A $400 million deal is drawing attention to what may come next in how artificial intelligence gets funded, according to TechCrunch AI.
The outlet reports that some of the first financiers to specialize in lending against GPUs — the powerful processors that made the current AI boom possible — are now turning toward inference chips. The arrangement is described as a chip-backed loan, meaning the hardware itself serves as collateral for the financing.
To understand why this matters, it helps to separate two phases of AI work. Training is the expensive, upfront process of building a model. Inference is what happens afterward: the everyday running of that model to answer questions, generate text, or power apps for millions of users. As AI products move from labs into wide use, the demand shifts from training horsepower toward inference capacity.
TechCrunch frames the $400 million loan as a marker of that shift, writing that it "points to the next wave of AI infrastructure deals." In other words, the financiers who built a business around GPUs are following the money into the chips that keep AI services running day to day.
The source does not detail which specific companies are involved or the full terms beyond the headline figure and the chip-backed structure, so those specifics remain unconfirmed here.
Why it matters: how investors choose to finance AI hardware shapes which companies can afford to scale — and a move toward backing inference chips suggests the industry's center of gravity is quietly shifting from building AI models to running them at scale.