When AMD announced its Navi GPU architecture and publicly showed the chip beating past Nvidia’s RTX 2060 and RTX 2070, at modestly lower prices, it was inevitable that Team Green would respond. By all accounts, that response is being drawn up and will debut in the very near future.
The Nvidia RTX 2060 Super, RTX 2070 Super, and RTX 2080 Super are all supposed to debut next week, but the rollout will be staggered. The 2060S and 2070S review embargos will lift same-day, with the 2080S following later in July. Supposedly, each GPU will use the GPU from the tier “above” it, but this doesn’t make a lot of sense for the RTX 2060 / 2070 — these two cards are both based on the same physical GPU (TU106). The Inquirer reports that there may also be a 2070 Ti Super and 2080 Ti Super, with their own new GPUs.
There are a few things we don’t know yet, including what happens to existing RTX products when the new Super cards launch. Nvidia, however, had two basic ways it could play this situation: It could cut price on the RTX family to bring it more in line with Navi, or it could bolster RTX performance in an attempt to increase the size of the gap between itself and AMD.
It isn’t hard to see why Nvidia would go the route of boosting current performance, and we expect Super cards to come in above current RTX pricing, not below it. When Nvidia built its current RTX family, it made the decision to dramatically increase prices at every price point. Introducing new cards at higher prices but with increased performance gives Nvidia the option to improve its price/positioning without cutting costs.
Imagine, for example, that the upcoming AMD RX 5700XT is 5 percent faster than the RTX 2070 but $ 50 less expensive. Nvidia could cut the price of the RTX 2070 to match the RX 5700XT on price-performance and simply eat the reduction — or it could either keep the RTX 2070 in-market at the same or a slight price cut and introduce a higher-performing RTX 2070S at an improved price/performance ratio. All Nvidia has to do to improve the price/performance ratio on the RTX family is increase price less than the amount it boosts performance.
The other reason Nvidia might well prefer to raise prices rather than cutting them has to do with Navi’s die size. As the screenshot above shows, AMD’s Navi is a 251mm2 part. The RTX 2060 and 2070 are both 445mm2 parts. RTX 2080 is even larger, at 545 mm2. We don’t know anything about relative yield for 12nm versus 7nm chips, but AMD has at least a potential advantage, in terms of being able to absorb a lower price point. Neither company, however, is likely very interested in kicking off a major price war. After being forced into less-than-ideal price banding with its current products and with its Vega family poorly positioned, AMD likely wants to improve its price banding and overall margins just as much as to rebuild market share.