there next year. The company has also previously announced plans to expand to Detroit, Denver, Las Vegas, Nashville, San Diego, Washington, D.C., and London in 2026.

A spokesperson said Waymo will wait until its technology is validated in Minneapolis, Tampa and New Orleans before committing to 2026 service launches.

“2026 is very much on the table, but we’ll be led by our safety framework,” Waymo spokesperson Ethan Teicher said in an email. 

With more than 250,000 weekly paid trips, Waymo’s robotaxi service currently operates in Austin, the San Francisco Bay Area, Phoenix, Atlanta and Los Angeles markets. The company has provided more than 10 million paid rides since launching in 2020.

Last week, Waymo began offering freeway routes in the San Francisco, Phoenix and Los Angeles markets. The company said it will gradually extend freeway trips to more riders and locations over time.

The addition of freeway rides marked an important milestone for Waymo and the robotaxi industry due to the challenges conditions of operating at such high speeds. Next year, Waymo will set its sights on achieving another key milestone: operating in markets known for harsh winter conditions.

Along with Denver and Detroit, the addition of Minneapolis means Waymo believes its nearly ready to begin serving riders in regions where its driverless vehicles would need to be ready to brave snow and frigid forecasts.

“We currently operate at freezing temperatures, including with frost and hail, and we’re validating our system to navigate harsher weather conditions,” Teicher said. “We’ll have small fleets to start that we expand over time.”

This week, Amazon-owned Zoox began allowing select San Francisco users to hail its driverless vehicles. San Francisco is the second market where Zoox now offers a free service, after its launch in Las Vegas in September. The company plans to remove its rider waitlist for San Francisco entirely in 2026.

WATCH: Waymo launches paid robotaxi rides on freeways