Picture a two-person team in Austin with $2,000 set aside for a second GPU. In January that budget covered two RTX 5080

Picture a two-person team in Austin with $2,000 set aside for a second GPU.

In January that budget covered two RTX 5080s.

MSRP was $999.

This week the RTX 5080 disappeared from Nvidia's US marketplace, the same way the RTX 5090 did before it.

The cheapest 5080 still in stock, a PNY model, is listed at $2,199.99.

That is 2.2x the launch price, and the memory shortage is driving it.

So $2,000 buys one card now, and the training run needs two.

One detail stands out.

Nvidia still keeps the RTX 5070 Ti and the RTX 5070 visible as out of stock.

Only the two highest cards were pulled entirely.

The likely read is that the silicon goes to RTX Pro Blackwell, where the margin is better.

A 24GB RTX 5080 is the rumored replacement, and it carries the same memory as the 5090 with half the CUDA cores.

None of that helps a small team with a deadline.

Waiting for a restock is not a plan when the model has to ship this quarter.

The math changes once you stop pricing cards and start pricing hours.

Two GPUs on a desk is a purchase, and in this market a bad one.

Hours on the same silicon is a cost you control.

The team above rented RTX 3090s at $0.15/hr, had a second node up the same afternoon, and kept the fine-tune moving.

Buy Now Pay Later, zero upfront, pay after one month.

https://ai.bhkcloud.com/?utm_source=linkedin&utm_medium=social&utm_campaign=bhk_social_2026w44&utm_content=morning What is the highest above MSRP price you have seen a card sell for this year?


Run your own model on local GPUs at $0.15/hr. Custom-model tuning and AI-slop removal by senior engineers, quoted per project. Talk to BHK Cloud

Originally posted on linkedin_personal

BHK Cloud