What's up: AWS suffers early morning outage, Anthropic's co-founder draws White House ire, Benioff's big apology, OpenAI backtracks.

Good morning. We hear a lot about the scale of capital pouring into AI. Understanding demand for AI is trickier for a number of reasons. For one thing, the metrics used to measure demand aren’t as readily comprehensible as the dollars by which we keep track of capital spending.

Here’s a way to look at AI demand that helps put the entire AI capital spending spree in a clearer perspective. Christopher Mims’ latest column in The Wall Street Journal explains the rationale for OpenAI CEO Sam Altman’s thinking behind the company’s deal with chip giant Broadcom.

Nvidia and Broadcom complement one another, like chocolate and peanut butter, he writes. Nvidia is the go-to company for companies building data centers to train their AIs, but Broadcom plays a role in custom chips “that can make the process of delivering AI—what’s called inference—faster and cheaper.”

The deal also helps OpenAI diversify its suppliers ahead of an expected boom in demand, according to Mims:

Altman has said that to deliver the artificial-intelligence services consumers want, his company’s data centers will need at least one AI-specific chip per user. In other words, billions of chips.

Experts echo this notion. If AI takes over all the tasks we’ve been promised it will, the world will need as many AI microchips as it currently has conventional microchips, says Ali Farhadi, CEO of the nonprofit AI research organization Allen Institute for AI.

As Mims writes, it’s a scale of computing power that boggles the mind. If that equally mind-boggling forecast for pervasive AI usage is correct, the scale of the data-center buildout is more comprehensible.

Internet users across the globe suffered disruption as a result of a widespread outage that hit dozens of major websites and apps using Amazon Web Services.

Sites including Facebook, Snapchat, Amazon and Roblox appeared to be down early Monday, with financial services Coinbase, Robinhood and Venmo also reporting disruptions, WSJ reports.

The company said the issues stem from problems with the Amazon DynamoDB system, which provides websites with database storage and computing power.

Users began reporting problems starting around 3 a.m. ET, according to Downdetector, a website that tracks service disruptions.

Anthropic’s problem might be that it’s the sober one at the AI rager, WSJ's Tim Higgins writes.

One of its co-founders drew a string of unusual rebukes from the White House this past week after penning a rather personal essay about his own uneasiness around the work his industry is doing.

“Make no mistake: What we are dealing with is a real mysterious creature, not a simple and predictable machine,” Jack Clark, whose official title at Anthropic is head of policy, wrote last Monday. “And like all the best fairytales, the creature is of our own creation. Only by acknowledging it as being real and by mastering our own fears do we even have a chance to understand it, make peace with it, and figure out a way to tame it and live together.”

Clark’s essay, adapted from a little-noticed conference speech he gave earlier this month, quickly drew condemnation from President Trump’s AI czar, David Sacks, and other tech luminaries, including Marc Andreessen and Keith Rabois.

Salesforce CEO Marc Benioff apologized for calling on the National Guard to go to San Francisco after a crescendo of rebukes from the city’s political and tech elite he has counted as friends, WSJ reports.

“Having listened closely to my fellow San Franciscans and our local officials,” he said Friday, “I do not believe the National Guard is needed to address safety in San Francisco.”

Meanwhile, at Dreamforce last week...

Benioff unveiled a partnership allowing users of OpenAI’s ChatGPT chatbot to access all manner of Salesforce app data, including CRM records, and control Salesforce AI “agent” tools, dubbed Agentforce, that can analyze such data and try to automate corporate tasks. Salesforce has banked on customers using its own chat app, Slack, to control those tools, The Information reports.

Wall Street investors are underestimating AI's potential to make entire industries obsolete, Blackstone's president said, adding that the impact of the technology was now "top of our list" when evaluating deals, the Financial Times reports.

OpenAI researchers recently claimed a major math breakthrough on X, but quickly walked it back after criticism from the community, including Deepmind CEO Demis Hassabis, who called out the sloppy communication, The Decoder reports.

The United States has been at the nexus of a data center boom, as OpenAI, Amazon, Google, Microsoft and others invest hundreds of billions to build the giant computing sites in the name of advancing AI. But the companies have also exported the construction frenzy abroad, with less scrutiny, NYT reports.

When China tightened restrictions on rare-earth exports this month, stunning the White House, it was the latest reminder of Beijing’s control over an industry vital to the world economy. Yet its dominance was decades in the making, WSJ reports.

President Trump’s vow to intervene against drug smugglers in Colombia widened a U.S. counternarcotics campaign in Latin America that began with military strikes on oceangoing boats but is increasingly focused on threatening governments in the region. (WSJ)

China said economic momentum decelerated to its slowest pace in a year, putting Beijing on alert in the midst of hardball trade negotiations with the U.S (WSJ).

Wall Street is starting to get a little defensive. In the midst of the market’s most unsettling stretch since August, investors have turned to utilities, healthcare stocks and consumer staples—industries that reliably churn out profits no matter the economic conditions. Electricity, drugs and groceries are always in demand, even when consumers buy fewer cars, phones and streaming services. Those three defensive sectors are on track to lead the S&P 500 index this month for the first time since June 2022. (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.