It's good to be Nvidia (NVDA) right now.
Though more than two years have passed since Nvidia launched the first GeForce gaming GPUs based on its Pascal architecture, and more than a year has passed since the most powerful Pascal gaming GPUs currently available hit the scene, the company apparently feels little need to quickly launch the first chips based on its next-gen Turing architecture.
When asked at Taiwan's Computex trade show when its next GeForce GPUs will arrive, CEO Jensen Huang said it will be "a long time from now," without offering further details. That has disappointed more than a few gamers who have been closely following stories related to the GeForce GTX 1180, a Turing GPU whose rumored specs suggest it will outperform the Titan Xp, Nvidia's most powerful Pascal gaming offering (it had an $1,199 MSRP when it launched).
Markets haven't been bothered much by Huang's remarks -- Nvidia continues trading near an all-time high of $266.59 -- and a quick overview of the competitive landscape helps explain why.
Though AMD (AMD) is holding its own in the mid-range PC GPU market with the help of its power-efficient Polaris architecture, Pascal GPUs continue dominating the lucrative high-end market. The performance of AMD's top-of-the-line Vega 64 gaming GPU (it launched in August 2017) is only about even with that of Nvidia's GTX 1080 (launched in May 2016), and trails that of the more powerful Titan Xp and 1080 Ti. In addition, reviewers found the Vega 64 to be much more power-hungry than rival Nvidia parts.
And from the looks of things, it will be a while before AMD rolls out the first gaming GPUs based on its next-gen Navi architecture. While AMD plans to launch a new Vega server/workstation GPU later this year, its 2018 roadmap doesn't include Navi. What's more, some recent reports suggested Navi will be more aimed at the mid-range market rather than the high-end. With AMD battling both Intel (INTC) and Nvidia while having a smaller R&D budget than either, the company needs to pick its fights.
Intel, which last fall announced plans to enter the high-end GPU market, might eventually make life tougher for Nvidia on the high-end. However, building a new GPU platform from scratch takes time -- 2020 has been thrown around as a potential arrival date for Intel's first gaming GPU -- and developing a highly competitive offering might take longer still.
In the meantime, Nvidia can take its time to roll out Turing. There is something of an opportunity cost to doing so -- launching high-end Turing chips would give Nvidia room to cut prices for its Pascal offerings, which in turn could let it take mid-range share from AMD ahead of Navi's launch. However, continuing to rely on Pascal for high-end sales probably serves to boost Nvidia's gross margins, since Pascal relies on a pretty mature 16-nanometer Taiwan Semiconductor (TSM) manufacturing process.
Also: Following months of graphics card shortages caused by cryptocurrency miners, Nvidia has argued that there's a lot of pent-up GPU demand from gamers that's beginning to be addressed as supplies improve. The company might be banking that gamers who were previously wanted to buy a high-end GPU but were unable to (at least without paying a big premium) will be content to buy Pascal-based cards.
Nvidia, it should be noted, is coming off a May quarter during which its Gaming segment revenue rose 68% annually to $1.72 billion. Nintendo Switch console processor sales, currency swings and (though much of it was satisfied via mining-specific cards) miner demand helped, but so did strong demand for Pascal cards among PC gamers looking to play AAA titles at high-quality settings.
It's still probably in Nvidia's interests to launch Turing before the holiday season arrives, to drive upgrades among its now-sizable Pascal base. But amazingly, considering how long Pascal has been around and how competitive the high-end GPU market has been in the past, there's no pressing need to move faster.