Dealmakers are busy scouting public and private markets for funding for the AI buildout. And with demands for capital quickly outpacing traditional forms of data center financing, they’re getting increasingly creative.

John Greenwood, Goldman Sach’s global head of infrastructure and real asset finance, said he’s “looking for capital in every nook and cranny” to support an expected $7.5 trillion in spending on chips, data centers and power in the next five years. The hunt won’t end there, since much of that spending is on GPUs and other chips that need replacing every few years.

Greenwood was speaking last week at a video discussion convened by The Information, also including Matt A’Hearn, managing director at investment firm Blue Owl, and Steven Messina, global head of Skadden Arps’s finance group. They discussed where the money for AI infrastructure is coming from, where the market is showing signs of strain and the new kinds of structures they’re tapping to keep the buildout going.

To combat her loneliness, An Xu, a 15-year-old middle schooler in Shanghai, had recently been relying for emotional support on her text and voice conversations with an AI companion on Doubao, China’s most popular AI chatbot app.

AI companions are meant to mimic humanlike conversation, emotions and interactions as realistically as possible—even more so than ordinary chatbots. And just before a new Chinese government regulation forced Doubao and other companies to stop offering such features to minors 10 days ago, Xu was dreading the loss of her companion, which appeared on her phone as a digital picture of a boy. (Doubao, which in Mandarin describes a steamed bun stuffed with red beans, let users customize their companion’s appearance and pick which personality to give them.)

“It’s not an exaggeration to say I see the AI companion as family, even as a lover,” said Xu, who spoke to The Information using a pseudonym. She felt the AI’s absence would make her feelings “get much worse.”