Following NVIDIA's (NVDA 1.45%) reveal of two high-end Pascal graphics cards in early May, Advanced Micro Devices (AMD 0.09%) finally unveiled the first graphics card based on its Polaris architecture during Computex. The Radeon RX 480 is a mainstream product that will be priced at just $199 when it becomes available on June 29th -- a bit more than half the price of NVIDIA's GTX 1070.
I argued in a previous article that AMD would need to launch a disruptive product in order to claw back market share from NVIDIA, and the RX 480 looks like it could fit the bill. The downside to this strategy is that AMD is handing over the lucrative high-end of the market to NVIDIA without a fight. But AMD desperately needs to regain its former position in the graphics-card industry, and going after the high-volume mainstream segment with a disruptively priced card looks like a pretty good idea.
The first Polaris
There aren't any independent benchmarks yet, so we'll have to wait until closer to launch before getting a clear idea of how the RX 480 performs. According to AMD, the card has compute performance of more than 5 teraflops, which would put it somewhere between the last-gen R9 390 and R9 390X.
Polaris brings with it plenty of architectural improvements, though, as well as a higher memory bandwidth, and substantially lower power usage. In terms of real-world performance, somewhere in the ballpark of the 390X would probably be a safe assumption.
This means that the RX 480 will be substantially slower than NVIDIA's GTX 1070. But with the 1070 priced at $379 when it launches later this month, the gap in price between the two cards may be larger than the gap in performance. Had AMD launched a $300 card instead, I would have been concerned that the 1070 would pull PC gamers up to a higher price point, like NVIDIA managed to do with its last-gen GTX 970. But the RX 480 is going after a different demographic entirely.
AMD is pushing this new graphics card as ideal for bringing the total cost down for buying a VR-capable system. Oculus VR had previously listed NVIDIA's GTX 970 and AMD's Radeon 290 as the minimum required to run the Oculus Rift, both of which launched at well over $300. The RX 480 lowers the cost of a VR-capable PC, but that's only half the problem.
The Oculus Rift headset retails for $599, and the HTC Vive costs a whopping $799. Even if a powerful PC wasn't required, the cost of the headsets alone would prevent VR from truly going mainstream until prices come down. AMD is making a big deal of the RX 480's ability to power VR, and I don't doubt that lowering the cost of the required PC will accelerate VR's ascendance to the mainstream. But if someone's willing to shell out $599 or $799 for a VR headset, I doubt they would then turn around and skimp on the graphics card.
Winning market share
Until NVIDIA launches lower-end Pascal graphics cards, which will happen within the next few months if history is any indication, the RX 480 should be a no-brainer for PC gamers looking to spend no more than $200. By pricing this card so low, it will be difficult for NVIDIA to undercut AMD on price.
Having said that, there's every reason to believe that the GTX 1060, or whatever NVIDIA calls its lower-end Pascal card, will be competitive with the RX 480. Another problem for AMD: NVIDIA currently enjoys a unit share of around 80% in the graphics-card market. There's certainly some stickiness in play, especially with NVIDIA building out its software offerings like GeForce Experience; so current NVIDIA users may be less willing to switch to AMD.
Despite these issues, I expect the RX 480 to drive at least some market-share gains for AMD, assuming it performs as well as its specs suggest. What the graphics card does for the company's profitability is another question entirely, though, and the focus on enabling VR seems early, with headsets being so expensive. With NVIDIA being the undisputed leader in the high-end of the market, I doubt the RX 480 will have much of an impact on that company's financial performance.
Pricing the RX 480 at $199 is an aggressive, and likely necessary, move by AMD. The onus is now on NVIDIA to respond with a mainstream variant of Pascal.