- AMD has been making efforts to go up-market, while Nvidia is trying to go down-market.
- AMD's latest quarterly results included only half quarter worth of Polaris sales.
- It could potentially offer far more upside than Nvidia due to a smaller size, along with impending Zen and Vega releases.
NVIDIA (NSDQ:NVDA) may have enjoyed a dominant position in the mid and top-end GPU segments for the past several years now, but market dynamics of the GPU industry appears to be changing quickly. Since the start of this year, both Nvidia and AMD (NSDQ:AMD) have released products in a bid to eat into each other’s pies. This is certainly good for consumers, as increased levels of competitiveness are encouraging both GPU-makers to come out with truly competitive products. But where does it leave AMD and Nvidia?
AMD vs Nvidia - Shaking Up The Market
Anyone following the semiconductor industry would know that Nvidia has had an upper hand over AMD in terms of raw performance and computing capabilities for several years now. The market was happy to pay a premium price for its offerings, which allowed Nvidia to pretty much dominate the mid and high-end GPU segments. But it’s the low-end of the spectrum that eluded Nvidia, as AMD offerings tended to fare better on performance-per-dollar scales.
So Nvidia made a disconcerting (for AMD) effort last month by releasing two budget GPUs, 1050 and 1050Ti, specifically targeting the potential clientele of AMD’s RX 460 and RX 470. Granted that Nvidia has historically offered a range of budget cards in the past, but as I explained in my last article, these two were offerings are geared to compete with AMD chips on all three factors -- price, performance and power efficiency. This release marked Nvidia’s aggressive effort to venture into the low-end segment.
AMD also appears to be following a similar approach, except on a larger scale, in proportion to its size and product portfolio of course. It currently dominates the low-end of the market but it released a slew of products earlier this year, in order to increase its presence in the mid and high-end GPU segments. This was particularly evident with the release of its external GPU dock, Radeon Pro WX series of cards priced between $399 and $799 along with ARM-based Seattle offerings. The increased level of competitiveness, in terms of both price, efficiency performance, is probably why Google decided to use AMD's FirePro GPUs in its cloud computing platform next year. But that’s just the past.
In fact, it’s Vega GPU and Zen CPU line-ups are expected to be launched in the first half of next year that would cater to the mid and high-end segments of their respective operating segments. The point that I’m trying to make here is that while Nvidia is trying to go down-market, AMD is making several efforts to go upmarket in a bid to grow its consumer base and ASPs at the same time. These incursive attempts from both sides, to venture into each other’s core market segments are, in turn, fuelling the battle of GPUs.
The ground reality
The fact of the matter is that AMD’s latest quarterly filing accounted for only half quarter worth of GPU sales; the cards started shipping in early August and the cut-off date for the quarterly SEC filing was September 24. Plus, the company mentioned in its report that weak microprocessor sales weighed in on the Computing and Graphics segment, and limited the division’s revenue growth to just 11% on a year-on-year basis. So we can’t really gauge the sales performance of Polaris chips on AMD’s last quarter financials due to these two reasons, and the full effect of these new launches is yet to be seen.
|GPU||Percent of Steam Users In September||Percent of Steam Users In October|
But early data points indicate that Nvidia has been faring better than AMD. A hardware and software survey conducted by Steam revealed that users preferred Nvidia’s latest cards over AMD’s. The table attached above highlights the adoption rates of the latest GPUs from both companies. Since this is an optional user survey, and not sales or distribution channel-based inventory discovery, the results could be skewed on either side by a good margin. Also, do note that RX 460 and 470 numbers haven’t reflected anywhere in the survey, at least not yet. But I think we can safely arrive at the conclusion that Polaris hasn’t been the blockbuster success for AMD that the market was hoping it would be.
This can be somewhat corroborated by Nvidia’s blockbuster earnings and recent market share gains. The larger GPU maker of the two, recently revealed in its latest earnings release that revenues from the sale of its GPUs rose 63.4% to reach $1.24 billion in the last quarter on a year-on-year basis. This segment was the key driver behind its overall revenues surging 53.5% over the mentioned period.
(Source: Jon Peddie Research)
After this fantastic financial growth, Jon Peddie Research reveals that Nvidia's market share grew by 2.2% during Q3FY16 on a sequential basis. Both AMD and Nvidia posted a healthy discrete GPU shipment growth of 20% and 39% compared to the last quarter, respectively, but the former's much higher GPU shipment growth, larger revenue growth, and recent market share gains leads us to the conclusion that this round was won by Nvidia.
The promise of Vega
But that doesn’t necessarily mean that things will stay the same. As I’ve mentioned earlier, AMD is expected to release its Vega line of GPUs, starting with RX 490 in the first quarter of 2017. It would be a high-end GPU line-up that should, in theory, go head to head with Nvidia’s best in terms of performance and efficiency metrics. Where Polaris was developed to tap the low and mid-range segments, the Vega line-up would cater to the premium end of the segment.
This appears to have forced Nvidia to reshape its product roadmap. It is reported that Nvidia will now be releasing its next-gen Volta on 16nm nodes, as opposed to the intended 10nm, that too ahead of schedule. The development has two parts to it:
- Firstly, releasing the cards ahead of schedule suggests that Nvidia management isn’t taking AMD lightly anymore. It probably knows that AMD could do serious damage if its Vega release goes unchecked, which is perhaps why it had to move up the timeline for its Volta release.
- Secondly, in the process of releasing Volta early, Nvidia is going with 16nm chips as opposed to the intended 10nm. This means performance gains in Nvidia’s next-gen cards will be limited to architectural changes only, and die shrink or lithography advantages won’t play any major role here.
I believe that this would level the playing field for AMD. It should theoretically allow AMD to close-in on (or even surpass?) Nvidia in terms of performance metrics of high-end cards. The former already gives stiff competition to Nvidia in terms of performance-per-dollar metrics and it would be interesting to see if its Vega release can take the fight to the Nvidia camp in raw performance of top GPUs also.
So, what should you expect?
Up until a few quarters ago, we were discussing whether AMD would survive or go bankrupt. But now the situation has changed so much that we only see discussions about how good a fight AMD it could put up against Nvidia. This dramatic shift in market sentiment, from extreme pessimism to market-wide fervor certainly encourages long-side participation in the company.
Moreover, Nvidia released just one set of cards (1050 and 1050Ti) to go down-market whereas AMD is doing a whole lot more to go upmarket. So for a company that was considered financially incapable to even cover its interest payments a few quarters ago, this is a commendable turnaround. I’m of the opinion that AMD could potentially offer far more upside than Nvidia over the coming quarters owing to its smaller size and upcoming Zen and Vega releases, but a lot would depend on its execution as well.
Looking To Invest In Technology Stocks? Here Are Latest Top Tech Stock Picks.