NVIDIA is doubling down on 3 big rocks: gaming, cloud & crypto. Lucky NVIDIA, which also has 3 interesting edges: new PC, smarter car & micro data center. Where NVIDIA may change the computing landscape as we know it.
Dawn of New PC Era
As we speak, old PC giants are waking up after seeing Apple’s big wins on MacBook &, etc. with M1, ARM SoC. This time, it may be really different for NVIDIA.
According to IDC, PC shipment grows 55% in Q1 2021 & Apple is banking it huge with the largest growth, 111%. Apple’s new “reinvention”: powering MacBook & iPad with its ARM chips will become the obvious strategy others will chase again. To Apple’s credit, they did not just make the “new PC/tablet” as others might think. In my opinion, they’re reimaging the best professional productivity computing devices to what the state-of-the-art of HW & SW can be, even if it requires to change to a new architecture — ARM64, to create new OS — iPad OS, to revamp existing OS — macOS, to add new tool — Rosetta 2… You can find a similar pattern in many great businesses: long-term customer obsession as an Amazon insider’s breakdown in the book, Working Backwards.
On the other camp, many experiments started much early actually. For example, back in 2012, Surface RT with the best of Windows 8 & NVIDIA Tegra 3 SoC did not last for a year. It ended with a $900M MS write-down in 2013. Despite a few more attempts, Intel/x86 is still the king of the classical PC. Now the game has been changed forever when Apple changed its heart to design its M1 chip.
Maybe with the lessons learned, NVIDIA is playing this game a bit differently. It partners with MediaTek to put NVIDIA powerful GPU with MediaTek high-value ARM chip to the new PC. Which could be a valid competitor to Apple M1. Can’t wait to see how good it can run new AAA Android games & professional productivity apps 😉.
Powering Smarter Car
Yep, I’ve to talk about smarter cars 😅. Nevertheless, NVIDIA must see this already: -23% on its Auto revenue to $536M in FY21. I don’t know why the drop for sure. But there is a hit: Tesla has designed its chip for HW 3.0. And, using off-the-shelf chips is not like Tesla’s thing anymore. Sure, others can only depend on NVIDIA for now. However, a few of them are growing up quickly. Also, the market keeps supplying a few with great capital ammunition. A few may change their hearts to better control their destiny sooner rather than later. Or at least, they will try & gain better bargaining power. So what can be another good edge for NVIDIA to double down on its $8B bets, while there is still time?
A short answer is designing a new powerful chip for both Self-Driving & Infotainment systems, with powerful NVIDIA GPU & more ARM CPU cores to power both Self-Driving & Infotainment virtual machines. Where Self-Driving founds the “Rich” computing power, and Infotainment feeds the “Reach” to a critical mass. Together they may advance the computer power on wheels while driving down the unit cost of each chip. This is a better way to make a valuable section to drive a faster computing evaluation. It’s especially applicable for the chip business because it’s a high initial investment with a lower marginal cost of production.
We do need to take NVIDIA’s CEO, Jen-Hsun Huang’s words a grain of salt, it takes 500 chip designers & 1,000 SW engineers about 2 years to bring a new chip to the market. Nevertheless, it does illustrate how the chip business is different from the traditional HW business. Furthermore, with TSMC, Samsung & Intel are racing to provide advanced chip manufacturing services, NVIDIA’s true moat in this section is the economies-of-scale. Just like what happens in mobile. There are only 2 camps in the top 6 SoC players:
- Top OEMs design their own SoCs: Apple, Samsung & Huawei.
- Top SoC vendors to supply the rest: MediaTech, Qualcomm & Unisoc.
Raising of Micro Data Center
With the ascendence of Cloud War, Micro Data Center will become a critical part of the equation much quicker While most eyes are on building big data centers & backbone connections, Micro Data Center as the quicker & cheaper supplement is inevitable. Because 3 trends will continue to drive computing & storage to be more distributed.
- Real-time application demands are soaring, e.g. multiplayer games, HD/AR/VR streaming & Self-Driving cars.
- The last mile connection pipes become bigger, e.g. 5G adoption is accelerating & Biden’s big plan to make high-speed broadband affordable to all Americans.
- Increasing desires to keep data in regions.
Thanks to the “edge computing hype”, increasing investment is accelerating new tech development. So Mellanox acquisition happened, which gave NVIDIA a head start on this journey. But most importantly, it’s new & early. The time is right to build the whole SW stack from the ground up to better optimize for new applications. Because these new ventures have less path dependency on the great x86 SW legacy & established ecosystem.
Bonus
- If you connect these dots, you can see why NVIDIA willing to pay $40B to buy ARM. Only with many oppositions & government scrutinies, it’s not likely happening anytime soon.
- How about others & MS to design their chips. tl;dr MS & others have to try at least anyway. Let me know if you are interested to spend more time to further extrapolate the game plays & possible outcomes for this.
Full Disclosure
The opinions stated here are my own, not those of my company. They are mostly extrapolations from public information. I don’t have insider knowledge of those companies, nor a whatever expert.
