The Morning Download: AI Goes Nuclear

What's up: Dr. Chat is always in; tiny robots; AI agents crowd online checkout; tax tests California’s gravitational pull.

Good morning. Meta’s sweeping plan to run its AI infrastructure on nuclear power raises questions about how companies will keep up with the accelerating pace of technological change.

WSJ reported this morning on Meta Platforms agreements that would make it an anchor customer for new and existing nuclear power in the U.S., where it needs city-size amounts of electricity for its artificial-intelligence data centers.

Highlights of the story, which you can read here in full:

• The Facebook parent said it would back new reactor projects with the developers TerraPower and Oklo and has struck a deal with the power producer Vistra to purchase and expand the generation output of three existing nuclear plants in Ohio and Pennsylvania.
• Financial details weren’t disclosed, but the arrangements are among the most ambitious so far between tech companies and nuclear-power providers.
• Meta aims to see the first new reactors delivered as early as 2030 and 2032, a speedy target even for more conventional power projects. Its purchase of nuclear power from Vistra starts later this year and will keep power on the grid.

The plan is remarkable for its ambition and is bound to make people think about the AI boom, yet again, in a different way. But it's the pace of the project that in many ways should matter most to corporate tech leaders. The AI boom continues to accelerate and enterprises must find a way to keep up. The signs of this acceleration are everywhere. On Monday, Nvidia unveiled its faster AI chips sooner than expected.

“We’re very eyes-wide-open that the schedule is challenging, but we think it’s important to be bold,” Urvi Parekh, director of global energy at Meta, told WSJ. That’s more than an intention. Parekh has a clear idea of how processes need to change to make that timetable feasible.

Hitting those timelines for new reactors would require the companies to quickly select sites that would be acceptable to nuclear regulators, start working with utilities to secure grid connections, and get their manufacturing operations up and running, she said. But it would also mean they have a chance to meet the urgent demand for more electricity to fuel AI computing.

Companies need to think about how they can transform or eliminate processes so that they can move faster. AI is just getting started and companies need to figure out what they are going to do with AI as it evolves, and how it reshapes markets and industries. Those are open questions. The answers are going to keep changing and the timeframe is being compressed.

How is your company keeping up with the faster pace of technological change? Use the links at the end of this newsletter and let us know.

OpenAI this week unveiled ChatGPT Health, formalizing what many users have already been doing informally for years: turning to the chatbot for medical guidance.

Built in consultation with more than 260 physicians, the new feature can analyze medical test results, offer guidance on a range of health questions and help users prepare for doctor appointments, Bloomberg reports. It can also connect with electronic medical records and wearable devices. OpenAI says the tool includes additional safeguards to protect sensitive health data.

By leaning into healthcare, OpenAI is doubling down on one of ChatGPT’s most common—and sticky—use cases. While the company emphasizes that ChatGPT Health is designed to support, not replace, physicians, the reality is that for many users “Dr. Chat” will increasingly serve as a first stop.

That dynamic matters strategically. Once users trust ChatGPT with lab results, the chatbot becomes harder to displace, deepening loyalty at a time when AI competition is intensifying. More below.

When OpenAI rolled out its Instant Checkout feature in September letting ChatGPT users shop directly inside the chatbot, it looked like the long-awaited killer app for AI agents. It also promised something equally important for OpenAI: a steady revenue stream, with the company taking a cut of each transaction.

Months later, the promise remains—but the checkout line is getting crowded.

The Information reports that the rollout has been slowed by back-end complications, particularly around inconsistent product data. OpenAI partners Shopify and Stripe are now working to better standardize and share merchants’ listings, a reminder that agentic commerce depends as much on plumbing as on intelligence.

Meanwhile, Amazon has entered the fray with its own AI shopping agent. That move was inevitable for a company that effectively defines online retail. The agent doesn't merely limit itself to what’s on offer at Amazon.com, but, in true agentic fashion, searches the web for what users want.

That ambition has come with friction. CNBC reports that some retailers are pushing back, saying Amazon’s agent has at times surfaced items they don’t carry or that are out of stock, an early sign that AI-driven commerce may create as many headaches as efficiencies.

Google is bringing its Gemini AI model directly into the inbox. New Gmail features will include tools similar to AI Overviews in Google Search, but this time summarizing email threads and answering questions as well as offering an inbox assistant that flags urgent messages and surfaces key to-dos.

AI inside productivity tools isn’t new (see Microsoft’s Copilot for Outlook), but for Google, Gmail represents a tighter coupling between its core workplace products and its AI strategy.

That push comes as Gemini gains momentum, as I noted earlier.  Apple is reportedly testing Gemini to power a revamped version of Siri, while Samsung plans to expand the number of devices offering “Galaxy AI” features largely driven by Google’s models.

That momentum is showing up in the market. Alphabet on Wednesday surpassed Apple in market capitalization, becoming the second-largest U.S. company behind Nvidia, Barron’s reported.

And speaking of Apple, the company's recent iOS 26 release—featuring the Liquid Glass interface—has gained limited traction: Just about 15% of iPhone users have adopted it since its September debut, according to StatCounter data cited by Cult of Mac.