A large part of Airbnb’s Q1 2026 earnings call was dedicated to talking about how the company is using AI tools for coding, customer support, and search. Notably, the company claimed that 60% of the code its engineers produced in the quarter was written by AI — echoing comments by others like Google, Microsoft, and Spotify, which have all talked about AI accelerating their programming.

Airbnb CEO Brian Chesky noted that the company finds AI particularly helpful for building tools for its API partners who manage their properties using different software.

“API partners say they want to be better hosts and need better tools. AI gives huge leverage — where you might have needed a team of 20 engineers before, an engineer can now spin up agents to do a lot of work under supervision. Adopting AI tools gives us leverage to build more software for API partners, accelerating work we previously did not have resources for,” Chesky said.

Airbnb has been slowly expanding its use of AI for customer support over the past year, and Chesky said on Thursday that its customer support AI bot now handles 40% of issues without escalating to a human agent, up from about 33% earlier this year. The travel company has also been experimenting with using AI to power its search function.

However, Chesky acknowledged the difficulty of truly employing AI tools in the travel or e-commerce spaces, pointing to weaknesses in the chatbot user interface.

“I do not think anyone has figured out AI for travel or e-commerce yet […] The design of a chatbot, as currently constructed, does not work for travel or e-commerce. There are four problems: too much text (most of e-commerce is photo-forward); no direct manipulation (you have to type everything rather than adjust sliders); poor comparison (you can get lost trying to compare thousands of options in a thread); and most bookings are multiplayer, while chatbots are primarily single-player, and not map-native.

Airbnb said net income rose 3.9% to $160 million in the first quarter, while revenue increased 18% to $2.7 billion, compared to a year earlier. Nights booked went up 9% to 156.2 million in the period. The company said its new “Reserve now, pay later” feature drew almost 20% of its gross booking value in the quarter.

Pity the grid operator PJM Interconnection. For decades it worked quietly and in the background, matching electricity demand with supply. Meanwhile, customers enjoyed some of the lowest electricity prices in the United States.

No longer. Politicians, businesses, households, and power companies think it needs an overhaul. Even PJM is in agreement.

PJM released a white paper this week that said the region “has years, not decades” to make fundamental changes to the way it operates. “The current situation is not tenable,” PJM CEO David Mills wrote in a forward to the report.

Normally, this sort of wonky report would land on the desks of a few legislators and regulators. But PJM’s territory includes a large number of data centers, including the compute-dense region of Northern Virginia. What happens to PJM will send ripples throughout the tech world.

The 70-page report is an exercise in navel gazing. But despite the deep introspection, not everyone is convinced the organization is up to the task of overhauling itself. One utility, American Electric Power (AEP), is considering pulling out of PJM altogether.

“The current state of PJM’s performance and stakeholder approval process does not give me great confidence that these issues will be resolved anytime soon,” Bill Fehrman, AEP’s CEO, said in an earnings call Tuesday. “In fact, if something is not done now, I expect we could still be having these same conversations in 10 years. The PJM market worked very well when supply exceeded demand; we are now in a very different time.”

Here’s what changed

Cloud computing and AI have begun to strain PJM’s existing generating capacity. Against the backdrop of surging demand, PJM paused applications in 2022 for new generating sources to connect to its grid, citing a years-long backlog. Just as the need for electricity was beginning to grow for the first time in decades, the grid operator prevented new sources from even applying to get hooked up.

PJM isn’t entirely to blame for the lengthy backlog. Many interconnection requests are duplicates — developers will propose essentially the same project in different grid regions to see which gets approved first. PJM’s sclerotic approval process meant that of the more than 300 gigawatts worth of projects in the queue in 2022, only 103 gigawatts ended up signing agreements, and only 23 gigawatts have been connected so far. Most developers withdrew rather than wait it out.

Demand in the region remains so large that, since PJM recently reopened the queue, power companies and project developers have filed more than 800 interconnection requests for 220 gigawatts worth of new power. PJM might have been able to pause new requests, but it did nothing to tamp down demand for new interconnections.

Here’s what PJM is proposing

In its white paper, PJM has proposed three options. One would require utilities and power generators to essentially make bigger, longer-term commitments. (PJM currently requires them to commit to supplying a certain amount of electricity for three years.) The second option would change reliability guarantees for customers — those who pay less might get their power cut first. The last choice would try to move PJM closer to a real-time market, where supply and demand dictate prices, without entirely eliminating stability from long-term contracts.

It’s hard to see how PJM emerges looking good in any of these scenarios.

First, the way PJM operates its market has somewhat locked it into a three-year mindset. That seemed to work when natural gas power plants were replacing coal-fired generators, but today solar and batteries can be installed at least two to three times faster. What’s more, the shortage of natural gas turbines means that power plants planned today won’t be able to install the equipment until the early 2030s. Plus, prices of turbines have skyrocketed on the back of demand for hyperscalers. Given those realities, it’s hard to see suppliers wanting to commit to an even longer timeline.

The second option would result in PJM splitting its territory, its customers, or both into groups of “haves” and “have nots.” For people and businesses stretched thin by years of rising utility bills, it’s hard to see them being happy with downgraded service. Politicians have seized on rising power prices and anti-data center animus, and so they are unlikely to back this one.

The last approach has the most nuance, but it also sounds like PJM trying to be all things to all people. It’s the type of plan that seems like it should appeal to large utilities like American Electric Power, giving them the opportunity to play in short-term markets to make more profit while also benefiting from predictable long-term contracts — having their cake and eating it, too. Yet if AEP, one of the largest utilities in PJM territory, isn’t thrilled with the menu before it, it’s hard to see how PJM can pick that one either.

Rising demand for data centers has just happened to coincide with disruption from renewables and batteries, which continue to drop in cost. Those trends are now colliding with an organization that doesn’t want — or doesn’t know how — to change the way it operates.

PJM may have thought its white paper mea culpa would buy it some time. But with politicians threatening price caps and utilities balking at future participation, the grid operator may not have years to sort things out. It’s looking like a messy few years ahead.