Nvidia Just May Dominate the Entire Globe, One DataCenter at a Time

HEADLINE: "Nvidia preparing version of new flagship AI chip for Chinese market"

NEWS EXCERPT, Reuters Exclusive, 07/22/2024: "Washington tightened its controls on exports of cutting-edge semiconductors to China in 2023, seeking to prevent breakthroughs in supercomputing that would aid China's military.

Since then, Nvidia has developed three chips tailored specifically for the Chinese market.

The advent of tighter export U.S. controls has helped Chinese technology giant Huawei and startups like Tencent-backed, opens new tab Enflame make some inroads into the domestic market for advanced AI processors."

OUR TAKE: Nvidia Corporation is an American multinational technology company headquartered in Santa Clara, California, and incorporated in Delaware (Wikipedia), and appears to have a tremendous and extended (multi-year, and maybe a decade +) head start over every competitor worldwide. This lead also likely extends to any collective of 2-3 competitors, as Nvidia has perfected the multifaceted ecosystem required for its level of technological breakthroughs. This multifaceted ecosystem includes in-house capabilities and long-term external partnerships in research, sourcing, fabrication, manufacturing, and software development, with purposeful selective compatibility with other industry player, next-generation upgradability within the Nvidia realm to drive retention by current customers, and strategic interoperability with related technologies to ensure Nvidia's inclusion within other specialty operators across many industries.

With respect to Nvidia's domination of the DCaaS (shared datacenter services metered out by datacenter operators and hyperscalers for subscription use by clients), not only is Nvidia essentially oversold by a factor of 2X+ (some of these unfulfilled sales for current generation GPUs will inevitably be filled with later generation GPUs given the new annual refresh cycle), but the majority of the already-massive existing base of datacenters are queuing up for hardware refresh across the board (CPUs, GPUs, Networking, etc.) to have more compute power available in the same existing footprint, and to be more energy efficient per compute unit (in some cases there is simply no more electrical capacity available for non-residential use). There is also a newly-available certainty for datacenter operators that Nvidia's new hardware preserves their ability to mix and match with Nvidia's later versions of software and hardware, so operators could, for instance, keep a resource of H100s for later compatible use (higher latency at that time) and nimbly update some of their server racks with upcoming Blackwell, later planned Rubin, etc.

Given these factors, it seems an equally wise course could be to allow Nvidia to solidify its global dominance — as a friendly ‘monopoly’, at least on a relative basis as far as monopolies go — rather than force an adversary or collection of adversaries to play catch-up, emulate, and/or attempt to surpass Nvidia's capabilities and then project their own potentially-adversarial influence in our direction. Consider this: if Nvidia was a foreign-based entity with adversarial ambitions against the West (such as denying access to latest advanced technologies), or if it were forced into such a position by its home country, it’s likely the West would stop at nothing to equal or exceed their capabilities, and likely support competitors to out-innovate that Nvidia. Hmm, it may be wise to let this monopoly thrive.