NVIDIA GPUS became guarantees for startup billionaire loans

Although it is not uncommon for ordinary consumers to consider GPUS as investments, startup Fluidstack has put this idea into practice to gain a major loan. Specializing in artificial intelligence, the company used the large number of Nvidia signs that as a guarantee that it will be able to return $ 10 billion.

The amount was invested by companies such as Macquarie, and repeats a market strategy that has been gaining space. Coreweave was a pioneer of the model, using its funds to invest in buying various hardware. To pay its loans, the company “lends” its structure to companies like Openia, which demand a lot of hardware power to function.

Founded in 2017, Fluidstack registered $ 65 million in revenue in 2024, and expects to increase this to $ 400 million this year. She expects to further increase this value by offering her Nvidia GPUS structures for the use of the Mistral IA French and for German Black Forest Labs.

The biggest risk that the company faces at the moment comes from the fact that Hardware of the type may not work for long when they are exposed to extreme conditions of use. Thus, she will have to balance replacement costs with her ability to generate revenues and pay their creditors.

According to Coreweave, the estimated life of a GPU within the structure of an AI server is about 6 years. However, Some factors may contribute to this time to decreasemaking the company’s bets and fluidstack risky for its investors.

Among them is the fact that Nvidia has launched Specialized AI products at a growing pace, and many of them arrive with shorter users. This also contributes to decreasing the value of hardware already purchased, making them less attractive to companies seeking reselling them to get their investments back.

While the manufacturer is the one who should profit from the situation most, some startups are already looking for alternatives. Companies like Tensorwaves are already working with similar AMD GPUS agreements – which, although not as efficient as the competitor’s, have more attractive costs.

Fonte: PC Gamer