Nvidia stock was headed for a fresh record Friday as the chip maker emphasized the durability of its artificial-intelligence processors, in a rebuttal to skeptics such as The Big Short investor Michael Burry.

Nvidia shares closed up 1.3% to $233.95 on Friday, just shy of its all-time closing high of $235.74, reached in May.

One major argument made by Burry and others who doubt the AI boom is that Big Tech overstates the depreciation schedule of chips bought from Nvidia. That’s the time frame used to spread the cost of an asset over its estimated useful life and would mean Nvidia customers systematically underestimate their annual expenses.

Nvidia now seems to be very explicitly pushing back on that claim.

“Over the years, every major operator has extended the depreciation schedule on its servers,” wrote Shruti Koparkar, a senior manager of product marketing at Nvidia, in a blog post on Thursday. The post noted that big customers like Meta Platforms and Alphabet’s Google now assume a six-year depreciation for AI servers, up from about four years previously.

This debate isn’t new but it’s becoming more important. Nvidia in August announced a partnership with Wall Street firms to deploy $500 billion in capital to support the AI sector. The proposed financing is likely to require loans at least partly backed by Nvidia’s AI hardware, so how long the chips hold their value is a vital question.

Barron’s previously wrote about how the rental value of even six year-old Nvidia chips actually has been rising, suggesting depreciation concerns have been overblown. Nvidia pointed to similar data in an investor presentation in late September.

Burry isn’t ready to give up the argument. In his own social-media post on Thursday he argued rising rental prices were a product of computing scarcity rather than chips holding their value. In his reading, prices for older hardware will drop sharply when more efficient processors come onto the market.

Nvidia has an answer to that as well—“CUDA, the software platform with which NVIDIA GPUs [graphics-processing units] are programmed, runs across generations, so nothing an operator already owns is stranded when a new architecture arrives,” Nvidia’s Koparkar wrote.

Of course, both sides are defending their own cases—Nvidia needs to convince financial firms they can lend against its chips, Burry has bets which stand to benefit from a drop in Nvidia’s share price. But so far it has been Nvidia which has been largely winning the argument, based on both chip rental data and its stock price.