What's up: New Estée Lauder hire points to digital push; How GenAI coding tools could redefine management; Apple commits $500 million for rare-earth magnets from U.S. suppliers; A reporter battles her favorite gym’s glitchy app

Good morning. Big Tech’s AI-powered hunt for land, water and gigawatts to power new data centers continues to take them far beyond their Silicon Valley terroir, potentially spawning new hubs of talent and innovation.

For local municipalities, having a Google or a Meta in the neighborhood can spell more jobs, a turbo-charged economy and a bit of that AI pixie dust.  But that's not always the case.

On Tuesday Pittsburgh was the center of the data center universe with President Trump in town touting new investments at an AI and energy summit on Carnegie Mellon University’s campus.

Among the highlights: Google said it would put $25 billion into AI data centers and related infrastructure in Pennsylvania and the surrounding region over the next two years. AI startup CoreWeave announced a $6 billion investment. 

The region has the talent to make it happen. 

Although still tethered culturally to its steel industry, Pittsburgh is now a growing tech, healthcare and robotics hub, the WSJ reports. Researchers and coders outnumber steel workers and CMU is a research powerhouse. In fact, CMU in May joined a lawsuit filed by other U.S. research universities against the federal government concerning a new cap on reimbursement for research.

Back to that real estate: “It’s the combination of technology, investment and power accessibility that makes a location key and successful,” said Commerce Secretary Howard Lutnick Tuesday, joining the president at CMU.

The WSJ reports that the event also highlighted the Trump administration’s embrace of fossil fuels and nuclear energy. Energy analysts warn that the U.S. won’t be able to meet data-center demand unless it also boosts production of renewables. 

Google also said Tuesday that it would put more than $3 billion into two hydropower facilities in Pennsylvania. Read the story.

Apple said it will buy $500 million of rare-earth magnets from MP Materials, expanding its U.S. supply chain at a time when the Trump administration has pressured it to do so, WSJ reports. MP will supply Apple with magnets it produces in a Texas facility and use recycled materials processed at a site MP will build in California. 

Today Apple sources rare-earth magnets from Asia, including from a Chinese supplier, according to its supplier list.

👉 Magnets go into the iPhone component that buzzes and vibrates to provide tactile feedback. They are also used in audio equipment and microphones inside various Apple products.

GenAI coding tools allow software developers to spend more time on their core work and less on project management tasks, a shift that could have implications for middle management, according to a Harvard Business School study involving some 50 thousand developers.

Researchers found that GitHub Copilot allowed coders to work more autonomously, reducing the need to reach out to managers or others to coordinate tasks.

“Gen AI can, in some instances, replace a manager, mentor, or any other individual that junior employees may ask for help,” lead researcher Manuel Hoffmann tells Harvard Business Review. 

ASML Holding shares slid after the Dutch supplier of chip-making equipment said it could no longer guarantee growth in 2026 amid increasing uncertainty from President Trump’s tariffs.

AMD said Tuesday it will restart shipments of its MI308 chips to China after receiving approval from the U.S. Commerce Department, Bloomberg reports. The news follows a similar announcement by rival Nvidia concerning the resumption of sales in China of its H20 chip.

Microsoft’s cloud computing work with the Defense Department partly relies on a workforce of engineers in China. And while U.S. citizens with security clearances, called  “digital escorts,” oversee their work, many lack technical expertise to spot potentially malicious work, according to a ProPublica investigation. 

Anthropic rolled out Claude for Financial Services, Bloomberg reports, joining rival OpenAI as well as Perplexity AI with AI tools targeted at finance professionals.

Troy Jones, Tesla’s top sales executive in North America, is the latest high-level executive at the automaker that is facing a steep drop in sales, WSJ reports. The move comes less than a month after the departure of Omead Afshar, a top aide to Chief Executive Elon Musk. The company’s director of human resources for North America departed in June, too. Earlier this year, a top AI executive also left the company.

Gandon, as Nestlé’s global chief marketing officer, led the company’s digital transformation across 188 markets and more than 2,000 brands, while developing partnerships with such companies as Amazon and Netflix, Estée Lauder said.

Reporter battles her favorite gym’s glitchy app

WSJ reporter Isabelle Bousquette might be one of the most avid Equinox Fitness Club members in New York. But she knows better than anyone that their app leaves a lot to be desired. Simple actions like booking fitness classes can be extremely slow and glitchy. Why is it so hard to search nine midtown clubs for the best pilates session between the hours of 6 and 8am on a Wednesday?

It’s a reminder of something we hear time and time again from customer experience-focused tech leaders: customers don’t always notice when an experience is good. But when it’s bad, they notice a lot. At an AWS customer event in New York this week, she had the chance to sit down with Equinox CTO Eswar Veluri to ask him: what is going on?

Here are edited excerpts from the conversation:

WSJ: I take a lot of fitness classes, and I have to tell you, the booking process can be pretty glitchy. Are you aware of that?

Veluri: I’m aware. The number of classes that we have, and the ability for people to view seven days of data and the metadata (who the instructor is, how many spots are open, what is the layout of the class) all of this information needs to be returned on the primary screen. We went into a mode where we can make it much quicker and faster but then you need to take an extra tap to see whether the class is open or not and that led to negative feedback. As the number of classes and clubs have grown we  realized that the problem has become bigger.

WSJ: It seems like it’s mostly a volume problem?

Veluri: Someone might say that's nothing compared to the volume of data other people handle but it’s the way we've architected it, which is what we’re changing. The metadata that is involved with the classes is like 14 different things. Each of these 14 are in different areas. It’s like every time someone is searching we're sending someone to go fetch a particular object from Room A. They come back, then they go to Room B, pick something else, then Room C, pick something else. Now the new architecture that we're doing is where all of these ABCDEFGH are cached on a shelf. It’s not earth shattering or anything. 

WSJ: In other areas of the club, your tech is great. You can scan a QR code on a treadmill and it’ll automatically save your run metrics to the app. I understand you’re developing a generative AI chatbot inside that will help in areas like booking classes and giving personalized workout recommendations. And now I’ve seen you even have robots that give massages? What’s your vision there?

Veluri: Aescape is the robotic massage. If you’re taking personal training with a coach and you had done extensive squats the previous day, we might recommend a 30 minute leg recovery the next day. And if you went into the massage room and you scanned your app, the robot would know that this is the recommendation. And then as it is doing some of the recovery work, it can actually notice certain areas of tightness or other areas that you might need working with, and that feedback is shared back with your coach.

We have these goals of personalization, personalization, personalization. So when they come in, we have a great understanding of their wearable data, what their goals are, what their interests are, some of them share more advanced data like heart rate and so on. So we’re in a position to provide a workout recommendation. It’s an evolution in increasing the number of inputs, making sure that input is actually useful to the recommendation that we're providing, validating that the end user found that information to be useful and then going on to the next thing.

Follow Isabelle on LinkedIn, Instagram and X for more behind the scenes on her reporting, and email tips and feedback to her at isabelle.bousquette@wsj.com.

President Trump is expected to sign an executive order in the coming days designed to help make private-market investments more available to U.S. retirement plans. (WSJ)

New York City’s mayoral front-runner, Zohran Mamdani, walked back his stance on the slogan “globalize the intifada” after a meeting with business elites. (WSJ)

President Trump’s moves this week to arm Ukraine and set a deadline for Russian President Vladimir Putin to begin negotiating in earnest followed a monthslong campaign by European leaders. (WSJ)

Israel’s recent war with Iran served as a cautionary tale for countries with sophisticated missile defenses and those that seek to have them. Over 12 days, Iran pierced Israel’s defenses with increasing success, showing that even the world’s most advanced systems can be penetrated. (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.