Bill Bishop was discussing what the midterms mean for chip policy and the new scrutiny of Nvidia. He argues that if the US slows its own data center build-out, it should tighten the loopholes that give Chinese companies access to advanced compute.
…can't slow down because we'll lose a race to China. I mean, you know, you want to slow down AI development, like data center development in the U.S. One of the ways you make sure that you're not then also sort of ceding ground to China is you slow down their ability to make money. China access.
So there is an argument or it's
possible to craft an argument. You say, well, okay, fine. You know, we need to slow data center construction in the U.S., but at the same time, you know, let's make it even harder for the Chinese to get this advanced compute that they're getting through all these different loopholes. And that may be something that maybe that's something that comes up, you know, probably not great for the stock market. But at the same time, right now, NVIDIA, it's not like because they're limited in China, they're losing sales. People can't buy enough of their chips. And Bloomberg had a story today that, you know, NVIDIA was going to sell their seller, or China had agreed or approved Chinese data center or Chinese AI companies and hyperscalers, they could buy these NVIDIA gaming chips that are useful for inference. And because of demand and sort of the ability to fill these orders construction or production, that Bloomberg today said that the Chinese companies make it 1% of what they ordered.
Right. Alibaba, 10 Cent, and ByteDance were the large customers that had ordered 6 million of NVIDIA's RTX Pro $55. Chips. And to your point, 1% of those orders will be delivered. It will be next year,
I think, is what the article said.
Yeah, I think they said over the next several months, but so the end of this year, certainly. And to the extent U.S. companies are more compute constrained in the future because of some of the data center restrictions, one would think that that puts even more pressure on NVIDIA to allocate compute…