The two major players in the discrete graphics chip industry, NVIDIA (NVDA 4.35%) and Advanced Micro Devices (AMD 1.36%), have chosen wildly different paths in recent years. NVIDIA has expanded into the mobile and enterprise markets in areas where its graphics technology is useful, all the while remaining the market leader in the PC GPU market, while AMD has been actively diversifying away from the PC, building semicustom SoCs for game consoles and planning an assault on Intel's server empire.
Over the past decade, there's no doubt that NVIDIA has been a far better investment than AMD:
While shares of NVIDIA have nearly quintupled during that time, AMD has seen its stock fall by 62%, thanks in part to Intel running away with both the PC CPU market and the server CPU market. Another factor, and one reason why NVIDIA's stock has done so well, is that AMD has been losing discrete GPU market share to its competitor. In 2010, both NVIDIA and AMD had roughly 50% shares of the discrete GPU market. By 2013, NVIDIA controlled around 65%.
The question, though, is which company will prove to be a better investment over the next decade. Based on the strategies of both companies, NVIDIA seems more likely than not to continue to outperform AMD going forward.
A difference in strategies
NVIDIA's core competency is graphics, and everything the company does looks to leverage that expertise. NVIDIA's main business is still PC GPUs, and with the PC gaming industry expected to continue to grow even as the PC industry as a whole contracts, it's not a bad business to be in. Beyond PCs, NVIDIA has entered both the mobile and enterprise markets, and both initiatives offer significant growth opportunities for the company.
In the mobile market, NVIDIA created the Tegra line of mobile processors, with the Tegra K1 featuring 192 graphics cores built on the company's desktop architecture. The K1 is not a mass-market device; instead it targets markets where NVIDIA's graphics cores give the company an edge. The automobile market is one, where the Tegra chip can run in-car displays as well handle the visual processing necessary for various driver assistance features. The Tegra segment is currently losing money, but strong growth so far this year has narrowed those losses, and it could eventually become a major source of profits for NVIDIA.
The enterprise market offers another opportunity for NVIDIA. Tesla, the company's GPU compute product, uses NVIDIA's graphics cores to accelerate enterprise applications, from scientific simulations to business analytics. Of the top 100 supercomputers in the world, NVIDIA's chips are present in 18 of them, with AMD's graphics chips powering just one. NVIDIA's enterprise business has been growing at a 64% CAGR since 2010.
In contrast to NVIDIA, AMD is involved in various disparate businesses. The company makes PC CPUs, both with and without integrated graphics, discrete GPUs, server CPUs, and semicustom SoCs, a business that was started due to the company's design wins for the current generation of game consoles.
AMD's PC business is being torn apart by Intel. During the most recent quarter, the computing solutions segment declined by 20% year over year, even as Intel managed to grow its PC revenue. Intel has major manufacturing advantages over AMD, having just launched its first chips built on a 14nm process, with AMD still stuck at 28nm until next year.
AMD plans to introduce ARM server chips in an attempt to win market share from Intel, but simply offering a different architecture doesn't change the fact that AMD has been uncompetitive in servers for years. Intel has run away with the market, and its manufacturing lead will make it difficult, if not impossible, for AMD to win very much market share at all.
The best part of AMD's business is its semicustom SoCs, but with any further design wins likely to be far smaller than the game consoles, achieving meaningful growth will be difficult. AMD expects to announce one or two additional wins this year, but no information about the size or scope of these deals have been unveiled.
The bottom line
AMD is a company spread too thin, and any bright spots are offset by weakness elsewhere. NVIDIA, on the other hand, has stuck to what it's good at and expanded into markets where it can gain an edge. From a valuation perspective, NVIDIA is also more attractive, trading at a cash-adjusted forward P/E ratio of about 14, compared to 34 for the debt-laden AMD. With every market that it's involved in being a growth market, and with an attractive valuation to boot, NVIDIA looks like a better buy than AMD.
