What's up: OpenAI and Databricks strike deal to sell AI agents; Trump administration considers new chip tariffs; Microsoft disables some cloud services used by Israel’s Defense Ministry; winning the robot wars.

Good morning. The scale of investment in AI infrastructure is both mind-bending and historic. Will it pay off and if so, when?

“The artificial-intelligence boom has ushered in one of the costliest building sprees in world history,” Eliot Brown and Robbie Whelen write in the WSJ. The money pouring into AI data centers, chips and energy exceeds what it cost to build the interstate highway system over four decades, when adjusted for inflation. Proponents liken the effort to the Industrial Revolution -- and no one is sure how they will get their investment back—or when.

Check, please. David Cahn, a partner at venture-capital firm Sequoia, estimates that the money invested in AI infrastructure in 2023 and 2024 alone requires consumers and companies to buy roughly $800 billion in AI products over the life of these chips and data centers to produce a good investment return. Analysts believe most AI processors have a useful life of between three and five years. This week, consultants at Bain & Co. estimated the wave of AI infrastructure spending will require $2 trillion in annual AI revenue by 2030.

How can any of this possibly work? It’s a fair question. Here’s what to watch out for, going forward. The crucial matter isn’t simply whether consumers and businesses continue to ramp up spending on AI. It’s whether that spending will soon cross an inflection point and soar from current levels as models continue to improve and the supply of data center computing capacity hits the market. That, in theory, should drive down prices just as corporate AI projects come out of pilot stage en masse and begin to ramp up demand. The result would be an economic flywheel that drives AI spending much higher in a short period of time.

So far, the main application for AI data centers has been training AI models. The question now is if and when training will be eclipsed by business and consumer usage of AI, which is known in the field as inference.

As the WSJ story notes, AI proponents such as Oracle Chairman Larry Ellison are betting that will happen soon enough. “Training AI models is a gigantic multitrillion dollar market,” Ellison told investors this month. The market for companies and consumers using AI daily “will be much, much larger.”

Is your company ready to help the AI economy “inference” its way out of AI armageddon? Use the links at the end of this email and let us know.

Consider this week’s OpenAI and Databricks partnership, the latest instance of tech vendors teaming up to make it simpler for businesses to use AI agent technology on data that they own.

The pact between the two San Francisco-based firms makes OpenAI’s models directly available to businesses that store and analyze their corporate data in Databricks’ platform, allowing them to build AI agents on top of it.

OpenAI, which has its own platform for businesses to build AI agents, doesn’t have the access to customer data that firms like Databricks have. That’s part of why OpenAI is teaming up with the data-analytics firm, Brad Lightcap, OpenAI’s chief operating officer, tells the WSJ Leadership Institute's Belle Lin.

“Data is the lifeblood of AI in the enterprise,” he said. “Our hope is that from experimentation to deployment, this partnership is able to accelerate enterprises, both with AI and data, as they look to build the next generation of software.”

In addition to working with OpenAI, Databricks has announced deals with other model makers, including Anthropic.

The Trump administration is considering new tariffs on chips made overseas. 

The goal is to have chip companies manufacture the same number of semiconductors in the U.S. as their customers import from overseas producers. Companies that don’t maintain a 1:1 ratio over time would have to pay a tariff, according to people familiar with the concept.

It could be a boon for companies increasing U.S. production. Taiwan Semiconductor Manufacturing Co., Micron Technology and GlobalFoundries would get more leverage in discussions with customers, the Journal says.

The exemption process could test supply chain logistics. Companies like Apple and Dell Technologies import products containing different chips from all over the world. Under the proposed system, companies would potentially have to keep track of where all those chips were made and work with chip makers to match the number of U.S. and overseas products over time.

Meanwhile, Intel CEO Lip-Bu Tan’s is hustling to secure the investments needed for the chipmaker’s comeback.

The hustling  predates the Trump administration’s decision to take a 10% stake in it. And it will likely continue long after.

Among the companies Intel has approached about investments or manufacturing partnerships are Apple and Taiwan Semiconductor Manufacturing, WSJ reports. Japan’s SoftBank made a $2 billion investment in Intel as its discussions with the Trump administration were under way in August. Nvidia followed that up with a $5 billion investment earlier this month.

Microsoft disables some cloud services used by Israel’s Defense Ministry

Microsoft has disabled the Israeli Defense Ministry’s access to certain services and subscriptions, after finding evidence that the ministry used the tech company’s cloud services to surveil Gaza citizens. The company, which shared few details of its findings, said its policies prohibit its technology’s use to facilitate mass surveillance of civilians.

Microsoft President Brad Smith in a blog post said that the company would continue to support Israel’s cybersecurity efforts.

Microsoft opened the probe after the Guardian reported in August that Israel used Microsoft's Azure cloud platform to store data on Gaza civilians and surveil them. The issue has been the source of protests at the company with current and former employees disrupting company conferences and in August occupying Smith’s office. The company fired five employees following the incident.

Exercising the terms of service. Microsoft's decision marks the second example of terms of service being invoked in a high-profile use case. Semafore has reported that AI startup Anthropic is limiting how federal law enforcement agencies use its models, citing surveillance issues.

Amazon reaches $2.5 billion settlement 

Days into a civil trial examining whether Amazon duped customers into signing up for its signature Prime service and created a confusing process to cancel, the company has agreed to settle, the WSJ reports. Amazon will pay a $1 billion civil penalty, the largest in Federal Trade Commission history, and create a $1.5 billion fund to pay back to consumers.

Musk's xAI welcomed to the fold. The General Services Administration, which oversees technology procurement for the federal government, agreed to make available to federal agencies models from Elon Musk’s AI company. The WSJ reports that the agreement, which came after months of talks, could indicate that Musk’s relationship with the White House is thawing after his very public departure from the administration in late May.

Another deal for CoreWeave, OpenAI. AI cloud cloud infrastructure provider CoreWeave announced a $6.5 billion deal with OpenAI. CNBC reports that the agreement increases the total amount of contracts between the two companies to $22.4 billion.

Winning the robot wars. The same national push that made China a leader in EVs, AI and clean-energy technology is turning the country into a robotics powerhouse, the New York Times reports, citing data from International Federation of Robotics. China added 300,000 new factory robots last year. The U.S.? 34,000.

A federal grand jury in Virginia indicted former FBI director James Comey on Thursday on charges of making false statements and obstruction, days after President Trump demanded the prosecution and ousted the U.S. attorney who determined there was insufficient evidence to bring the case. (WSJ)

Secretary of Defense Pete Hegseth has ordered generals and admirals from around the world to gather for an urgent meeting near Washington on Tuesday morning, surprising senior military officers and stirring some consternation in their ranks. (WSJ)

President Trump announced many new tariffs Thursday, including a large one on pharmaceutical companies that aren’t building plants in the U.S. (WSJ)

The Justice Department has told federal prosecutors to draft plans to investigate a group funded by liberal philanthropist George Soros, a person familiar with the plans said, amid an escalating crackdown on President Trump’s opponents and critics. (WSJ)

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

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