Oh, yes. I would like to say hello. And I also want to thank them for their very warm welcome. Everywhere I go, everywhere we go at ARK in Korea, we just feel the love and I'm very grateful for that.
So, it's interesting. When I was here in June, the semiconductor stocks cracked because we had a very strong employment report in the US and interest rates went up. And I remember saying at the time that, especially in the memory space, it's the most volatile space within semiconductors, so to expect a lot of volatility. However, we believe we're in a technology revolution that's going to last for years. So I took a look at SK Hynix. It's dropped from three million to 1.2 million or something like that. And I just looked at the valuation. It's a very low valuation. I think there is some fear that Chinese chips, Chinese memory chips, are going to start competing with Korean chips, and that might be true. But I think two things will happen to help. One is Korean companies are probably going to shift more aggressively towards high-bandwidth memory, which is higher value-added. So that's the first.
The second thing that I think is going to happen, and from these levels my confidence is up tremendously, is that this AI revolution that we're in right now, if you use a baseball analogy, we believe that we're not even in the first inning. We're still in the parking lot waiting to get into the stadium. We are in such early days and we think the demand growth is going to be explosive. So we can't use past cycles to try and understand this cycle. We have to understand that AI has been unleashed in a very profound way. It's going to increase productivity massively. Real growth rates are going to continue increasing, accelerating, we think, and with them interest rates will also move up.
Now, many people think that semiconductor stocks are inversely related to interest rates. When interest rates go up, they go down. That is not always true. Oh, in fact, no, it's not. In fact, many people remember post-COVID when interest rates went up and everything got wiped out almost. This time around, if you look carefully inside the interest rate structure, you can break it down into inflation and real growth. The recent increase in interest rates is because of real yields, which are reflecting real GDP growth. The increase has not been because of inflation, and companies in the AI revolution are probably going to grow so quickly that higher interest rates are not going to affect them. They might affect more mature companies. Yeah. But any companies involved in the AI revolution, we think are going to do very well over time.
The last thing I'll say is, does this mean that memory stocks here in Korea are not going to be volatile anymore? No, they will continue to be volatile. It's just the nature of the animal. They're the most cyclical part of the market. They're the most commoditized. That's why this move to high-bandwidth memory is so important. So the volatility should continue, but there's been so much correction to the downside. I can't imagine that we have much further to go on the downside.
Well, you know, it's interesting. We're at a moment where an ecosystem like Amazon is closing. It's becoming a closed ecosystem. They're not allowing Muse. They're not allowing Grok. And in fact, even Elon Musk said, 'Oh, wow. Maybe I need to start my own platform, my own shopping platform.' So, I think Muse is in a good place, meaning Meta. I think Shopify is in a very good place. It is.
Yeah. Shopify is the backend for a lot of merchants. Even Instagram. Instagram uses Shopify to help fulfill any orders it gets. So, we're looking at names like that. I think now OpenAI, interestingly, tried shopping before its time. They, before agents even, and so I think they tried to partner with Walmart, and Walmart turned around after three or six months and said this is not working.
Oh yes. So we're so early. We're so early. It'll be interesting to see if Amazon does open up, and I think it should, because odds are very high that it will offer the lowest prices and the greatest selection. And I'm trying to figure out why it seems so insecure, because this new world should work in its favor if it really does have lowest prices, best delivery, best selection. And so it doesn't really make sense for them to be basically saying no Muse and no Grok, but that's what they're saying. So we're early on.
I do think Meta is doing a very good job. In fact, Nick Gruce, who's our director of consumer AI and fintech — he's director of research for consumer AI and fintech — he is addicted to Muse already. One of the best things that happened to him, and maybe your viewers will like this tip, is he put in a prompt basically saying, 'I know that many people owe me money,' and it's not friends, it is former buildings where he lived, they never returned the security deposit, or it's the government, he got a tax refund but he had moved and he never received it. So he was able with one prompt to find $1,500 worth of money that people owed him and it was just in limbo.
Now, I don't know if the Korean system here in the United States is similar, but if it is, you know, you might be giving your viewers a very nice tip right now and a good way to use Muse. He is also using it for nutrition purposes: what should I eat today? What should I buy for the week? He is using it for workout purposes, you know, a different workout every day so he gets all the different muscle groups. And, you know, of course, he's using it to organize his calendar, make reservations, whether it's for vacation or dinner or what have you. So he's becoming very dependent on his agent, his Muse.
No, we do not believe that upside is capped already. In fact, Muse is up to 2 million weekly average daily average users and it has been number one in the app store, the iOS app store, the number one productivity tool for 28 consecutive days. Now, no one has done that since ChatGPT. So it's very popular. So when we made that trim, it was not a trim focused on Meta at all. It had seen some appreciation, but the reason we did it is we had the opportunity to buy a private company in our ETFs, Kalshi, for prediction markets. And because we run with so little cash in our portfolios, we had to take a little bit from almost all of our stocks. So that's why it showed up as a sell. But it wasn't an investment call. It was we need to make room for this opportunity we're so lucky to get. I think we might be the only ETF that owns Kalshi in the United States. And we also, in the private world, own OpenAI in our ETF. So we're very lucky to be getting these opportunities.
It is our thesis that we're not even in the first inning of this AI revolution and we are going to need so much more cloud capacity. Now, we didn't know that Jensen Huang would come out one day after we bought, or the same day we bought it, two days in a row.
And so Jensen came out and said if anything their credit rating should be going up. We're fully backstopping. They don't need the government. And most of the contracts, which are with hyperscalers, they're good for these contracts, meaning they will pay the contracts, they will satisfy the contract.
So, you know, this happened to us with Coinbase too. When Coinbase was absolutely getting killed by SEC Chairman Gensler's commentary and by the SEC serving it with a Wells notice, we knew back then that — and we bought on that, the stock collapsed. This is the same kind of thing with CoreWeave, but the stock collapsed back then. And why did we buy it if the SEC was against it? Because the judicial system was starting to rule in favor of crypto and against the SEC. So when we buy, it is with knowledge. It's not just to take advantage of a stock plummeting, although we are a liquidity provider. We will buy when stocks are down, whereas momentum players will not. And we will sell stocks when they're up, whereas momentum players will not. So this is what makes the market.
Yes. Never underestimate the Chinese when it comes to manufacturing. I've always wondered why they didn't get into memory sooner. And the answer is it's hard. It's hard to manufacture. So maybe they've figured it out. They have gotten better with technology, especially since President Xi Jinping is saying new productive forces and anti-involution. And although we don't see, you know, we don't see the competitive dynamic changing there too much.
So I think our answer to this is: one, we're going to need so much more memory than people understand. So we have another player coming in. We do think that the Korean players have an edge and will maintain an edge in high-bandwidth memory, and so that's probably where the revenue will begin to shift more. So that's good, and we need high-bandwidth memory.
Now, many people look at our purchase of Cerebras, and that's in the public market, and we owned Groq, that NVIDIA effectively bought, because neither one of them is dependent on high-bandwidth memory. But that was just a way of saying high-bandwidth memory is in short supply. And I think with the Chinese coming in, let's just say they are successful on the more commoditized end, the consumer device end. Let's say, you know, the recent increase in prices is not good for devices. The technology works best when prices are falling or when the quality is increasing at the same price. So I think that getting those prices down will restimulate the consumer device part of the economy, and that will be important.
Whenever I see prices going up in technology, I do get a little nervous because the price elasticity of demand is so high. And what that means, you know, when you raise prices on something, consumers will respond, and they have responded recently.
Yeah, it's very interesting there. If you look since 2019, when institutions became interested in crypto and Bitcoin specifically, the correlation between gold, physical gold, and digital gold, Bitcoin, is 0.1 — hardly any correlation at all. And asset allocators are looking for assets that are not highly correlated. Well, what had sold off recently, it was Bitcoin. What had appreciated recently, it was gold. And so what we're seeing is the Bitcoin-to-gold ratio has turned around. And if you look at the chart — and we can provide it to you if you would like — you look at the chart, you'll see that even though it's very volatile, Bitcoin to gold, the long-term uptrend is in place.
And so we believe that asset allocators are short Bitcoin and are probably moving in gradually here and long gold. So they're probably rebalancing a bit and maybe taking their first positions ever in Bitcoin.
