NVIDIA’s market cap has grown significantly over the last couple of months. Recently, the company overtook Intel, becoming the most valuable chip designer in terms of market capitalization. The deltas between the two have been growing wider and wider ever since. As of 24th August, NVIDIA is valued at $313 billion, a whopping 50% more than Intel’s $209.12 billion cap. Even if you combine AMD and Intel’s collective market value ($306.75 billion), it’s still less than NVIDIA’s market cap of $313.82 billion.
The reasons behind this are threefold: Firstly, the growth rate of the Data Center market has exceeded all expectations over the last 4-5 years. This sector is dominated by NVIDIA’s solutions, namely the Tesla accelerators, with the recently launched A100 Tensor core beating all existing hardware by a mile. The Mellanox acquisition further bolstered the company’s ambitions, providing it with its own interconnect and networking technologies. Although the buy-out cost Jensen quite a large sum of money (~$7 billion), it already seems to be bearing fruit, with the 2nd fiscal revenue accounting for 14% of NVIDIA’s overall earnings for the quarter.
The second factor driving NVIDIA’s growth is the stagnation in the discrete graphics card market and a lack of competition over the last couple of generations. The last AMD graphics card to challenge NVIDIA in the ultra-high-end segment was the R9 Fury X back in the first quarter of 2017, and to be fair, it didn’t do too well. AMD had to significantly reduce profit margins to sell the GPU and the limited 4GB VRAM buffer didn’t really help. Since then, the company has shied away from entering the high-end GPU market which has resulted in NVIDIA taking certain liberties. For example, the RTX 2080 Ti was nearly 50% more expensive compared to its predecessor despite the fact that the raster performance wasn’t all that impressive compared to the next fastest GPU, the RTX 2080 Super.
Over the last one year alone, AMD’s share in the dGPU market has fallen by almost 10%. RTG desperately needs a good season, otherwise, this trend will continue for another 2-3 years. The Navi 2x GPUs and the much anticipated Big Navi may just do the trick, but history tells us that AMD has time and again set expectations too high, setting up for an eventual disappointment.
The last factor is somewhat common to the industry as a whole for NVIDIA appears to have profited the most from it: That’s right, the pandemic and the accompanying stay-at-home lifestyle. While Google’s Stadia has largely failed, NVIDIA’s GeForce Now streaming service has been quite successful, attracting more than 1 million registered users as soon as the beta ended. The actual number is even higher by now, but we don’t have any hard numbers regarding the latest member count.
While Intel is fighting an uphill battle, AMD has a lot of opportunities to exploit. The server market, in particular, is a lucrative target. With the Epyc Rome CPUs, Team Red was able to capture nearly 10% of the overall market for the first time in more than a decade. The Data Center market, however, is a more difficult challenge. NVIDIA is dug in pretty deep with a diverse portfolio and a highly competitive product line. AMD’s recently announced CDNA architecture aimed at compute and AI workloads has been specifically designed for this purpose. Still, it’ll take multiple generations to establish a foothold in this space. At the very least though, you can be sure that AMD is headed in the right direction.