What's up: Trump calls on Intel CEO to resign; three questions for Lowe’s chief digital and information officer; Silver Lake to tackle data-center power bottleneck; Meta’s superintelligence SWAT team

Good morning. OpenAI on Thursday announced the release of GPT-5, its most advanced AI model, a culmination of more than two years of development that saw a series of setbacks and delays.

Also on Thursday, President Trump took to Truth Social to call for the resignation of Intel's CEO. More on that developing story below.

Interacting with the new technology “really feels like talking to an expert in any topic, like a Ph.D.-level expert,” Chief Executive Sam Altman said on a call attended by The Morning Download Wednesday.

Perhaps Dr. GPT-5 could help CIOs tackle the biggest challenge tied to any technological upgrade of this kind, namely preparing people and the processes to make the most of it.

At the same time, some two-plus years have passed between the launch of GPT-4 and GPT-5 and businesses might be better prepared for what comes next. Many now use a mix of multiple models from different providers. 

“I think employees are more focused on, am I getting my answers? Am I getting them faster?” said Lowe’s Chief Digital and Information Officer Seemantini Godbole, one of a number of executives who had a chance to experiment with GPT-5. Read the story.

Godbole: GPT-5 is so new that we have given it to only certain users. The full rollout will happen only in the middle of August. Then it takes about four to six weeks for us to get a more comprehensive point of view.

I think employees are more focused on, am I getting my answers? Am I getting them faster? I don't know if I can just go to a financial analyst and say, ‘Hey, what is the difference between a GPT-4o and a GPT-5? I think they may be just saying, ‘Hey, it's getting better.’ And as we connect it to more and more data sources, they’ll feel like, ‘Wow, it's found this information I used to have on SharePoint, and it just found it all for me.’

WSJLI: When it comes to actually rolling out GPT-5, what will that process look like?

Godbole: We have built a foundry, which basically means that we are able to talk to different models and switch models as required. We are going to connect to GPT-5 through that same foundry infrastructure that was built in-house.

The way we roll it out is, we'll always have the ability to say, ‘Hey, we send 50% of traffic to GPT-4o, and 50% of traffic to GPT-5. So between GPT-4o and GPT-5, I can always get a fifty-fifty reaction. Generally, that is a statistically significant sample, about 25% to 50% of users.

WSJLI: Based on your feedback so far, where do you think GPT-5 can make the most impact on certain business functions?

Godbole: The other day, I was trying to write a paper together for an upcoming offsite of our technology staff. It's a small meeting, and I was putting a document together about, how do we decide the prioritization?

It was really a thinking exercise for me more than anything else. When I was going through that exercise, I was using GPT-5, and it felt like it was almost going back and forth in a very intellectual discussion, putting together a comprehensive view about how we would think about this problem.

Also, parts of marketing who are trying to do analysis about cohorts of customers. Are we providing enough personalization? Are we making sure that we are creating messages that resonate with our customers? We could absolutely make use of these models.

The other area where I feel really optimistic is when you are doing a lot of financial analysis and you're trying to figure out trends and numbers and you're trying to put together correlations, I feel like that we could really benefit from this model.

Silver Lake invests $400 million to tackle data-center power bottleneck. GPT-5, or any new AI model, would not be possible without the massive amounts of power needed to run the required computing workloads. OpenAI recently struck a data-center deal with Oracle that totals 4.5 gigawatts of capacity, the equivalent power of more than two Hoover Dams.

Addressing the need, private-equity firm Silver Lake is expected to announce Friday a $400 million project aimed at securing powered land it will sell to data-center developers and hyperscalers, WSJLI's Isabelle Bousquette reports.

“The real limiter at this point on all of that isn’t: can you get the GPUs from Nvidia or others, or can you actually find the money,” said Silver Lake Managing Director Lee Wittlinger adding, “the real limiter is really power.”

The new platform is a joint venture between Silver Lake and Adam Fisher and Peter Rumbold, principals at Commonwealth Asset Management.

The joint venture said it has already secured 3 gigawatts of power and is developing sites in Texas and Georgia. The partners said they expect this phase of the project ultimately will result in securing over 6 gigawatts in various sites across North America, by leveraging both relationships with utility providers and in some instances by building their own power generation. Read the story.

Intel’s troubles deepen. President Trump Thursday called upon the chipmaker’s chief executive officer to resign, saying he was “conflicted,” deepening the president's involvement in the affairs of U.S. tech companies.

The WSJ reports that Trump appeared to be referencing Intel CEO Lip-Bu Tan’s past business dealings in China, which Sen. Tom Cotton (R., Ark.) called out in a letter to the company’s board earlier this week.

Tan is already at odds with some board members over questions as central as whether the company should stay in the manufacturing business or exit it entirely, according to people familiar with the matter. 

Tan took over as CEO in March with the hope of turning Intel around by cutting costs and improving the company’s performance in its existing business lines.

• Five Things to Know About the Intel CEO’s Links to China