What's up: Nvidia profits soar; U.S. approves deal to sell AI chips to Middle East; Trump weighs executive order targeting state AI laws

Good morning. Nvidia did more than beat investor expectations for the latest quarter. It also reframed the understanding of where the AI economy is headed for now.

AI companies have been under pressure in the stock market given concerns that their valuations and business models were unsustainable and that revenue won’t support trillion-dollar-level capital spending on data infrastructure. Nvidia’s results have raised expectations for AI. Risks persist, but the idea that an AI bubble is about to burst doesn’t have the same ring that it had prior to Nvidia’s latest numbers.

Anxiety over an AI bubble is understandable, given the magnitude of spending and the incredible valuations. It’s unprecedented, but less mysterious in the context of a few under-the-radar trends:

Demand for AI tokens is soaring. As I wrote in a column last month, the measurable nonfinancial metrics largely ignored by investors have been soaring. In a proxy for the broader market, Google has been reporting stunning increases in demand for AI, measured in tokens, or units of data processed:

“This time last year, we were processing 9.7 trillion tokens a month across our products and APIs [application programming interfaces],” Alphabet’s Google said in a May blog post, referring to units of AI usage. “Now, we’re processing over 480 trillion—50 times more.”

Google was just getting warmed up. “Since then we have doubled that number, now processing over 980 trillion monthly tokens, a remarkable increase,” the tech giant said in July during its second-quarter earnings call.

This month, Google said the figure had reached 1.3 quadrillion.

Price-performance is improving. This is possible because the price-performance of processing AI has been declining, as Nvidia Chief Executive Jensen Huang said last night on the company’s third-quarter earnings call. “Performance per watt, the efficiency of your architecture, is incredibly important,” Huang said.

Yes, AI infrastructure is expensive, but the customer’s total cost of ownership is falling from one generation of systems to the next, and that helps keep the upgrade cycle alive.

Some enterprises really are beginning to see returns on their investment in AI. “It’s very, very hard to build an agentic framework,” Leigh-Ann Russell, chief information officer and global head of engineering at financial services provider BNY, told me during our conversation on stage last month at the Gartner IT Symposium/Xpo. “But we have 117 solutions [including agentic] touching everything that happens at the bank and we’re seeing really, really tangible outcomes that impact our bottom line in terms of growing capacity.”

Walmart uses AI agents to help source products, informed by trend signals such as what teenagers are buying at the moment, Vinod Bidarkoppa, executive vice president and chief technology officer at Walmart International, said on the same panel.

He views AI as a “force multiplier” that will lift productivity across all lines of business, particularly in engineering, where significant increases are already visible, but not a means to replace human decision-making. “This is real value, if all of these agents are working together,” he said.

AI still entails plenty of risk and its benefits will be unevenly distributed. But our understanding of that risk is being rebalanced right now.

Nvidia by the numbers. Revenue for the third quarter ended Oct. 26 was $57 billion. That was up 22% from the previous quarter and up 62% from a year ago. Quarterly net income was $31.9 billion, 65% higher than a year earlier. Revenue from Nvidia’s data center segment set a record at $51.2 billion, beating analysts’ expectations of $49 billion. Nvidia expects fourth-quarter revenue to be $65 billion, "plus or minus 2%."

The reaction to Nvidia’s blowout results didn’t just signal the strength of one company, but the degree to which U.S. policy is increasingly organized around fueling and protecting the AI boom Nvidia is helping drive.

The Commerce Department approved the sale of up to 70,000 advanced AI chips to the United Arab Emirates and Saudi Arabia, two countries that have promised significant investments in the U.S. and deals with President Trump's family’s businesses, WSJ reports.

The deal would allow U.S. companies to sell up to 35,000 of Nvidia’s GB300 servers or their equivalents to both G42, a state-run AI company based in Abu Dhabi, and Humain, a Saudi government-backed AI venture, government officials said. Nvidia competitor Advanced Micro Devices has an agreement worth billions of dollars to work with Humain.

The agreement comes over security concerns that Chinese companies and others could find ways to access the computing power. Both Saudi Arabia and the U.A.E. have deep partnerships with China.

Axios reports that officials in the White House are pushing lawmakers to drop mention of China AI chip restrictions out of the National Defense Authorization Act, a move that would help Nvidia.

In a separate deal this week coinciding with Crown Prince Mohammed bin Salman’s visit to the Oval Office, xAI said it would work with Nvidia and Saudi Arabian partner Humain, on a data center in the kingdom that would consume 500 megawatts.

The White House is considering an executive order that would block states from regulating AI, an issue that has irked the administration and its Big Tech allies.

The order would call for a task force established by the attorney general to challenge AI laws deemed to be onerous, and federal funding to be withheld from those states through a federal internet access program, according to a draft of the executive order viewed by The Wall Street Journal and people familiar with the discussions.

Lawmakers on the other side say a moratorium would only be appropriate if Congress had traction on a federal AI bill. 

One area ripe for regulation: Kids and chatbots. A new report finds ‘systematic failures’ in how chatbots recognize psychiatric conditions. 

Common Sense Media and Stanford Medicine’s Brainstorm Lab for Mental Health Innovation teamed up to test how well OpenAI’s ChatGPT, Anthropic’s Claude, Google’s Gemini and Meta AI handled mental-health conversations with researchers posing as teens.

All of the platforms often failed to recognize signs of mental-health conditions, including descriptions of hallucinations, paranoid thinking, disordered-eating behaviors, manic symptoms, self-harm and depressive symptoms, WSJ reports. The platforms continued to offer general advice rather than directing teens to urgent professional help.

The growing prevalence of Chinese-made equipment in Europe’s energy, telecommunications and transportation infrastructure is sparking investigations into whether such devices are vulnerable to hacking.

The WSJ reports that last summer Oslo’s public-transport authority drove a Chinese electric bus into a decommissioned mine to find out. Isolated by rock from digital interference, cybersecurity experts found the bus could in theory be remotely disabled using the control system for the battery.

China’s embassy in Norway warned against “hyping up” theoretical risks. And politicians around Europe are also weighing cybersecurity risks against the potential benefits of China’s low-cost equipment and investments.

Frank X. Shaw, head of communications at Microsoft, said AI was essential in helping the company target their messaging in the critical opening hours of the July 2024 Crowdstrike software outage.

Microsoft used AI-powered tools to measure the “blast radius,” that is, the spread of news related to the event and what people were saying.

“And what people were saying, this is Microsoft’s fault,” Shaw told WSJ Leadership Institute’s Wendy Bounds at Wednesday’s CMO Council Summit in New York. The company knew it needed to respond fast and it knew where to do it, enlisting employees to appear on broadcast TV and radio spots targeting those areas identified by AI.

Demand for artificial-intelligence products continued to drive growth for Lenovo Group, which reported a record quarter. The world’s largest PC maker said revenue rose 15% to $20.45 billion, exceeding a market consensus of $20 billion. 

It's official. Yann LeCun, Meta’s chief AI scientist, is leaving the company, WSJ reports. LeCun said in a LinkedIn post Wednesday that he was creating a startup “to bring about the next big revolution in AI.” The company plans to focus on “systems that understand the physical world, have persistent memory, can reason, and can plan complex action sequences,” he wrote.

Nokia announced a new strategy in which it plans to accelerate growth in AI and cloud, while developing new technology that incorporates AI into the next generation of mobile networks, WSJ reports.

Former Harvard University President Larry Summers has resigned from OpenAI’s board after receiving criticism for previously undisclosed correspondence between him and sex offender Jeffrey Epstein, WSJ reports. 

Robotaxi company Waymo might be coming to a city near you in 2026. It already has plans for London, Washington, Detroit, Las Vegas, Nashville and San Diego. And this week it added Dallas, Houston, San Antonio, Miami and Orlando to the planned rollout, CNBC reports. 

Walmart reported strong sales for the most recent quarter and raised its financial outlook for the full year as consumers flocked to its value and fast-shipping services ahead of the holiday season. (WSJ)

A Centers for Disease Control and Prevention webpage that previously made the case that vaccines don’t cause autism now says they might. (WSJ)

European officials pushed back against a U.S. proposal for ending the Ukraine war, saying that Kyiv must approve any plan and that the conflict must not end with a Ukrainian capitulation. (WSJ)

The House voted unanimously Wednesday evening to repeal a controversial Senate-crafted provision that could grant some GOP senators at least $500,000 each in taxpayer-funded damages, after the provision prompted bipartisan outrage. (WSJ)

The WSJ CIO Journal Team is Steven Rosenbush, Isabelle Bousquette and Belle Lin.

The editor, Tom Loftus, can be reached at thomas.loftus@wsj.com.