The other side of the meter: Who’s financing the AI buildout and why it matters

This story is about how the AI data center boom gets paid for, not about home energy upgrades directly, but it points to a trend worth knowing about: much of the huge cost of building AI data centers is being financed through off-the-books deals rather than company cash. Meta's two-gigawatt data center campus in Louisiana, for example, was built through a joint venture with a private-credit firm, with roughly $27 billion in bonds funded by outside investors rather than Meta itself. This kind of deal has become common, with data-center debt issuance nearly doubling to $182 billion recently.

The financing structure matters because it pushes projects toward speed. Lenders want power flowing fast to start earning returns, and connecting to clean power sources can take five to seven years, while building an on-site natural gas turbine takes about 18 months. That gap is steering many of these massive, decades-long facilities toward gas power rather than cleaner options, and toward running equipment hard rather than pausing for efficiency upgrades.

There's also a wider financial angle: some of the bonds funding these projects are held by pension funds and insurers, meaning ordinary long-term investment pools carry some of the risk if AI demand doesn't pan out. Oracle bondholders have already sued over losses tied to its data-center buildout.

None of this changes what's available for home energy upgrades or rebates, but it's a reminder that the same electric grid you rely on is being shaped by decisions about how AI infrastructure gets built and powered, often in favor of speed over cleaner energy sources.