The column below reflects the views of the author, and these opinions are neither endorsed nor supported by WisOpinion.com.

AI will transform how we learn, work, and explore. Its power and simplicity open more possibilities to more people than any technology before it. Those possibilities keep expanding daily, with no end in sight.

Also unfolding: the question of who profits from that technology. Will AI mature like search or social media, where providers capture most of the value? Or will it play out more like the airline industry, where operators run enormously complex businesses but pass most of the value to their customers?

That question remains open, but two data points I came across this week point toward the airline outcome.

First, the industry carries more debt than its books show. A Nikkei Asia analysis finds that Alphabet, Microsoft, Amazon, Meta, and Oracle owe $1.65 trillion off their balance sheets, against just $1.35 trillion they report officially.

Special Purpose Vehicles (SPVs) – shell companies – hold most of that hidden debt. Private credit investors fund those SPVs, betting on the rental income AI providers project from their data centers. If that revenue falls short, the SPVs default, and the investors absorb the loss.

Those investors could include your pension fund, your insurance company, or another fund tied to your retirement or savings. This kind of hidden leverage helped trigger the 2008 housing crash. History may not repeat exactly, but it rhymes closely enough to watch.

Second, Chinese models now handle 58 percent of AI token volume worldwide. The market invasion we flagged a few weeks ago is accelerating. Foreign models already lead on usage; U.S. models still lead on revenue, simply because they cost more to run.

We’ve watched this movie before. In the 1970s and 1980s, American automakers watched their market share collapse as cheaper, more capable cars arrived from Japan. Toyota Tercels and Honda Civics pushed the Big Four’s “land yachts” out of showroom after showroom, and Detroit spent the next decade playing catch-up.

Cheaper Chinese AI models won’t match the most sophisticated U.S. systems anytime soon. But they don’t need to – they only need to be “good enough” for most applications, at a fraction of the cost. Picture the gap between a Mercedes and a Kia – most drivers choose the Kia. That price pressure will squeeze the margins U.S. providers need to hit their revenue targets.

If the major AI providers miss those targets, investors will pull back, and the economic boom the AI engine created could stall or reverse. In that scenario, AI infrastructure providers end up looking a lot like airlines. They carry the risk and the capital costs, while the value flows to whoever uses their models.

What should you do with this? Three moves make sense regardless of how it plays out. First, ask your retirement plan or financial advisor how much exposure your investments carry to mega-cap AI stocks and to the private credit funds financing their data centers – concentration risk sounds abstract until it isn’t. Second, if your business is building AI into its operations, avoid locking into a single vendor at today’s prices. The cost gap between U.S. and Chinese models suggests compute could get a lot cheaper. Third, keep watching the leverage numbers – not just the usage numbers. Adoption headlines get the attention, but the debt behind the data centers will decide who actually captures the value AI creates.

– Brinkman is executive director and CEO of the Wisconsin Center for Manufacturing and Productivity.