With the first generation of Radeon RX Vega graphics cards now out the door, attention is turning to what AMD has in store for us next. AMD has just ordered production of its Vega 11 GPUs to commence. The Vega 11 chips will slot in alongside the current Vega 10 GPUs as the lower-end model, powering the successors to the Radeon RX 580 and Radeon RX 570, both of which use Polaris chips.

AMD’s orders for Vega 11 have been placed with Globalfoundries once again, the same manufacturer used for its current crop of Vega 10 GPUs used on the Radeon RX Vega 64 and the Radeon RX Vega 56. Once again they’ll be manufactured on the 14nm fabrication process, although in a major change from the current Polaris chips, AMD is reportedly planning to use HBM2 rather than GDDR5 memory for the successors to the RX 580 and RX 570 graphics cards.

As well as this, AMD’s plans for its next generation Vega 20 GPUs have emerged. Due for launch at some point in second half of 2018, Vega 20 will move the Radeon Vega architecture to the new 7nm process with support for faster HBM2 memory and up to 32GB capacity.

AMD Radeon RX Vega 20 promises to be an absolute monster and will go toe-to-toe with Nvidia Volta. It will make its debut with Radeon Instinct graphics accelerators, for use in AI and supercomputing before the eventual gaming variants roll out.

Interestingly AMD is expected to opt for TSMC rather than Globalfoundries for fabrication of Vega 20, possibly due to the limited manufacturing quantities of Radeon RX Vega 64 and Radeon RX Vega 56, which has burned AMD somewhat. Supplies of both have been extremely constrained, in turn driving up the price in some territories. The problem was only exacerbated by Vega's excellent cryptocurrency mining capabilities, a market which AMD courted with a dedicated driver release despite claims there would be enough Vega GPUs in stock at launch for everyone.

Plenty of moving and shaking for AMD over the next 12 months then, and an indicator that Team Red doesn't plan to rest on its laurels for long.