The Six Five team discusses AMD AI Chip in China.
If you are interested in watching the full episode you can check it out here.
Disclaimer: The Six Five Webcast is for information and entertainment purposes only. Over the course of this webcast, we may talk about companies that are publicly traded and we may even reference that fact and their equity share price, but please do not take anything that we say as a recommendation about what you should do with your investment dollars. We are not investment advisors and we ask that you do not treat us as such.
Transcript:
Daniel Newman: It appears AMD and NVIDIA are going to continue to be challenged to make a ton of money in China. So we’ve heard about NVIDIA’s woes and chip controls that have continually thwarted efforts to build what they kind of call throttled back AI chips that can be used for hopefully what would be considered non-essential, non-military, in many ways, non-economic advantage building capabilities, yet still allow US companies to participate in the significant market opportunity that is China.
Basically this week, Pat, there was a Bloomberg story that broke. It was a China-tailored product by AMD referred to as the MI309, is what they’re calling it according to people. And it’s not clear who’s trying to buy it. But basically more or less that US regulators told AMD that it is too powerful to sell without a license. And so basically, AMD is going to need to go back to the drawing board on this particular piece at this moment. So apparently AMD thought it had met the requirements for the throttled down. Now we’ve talked about it, the requirements have been set, they’re clear, they’re understood. They haven’t changed, but they’re eligible to change at any moment. And there is a malleability to these metrics right now.
And I’m going to do a reiteration here because I think it’s really important. There’s three major reasons that the US is so focused on the semiconductor industry. It’s national security. It’s technology leadership. And it’s supply chain resiliency. And why do I say these three things? Because global economic dominance comes down to having the strongest national defense, the strongest and most robust supply chain, and of course being the global leaders in technology.
So right now, of course, this is where there’s a push and pull with NVIDIA, with AMD. They want to sell these chips into China. There’s a huge amount of demand for them. There was companies like Baidu, Tencent that basically said they have multi years of NVIDIA chips that are powerful enough to continue to advance their products and services. They have multiple years worth of products still. This is where we’ve talked about the fact that we have seen a load of this stuff sold into these cloud-scale companies. They’ve been buying truckloads of these things. Do not, for a minute, think that all the stuff NVIDIA has sold over the last few quarters has been implemented or deployed.
So the question mark basically is, if you’ve got Huawei building their own, we know that’s happening. We know China is going to focus on empowering its own. So if we don’t sell them into China, China will double down. And there is the EUV conversation, but we’ve seen recently, now PC cores are not the same thing. Our mobile device cores are not the same thing as building data center cores. But we’ve seen the newest Honor phone. Absolutely, Pat, we didn’t talk about it, but absolutely gouging market share from Apple with a new design that people are loving, built on seven, I believe, seven nanometer. So without EUV.
So some fascinating stuff going on, Pat. But bottom line here is, another roadblock hit. AMD is not yet cleared. And this is going to be a headline that’s going to be recycled several times, I think. I think we’re going to keep tuning them down because we do not want to let China catch us in the race for artificial intelligence.
Patrick Moorhead: Yeah. So I wrote a research note related to NVIDIA when there was a lot of discussion about this. And essentially, there’s three tiers that the US Department of Commerce looks at. And think of an XY axis. On the X axis, you have performance density. And the Y axis, you have total processing performance. And there’s kind of this red, yellow, and green, right? Green is, hey, don’t bother us. You don’t need a license. Typically, below 1600 TPP and increment on performance. On the yellow, that’s called moderate performance. The vendor gives the US government notice, waits up to 25 days of approval. And these have a TPP of 1600 to 4,800, but are primarily governed by performance density between 1.6 and 6. Then you’ve got red. That requires an all-in-license, right? TPP above 4,800, and nearly anything with a performance density above six. And these are reviewed with a presumption of denial. So if you want to look at the gory details of how this stuff happens.
And here’s the odd thing, is whether you’re AMD or NVIDIA, you know what the rules are and you’re going to go in there, unless of course it’s a red. So it’s just shocking to me that people would put together a chip and spend the time to do that and then it fails that level. But listen, this is the US government. I find it amazingly ironic. Dan, you and I could cross the border and stick one of these cards in our luggage. Not that we would. And I know people are like, “Listen, NVIDIA only sells racks.”
Daniel Newman: Systems, maybe.
Patrick Moorhead: VGX and systems. And they’re right. But you could hypothetically come across the border with a card, AMD or NVIDIA, and marry it up with a system afterwards that you could go over the border. The other thing is that neither AMD or NVIDIA does not phone home for drivers. So if you have the drivers, you have the card, you pull in the system from the other end, you have it. So I don’t think this stuff is even controllable, Dan.
Daniel Newman: Yeah, that, what you just said is probably intentional in many ways. And let’s be very candid, there’s a pretty good understanding that around the world there is a very robust gray market to move this product around. It’s not done nefariously, but it’s kind of like watches, Pat.
Patrick Moorhead: It’s funny, sometimes it is. When I was at AMD one time, in the front of USA Today business section was a picture of my product line in Iran.
Daniel Newman: But that’s what I mean.
Patrick Moorhead: It’s like, how did that get there?
Daniel Newman: This is like arms dealing though right now.
Patrick Moorhead: Oh, 100%.
Daniel Newman: It’s just the arms is artificial, arms-tificial intelligence? I don’t know. We’ll figure out some tacky thing that I can put a wrapper around.
Author Information
Daniel is the CEO of The Futurum Group. Living his life at the intersection of people and technology, Daniel works with the world’s largest technology brands exploring Digital Transformation and how it is influencing the enterprise.
From the leading edge of AI to global technology policy, Daniel makes the connections between business, people and tech that are required for companies to benefit most from their technology investments. Daniel is a top 5 globally ranked industry analyst and his ideas are regularly cited or shared in television appearances by CNBC, Bloomberg, Wall Street Journal and hundreds of other sites around the world.
A 7x Best-Selling Author including his most recent book “Human/Machine.” Daniel is also a Forbes and MarketWatch (Dow Jones) contributor.
An MBA and Former Graduate Adjunct Faculty, Daniel is an Austin Texas transplant after 40 years in Chicago. His speaking takes him around the world each year as he shares his vision of the role technology will play in our future.
