Nvidia is currently engaged in a civil war of sorts with its own AIB partners according to a report from DigiTimes. Server suppliers are increasingly providing the gaming-focused GeForce graphics cards for server and HPC usage rather than the more expensive Quadro and Tesla enterprise products specifically designed for server-based tasks.

Now you may be thinking ‘what’s the harm?’. If you’re selling graphics cards, you’re selling graphics cards, but the truth is that there’s much more money to be made from shifting Quadro or Tesla graphics to cash-rich businesses. Quadro is designed for workstation computers while Tesla is used for both servers and deep learning.

An individual Quadro graphics card can cost as much as £6649, while the supercomputers built with banks of these can go for well in excess of £150,000. For that money, they’re getting the best of the best in terms of graphics chips, but a single £6649 Quadro GP100 costs as much as nine GeForce GTX 1080 Ti graphics cards. Server farms and other associated companies can choose to use the much cheaper gaming GeForce graphics cards, saving themselves a fortune and cutting prices for their customers.

The caveat to all this is that Nvidia’s professional products offer the absolute best when it comes to performance, yet business users are paying over the odds in comparison to Nvidia’s already powerful gaming products.

Nvidia’s none too happy about this, believing server and GPU partners such as Asus, MSI and Gigabyte are sidestepping its enterprise products and lowering potential profits. According to DigiTimes, It’s now putting in stricter measures to prevent this happening again, including the potential threat of withholding GPU distribution from some partners if they don't adhere to its rules.

The knock-on effect is difficult to predict - either this forces Nvidia’s partners to run with its more expensive Quadro and Tesla products, or it pushes bargain seekers away, potentially even to competing AMD FirePro, Instinct or Radeon products.