ELITE策略师：如何布局下一轮AI交易

Warren, in the last few weeks you've gone from being overweight on the stock to neutral, and now you're back to being overweight again. Explain your train of thought and why such a sudden shift occurred in such a short period of time.

Let me go back and, as you said, analyze our reasoning over the year. Hopefully, that gives you some idea of how we got to where we are today.

We started the year with optimism, and then the Iran war shook us a little. We saw certain things and downgraded the stock briefly between February and April. What really brought us back into the market in full force was the leak of the Mythos model in early to mid-April.

We really thought AI would be the leader in this market, so we increased the share of stocks in the portfolio again. I think the stock market, with its current overweight in the index, is virtually synonymous with the topic of AI. It really outperformed the index. We stayed with this through August because we wanted to be present during reporting season.

Our main thesis at the beginning of the year was that this would be a record year for market earnings, and we saw that in the way forecasts changed at the end of last year. So we wanted to be there. When we went through reporting season in August, we thought the market was really underestimating the risks from the Fed.

You have micro, or certain idiosyncratic, risk factors during reporting season, and then everyone starts looking at the macro or systematic risk factor. Coming out of reporting season, you start thinking about things like the Fed, the economy, and more general topics. That's when you usually see correlations rise, and in this market, correlations had been very low.

We thought that would happen. We believed the probability of a Fed rate hike in September was higher than the market thought, so we downgraded the stock. We downgraded it on August 10, when the market was at 7,753.

Slowly but surely, after Jackson Hole and the new data came out, you could see the probabilities begin to reassess the odds of a Fed rate hike in September. The market actually—the bond market and the rates market—became overvalued, and the stock market faltered a little bit, but it did pretty well.

The Fed raised rates, and we got through it. In our view, the Fed's message was that we were likely to see 2 hikes, followed by a long pause. I think that's quite acceptable for the market, at least in the near term.

This reminds me, if you look at history, of the 1997 cycle, when there was 1 hike, then the Fed paused, and in 1998 it went to a cut. The market then fell by 6% after the first increase and went up sharply.

I think the reaction will be more like the 1997 cycle than a traditional rate-hike cycle, where the Fed raises rates over and over again. We could be wrong, but that's how we interpreted their statements in September. We really wanted to be in a position when the earnings story comes to the fore again in a few weeks.

That's the big picture of how we got to where we are today.

1. Tech-heavy market and AI leadership

Let me be clear: Is your expectation of future strength in stocks despite the start of the rate-hike cycle because you expect markets and investors to be completely focused on earnings and to all but forget about the Fed raising rates? Is that part of your thesis?

To some extent, I think it's really about the structure of the market right now. The market is so saturated with technology and big companies, and I think it takes more than a couple of rate hikes to kill the AI story, which is very powerful. This is a long-term trend. All the metrics we track are strengthening when it comes to demand for computing power and so on.

For me, what I'm looking for—maybe the Muse app or something like that—is some kind of catalyst. You need that catalyst. Perhaps it will be new model releases before Anthropic's IPO. I don't know, but we need a catalyst that will really force the market to re-rate the mega-tech companies or hyperscalers, along with the semiconductor group.

I think this is exactly the mix we're hoping for. As rates continue to adjust, we would expect the 10-year bond yield to rise to 5.1% or 5.15% after this Fed rate hike. Our fear was that the market would become very narrow because the equal-weighted index is more sensitive to interest rates.

I think the market can grow even with narrow leadership, given how it is currently structured. The recipe is to get back to the AI topic and simultaneously increase the prices of semiconductors and mega-cap tech companies.

It's exactly this kind of leadership that I think could take us to the 8,000 mark for the S&P 500 within the next few months.

2. Macro shocks

You see, I'm probably just as optimistic as you are, if not more so. I am very optimistic about the coming months, but one of the things you write about, in addition to your “bullish” sentiment, is that the macroeconomic backdrop is particularly vulnerable to shocks. First, what does this shock look like, and where do you expect it to come from? And secondly, how do you reconcile the fragile macroeconomic backdrop with your more optimistic outlook?

The shock we thought was most inevitable was a Fed rate hike, which the market was not ready for. We were concerned, first, that the Fed would raise rates and, second, that it would target a full cycle of increases.

The guidance we've received—namely, the SEP, or Summary of Economic Projections—is released every quarter. The members of the Fed, not just the voting members, write down their projections for various economic and financial indicators. We received this SEP in September.

I was a little worried that the Fed would raise rates. Warsh is such a slippery guy. I didn't know exactly how he would communicate, and I still don't think his communication was anything special.

I was worried that the dots would indicate a major rate-hike cycle, confirming market expectations, as the market had already priced in 4 hikes. But that didn't happen. I think we got a few dots in the SEP that give me some confidence that there will be only 1 more hike this year, and that's it.

The Fed is forecasting year-end core PCE of 3.4%. Once we get a data revision, the figures that make up the PCE and have been greatly overestimating it will be adjusted. Portfolio management fees are the main one.

I think we will get this revision with the next PCE report, and it will cut a few tenths of a percentage point off the PCE figure. So I think this means that the Fed has set the bar very high.

By the end of the year, I think core PCE will be below 3.4%, and the Fed will be able to step back and say, “Yeah, things are going better than we expected.”

Another thing the Fed did in that forecast, the SEP, was say that the unemployment rate would be 4.1%, meaning it would stay at its current level through this year, next year, and through 2028. If we were to see a small wobble in the labor market, the Fed could theoretically look at that and say, “Okay, maybe the economy or the labor market isn't as strong as we predicted in September.”

In my opinion, these are 2 factors that point to the 2 rate increases embedded in the SEP. To go beyond those limits and have more hikes, it would require a strengthening of the labor market compared with its current state. It would also require inflation to reach a very high level, which I think is unlikely to happen.

For me, that macroeconomic risk has now decreased a little bit. Next, of course, will be elections and things like that. I believe the midterm elections will pose some risk.

We saw semiconductors start trading based on the Democrats' chances of winning or the Republicans' chances of losing the Senate. For me, that's a risk. The opposition to data centers is real. It's really starting to get bipartisan support, so that's also a macroeconomic risk.

However, I don't think the market is ready to see this as something that needs to start a sell-off. Those are the 2 macroeconomic risks—the Fed and the approaching election—that I think the market is vulnerable to.

The reason I was worried about this, answering your question directly, is that the implied correlations in the market were depressed. When correlations are low, they keep volatility low.

Volatility is essentially a function of how quickly the volatility of individual stocks changes, how much those stocks move, and how much those stocks move together—that is, correlation. If you have a group of stocks that move with considerable volatility but are not correlated with one another, the volatility of your portfolio or the index as a whole may be low.

That's exactly what we observed. The risk is that, in the event of a macroeconomic factor, all these stocks that previously did not move in sync will begin to fall together. Therefore, we have been particularly concerned about macroeconomics since the end of the last reporting season.

3. Low correlations and pair trades

When you talk about correlations, can you explain that in a little more detail? Do you mean specifically AI trading and everything else, or is there something at the index level that is hiding what is happening beneath the surface? Can you tell us a little more about that?

Yes, of course. I think it’s about how all the stocks in the market matrix move together. This can be seen both between and within sectors. I believe this market is unique in that we call it a long-short, or pairs-trading, market.

Many funds trading in this market are looking to play for the upside on the AI theme this year. Mostly, this means buying semiconductors and equipment in the AI development basket, which includes companies like CAT and other industrial names. Then they want to hedge it. They want to open short positions, funding these long positions with shorts that are not related to AI.

What’s interesting about our research is that throughout 2024, we saw a steady increase in interest in short positions in sectors completely unrelated to technology and AI. We’re seeing increased interest in shorts in the financial sector, consumer staples, and utilities. We saw that even in software there was a massive spike in short positions, and that’s obviously another sector in technology that’s often pitted against semiconductors to play on the AI theme.

Finally, what really caught my attention, and what I don’t think makes much sense, is playing semiconductor stocks up and hyperscaler stocks down. For a long time, this was the main trading strategy, or one of the sub-strategies, in the market: to hold long positions in semiconductors and have an underweight or short position in hyperscalers.

So, all these strategies, how people are implementing them, and the way AI has become such a huge factor—semiconductors now account for almost 20% of total market capitalization. It’s simply the dominant industry in the market. In my opinion, this is the real reason why index-level correlations, as measured by the correlation matrix of each S&P 500 stock, are lower this year and indeed close to historical lows. This is definitely a unique market structure.

4. Semis and broadening trade

Warren, one of the things we’ve heard a lot about over the last year or so is market expansion. People have been looking at the equal-weighted index, which has actually lagged the cap-weighted index for most of this year. You’ve written a lot about how semiconductors would actually be the “ideal leader” for this bull market.

Why are you leaning toward this when so many other people are saying we need this rotation, this market expansion, to pull the whole index up? Perhaps these are completely independent ideas, but do you see this as a conflict or perhaps a contrary view on your part?

I think it’s just a matter of understanding the rhythm of this bull market, and we’re all completely focused on the topic of AI. There is no scenario in which this market will feel good if AI fails. If everything is as the bears say, then this market will really be in trouble, and no amount of Exxon will save it. You can’t just switch to those stocks.

We’re all completely focused on AI. For this to work, if the bull market is going to continue, you’re going to have to incorporate some of these AI metrics into your macro analysis. Then I think leadership will persist in this technology space, and semiconductors are the perfect place.

If you look at one of our studies, we tracked every industry in the S&P 500—about 60 industries, something like that. We tracked how the S&P 500 behaves on days when the industry is rising, if you’re bullish, and how it behaves when the industry is falling. We tracked the spread between the days of growth and the days of decline. This is a way of estimating beta, or market sensitivity, using our own approach.

We’ve been tracking this spread in semiconductors, and it’s the largest this year. This year, if you hold the S&P, the difference in performance between days when the semiconductor index is up and days when it is down is 90%. No other industry had an indicator like that. This confirms that semiconductors are the heart of the AI trade.

As I said, if you believe this market will rise or fall based on AI, then semiconductors have to be in working order. They cannot fall apart if we want the market to truly reach stable new highs. They may not be the leaders, as I think we may see hyperscalers take the lead going forward, but I would really start to question our preference for stocks if semiconductors fell back below the July lows as the market turned more volatile.

For me, that’s one of the key risk-management metrics that we’re watching. I don’t see any other way, even though many are calling for the market to widen. Sometimes it does widen for a moment, and that’s the real rhythm of a bull market: semiconductors and AI lead, then the equal-weighted indices catch up a bit. Maybe rates are coming down, or we get positive data, and the rest of the market pulls up to catch up for a while.

This is the right rhythm for this bull market. Our rule of thumb is that you shouldn’t wait more than 3 months, so we’re not too worried about the narrow market or coverage issues that many people talk about. The only thing we want to see is that when the S&P updates its high, the equal-weighted index confirms it within 3 months, or 1 quarter, because historically this is the minimum for a healthy market.

I really don’t think this market expansion will be able to take us to new significant highs. We rely completely on the AI trend.

5. GPU availability and rental rates

If you’re generally optimistic about AI, you’re tracking a lot of these granular metrics, and one of them is the availability of computing power. Why are you looking at this? It seems to me that you’re talking about computing power more often than profits, which are a general indicator of the strength of a bull market. Tell me about this compute-availability indicator.

The compute-availability indicator, or GPU availability, is something we’ve been tracking since 2023. Fernando, who is doing research with me, started tracking GPU availability back in 2023. It came out of skepticism when the AI story was just starting to unfold, and we thought, “Okay, we’ll see about GPU availability.”

We started reaching out to neoclouds and pay-as-you-go cloud providers to see whether we could get a GPU within a day. Then we reported it as a percentage: what percentage of the time were we able to get a GPU based on the request? Zero would mean there was no chance—we couldn’t get the GPU. One hundred percent means that every time we applied, we were able to get a GPU.

One hundred percent is obviously a very free market, and 0% would mean an extremely limited market with strong demand. Over time, new GPUs appeared and Nvidia released new models, so we were able to combine them into an index. You can look at the A100, H100, H200, and B200 as they come out and start to understand demand within the entire GPU ecosystem. You can see it every day, just as rental rates provide daily updates.

6. Compute indicators flashing bullish

I think both of these factors—availability and rental rates—are better indicators of the state of AI and the real-time computing market than profits, because profits are always looking backward. They tell us about what has already happened. We get an idea of what is happening in real time, so for me this is more important than profits because it will directly affect future financial results.

Our work shows that availability determines rental rates. When you see a drop in on-demand availability for these GPUs, rental rates stabilize and then go up. The market is simply trying to reach equilibrium. If your on-demand service providers are renting Blackwell for $5 an hour and there is no availability, prices will start to creep up until demand deflates a bit. That’s just how the market works, and that’s why availability tends to lead rental rates.

We look at availability, rental rates, and the third thing we like to track when we have the data: lab ARR. There are concerns about circular financing, where one hand gives money to another within the AI ecosystem. I think lab ARR, or the ARR of AI applications in general, shows how external money comes into the ecosystem and validates the utility and spread of the technology.

These are the real-time metrics that we’re developing. We’re hopefully adding context by exploring all of this, but if you keep an eye on these 3 things, I think they will give you a good idea of how healthy real-time AI trading is.

What do these 3 things say today, or this week? Is this optimistic? Does this support a positive outlook for stocks right now?

Very optimistic. This is the most optimistic factor for the market, and partly because of that, my mood is a bit skeptical and I’m more cautious about taking risks.

Traditionally, I’m from that world where they say, “Don’t go against the Fed.” The Fed is raising rates now. One of the reasons we became optimistic in 2023 and essentially held that position as a company is that what we saw was the Fed was going to start cutting rates. We wanted to have that tailwind at our backs.

When the Fed supports you, it’s not the same as when the Fed tries to break your spine. That’s a big difference. I’m aware of that, and we’ve talked about macroeconomic risks. I think you’ll have to maneuver your positions more now that the Fed is tightening policy.

However, something is happening now. This is a moment for new, truly transformative technology, and we see it in all the statistics. We see it in availability and rental rates, and I worry about trying to convey that signal to customers.

I'm worried that the “right tail” of the distribution is underestimated—the right tail of the market and where we can move. We've seen this explosion of revenue, and you're starting to see adoption grow. It seems to me that the potential of the right tail hasn't been fully realized yet in the context of what could happen with this AI story, and I don't want to miss that.

You have the Fed raising rates multiple times, or you have this new technology that's really transforming how everyone perceives the economy in real time. And I know that it increases productivity; that's true. People like to ask, “Where's my productivity?” Well, look at the profit per worker in the knowledge-based fields that have already felt the most impact of AI—it's growing. Look at the revenue per employee in the software industry. Look at the profit per employee in the financial sector.

We've looked at both of those aspects, and for me, it's an area that we've integrated into our business. We've seen changes, and we're talking to our customers who are doing the same thing. While there are people who haven't gotten there yet, I think we're on the cusp of something big. All indicators point to this. All indicators point to this.

The availability of computing power—the availability of GPUs—is at a minimum. You can't get Blackwell on demand right now. We had a moment last month when H200 availability increased, but the rental rates for these older chips were very high. So the economics of inference dictated that you needed to move to Blackwell, which probably should have mitigated the demand for older chips. It did, and now you see the availability of older chips dropping as well.

When I see this, I think it's a really optimistic signal. I think the cloud revenue estimates for hyperscalers are underestimated. I don't want to miss it. I don't want to miss that same “right tail” if it comes up.

I agree with you. You personally give me confidence in my portfolio and in some of the semiconductor and hyperscaler companies that I still hold. Although we've seen some decline in the semiconductor sector in recent months, I think things are improving again.

So I want to ask you about frontier labs' revenue. A significant portion of it comes from the startup world. Early-stage companies are using these frontier labs' products and services. But it seems to me that if the Fed raises rates, costs go up for early-stage companies that depend on venture capital. Is this a risk to frontier labs' revenues and perhaps to investments in AI in general?

Yes, that's a big part of what we do. As I said, we can be optimists, and we need to understand the “right tail,” but we are first and foremost risk managers. We always ask. I think the AI world is now divided, as a short editorial comment, into certain camps. There are bulls and bears: bears are constantly looking for evidence that they are right, and bulls are constantly looking for evidence that they are right. We certainly lean toward the bullish side because the evidence points to that.

But it is important to be objective and clearly see the risks that may arise. Right now, as the Fed tightens policy, one of the common threads—I've talked about this on this podcast—is that it would take more than 25 basis points to stop trillions of dollars of data center construction. I think that's mostly true, but we wanted to explore this question further: Are we missing something? Is there any channel through which this can actually start to influence investment in AI?

As I said, these labs' annual recurring revenue, or ARR, is a very important metric. These labs' ARR has to continue to grow to justify all the costs and the capital raising that will be desperately needed. As many skeptics are very keen to remind everyone, there is a huge share of cloud revenue dollars going to hyperscalers. So you need to increase that ARR.

When the Fed raises rates, if you go back to the last rate-hike cycle in 2022, venture funding on a 12-month basis was cut by almost 70%. So this segment of the economy is vulnerable to higher rates and tighter financial conditions. If we look ahead, we've analyzed and sifted through the data, and on a 12-month basis, venture funding for startups has been about half a trillion dollars in the last 12 months. If you subtract the money going to the labs, since it is included in this amount, it comes out to somewhere around $290 billion.

We took a step back and said, “Okay, most of these startups in this new cycle are centered around AI.” They are big spenders and superusers, and it shows in the data: The top 1% spends orders of magnitude more than the next 10% of spenders on AI. The risk is that the Fed will hit the venture capital market and it will dry up. Funding is running out there, and the AI spending they are making is negatively impacting the labs.

We believe that in a worst-case scenario, about 25% of the labs' ARR is at risk if we have another 2022-like cycle of startup funding for this ecosystem. That would be a huge obstacle to investment in AI. If these labs went public and their ARR fell short of targets due to the Fed-related tightening of the venture capital market, you would see a market sell-off. The market would be very concerned about this.

7. Hyperscalers leadership

So this again points us to macroeconomics. While we need to talk about GPU availability and the economics of computing, we also need to watch the Fed, oil prices, and other macro factors, all combined, because the Fed could impact this AI story. It is not completely independent of the Fed.

This is a very interesting detail, a kind of private-market component here. One of the things I've written a lot about is that AI has surpassed macroeconomics. This is my opinion. I'm not sure that 25 basis points one way or the other really changes the history of AI trading, and I think you'll agree with that.

One of the things I often see is that hyperscalers are a trade to start abandoning now because they are wasting all their money, and this is capital that an investor may want to keep an eye on. For example, semiconductors. Do you support that? I know you also said that hyperscalers will eventually start to take the lead again thanks to semiconductors. Tell me about it.

Yes, I know. I think this was a trend in the first half of the year because the “payers versus recipients” chart was popular. If you listen carefully, they talk about it much less now. Far fewer people ask, “Where is the ROI? What is the return on investment?” I believe this last reporting season was decisive. I think it showed that we are already seeing a return on investment, and these things will only accelerate.

Going back to rental rates and what's happening with them, data center economics assume that GPU rental rates will decrease every year. But even rental rates for the oldest GPUs remain stable over a multiyear period. Rental rates for B200s have increased by 30% this year. This will affect hyperscalers' revenues.

Hyperscalers will show amazing results in the next few quarters. There is a shortage of computing power, and this will inevitably be reflected in the indicators. Personally, I don't even think this discussion makes sense anymore. I think this issue has already been resolved.

8. AI CapEx and the Fed

You saw what happened after Microsoft released its report. You saw what happened after Amazon's report. Even Google. In my opinion, that's all—the issue is closed. This discussion is over. In my opinion, you should come up with a new pessimistic scenario.

I like it. Again, for me personally, it's very calming.

So, Warren, you recently put together this great chart. It showed the growth in capital expenditures of companies from the S&P 500 against the backdrop of the Fed rate. I think this is very interesting as AI capital spending is increasing and the Fed is obviously raising rates. Explain this dynamic, as well as why you decided to place these 2 indicators on the same graph.

I think it's interesting to note that we've repeatedly pointed out that before the Iran war, the futures market was pricing in 3 Fed rate cuts over the next year. Now 4 increases are expected—a transition from 3 reductions to 4 increases. I think the Iran war played a huge role in that: the rise in oil prices and the fact that it wasn't a temporary spike. We seem to be stuck at around $100 per barrel.

I think that's a big part of it. This changed the Fed's response and the political pressure on it, but at the same time, the economy turned out to be stable and strong. Going back to the Fed's Summary of Economic Projections, or SEP, that I was talking about, we saw that the Fed raised its interest-rate forecast for 2028. This is a rather distant prospect; the increase was 50 basis points, to 3.8%.

It's starting to look like the Fed is admitting that the neutral rate—the interest rate that balances inflation and employment—has become higher. When it increases, it means that the “speed limit” for the economy has also increased. A real war doesn't raise the neutral rate. It does not raise the speed limit of the economy. This makes the capital-spending cycle we are in.

So the most interesting thing about this chart, in my opinion, is that there has never before been a period when the Fed has cut rates amid a capital-spending cycle. So I think we had a strange coincidence of disinflation, coming out of this “soft landing” scenario, where the Fed started cutting rates because of concerns about economic growth, and then AI, like a deus ex machina, suddenly appears on the scene, and you suddenly see 3% of nominal GDP being spent annually. Companies that used to buy back shares, pay dividends, and hoard cash are now spending all their money, borrowing more, issuing new securities, and injecting it all into the economy. This changes the marginal velocity of the economy. This is the main factor in why the neutral rate is rising, if it is indeed rising. It’s hard to do something like that and not raise the neutral rate.

The marginal velocity of the economy is increasing precisely because of this. I think this graph just shows how unprecedented this moment is: to see the Fed cutting rates and, at the same time, see a sudden, unexpected boom in capital spending. Capital spending is usually very cyclical. Usually, when capital spending increases, the economy heats up and the Fed raises rates in response, but this time the Fed lowered them as we were emerging from the effects of COVID and seeking a soft landing.

9. What ends the bull run?

And now we have capital investments in AI on a scale we haven’t seen in generations. So, for me, the main message of this graph is that if you are looking for a reason for the increase in the neutral rate and the increase in the marginal velocity of the economy, it is precisely because of the increase in capital spending, against the background of which the Fed lowered rates. I don’t really blame the Fed. I don’t think that—it was hard to predict that this would happen—but looking back, it’s the most significant factor in the transformation of the economy.

The idea of a “speed limit” is a great way to think about it, and at least for me, it makes everything clearer. I know you don’t think we’re at the peak of a bull market or the current market cycle, but what would have to happen for you to change your mind and say, “Hey, it’s time to take profits”?

I think it’s important. As optimistic as I am about AI and everything we’ve talked about, I don’t really want to bet too much on the upside of stocks if the Fed is really going to raise rates and, let’s say, enter a rate hike cycle—a repeat of 2022. If it’s a short-term adjustment and then a pause, I think we can get through it. But if it starts, if the data stays high, then I’m a little worried that the Fed got this kind of market reaction after the meeting.

This somewhat emboldens the Fed in its actions because it tells them, “Hey, the market has taken this calmly and supports your decision.” So I’m a little worried because the Fed also employs people. It’s just a committee of people, and they’re influenced by what others say. I think deep down Kevin Warsh wants to be a “hawk,” even though he was supposedly selected to lower rates. So I’m worried that they might get too relaxed and start raising rates more than I expect. That would be a problem.

And if we look further into the future, considering that AI, in my opinion, is the main driving force, I’m worried about the political backlash. I’m worried about the messages coming from the leaders of AI labs. These guys don’t resonate with average Americans at all. For example, Dario Amodei—every time he speaks, he only makes it worse; he buries himself even deeper. I think the average person on the street has already made their verdict that they don’t trust Sam Altman.

This is all getting political, but you have to look at it from a layperson’s perspective: Anthropic selected a team of AI supervisors, a group of these effective altruists, and these guys—nothing personal against you; you’re a young guy—but these are all very young people who exude a sense of detachment from reality. Bad marketing. Nobody wants this. Sometimes you look at this and think that maybe it would be better to have a situation like in The Terminator than to give these guys control over your life.

These are the things that worry me: policy. So far, all of this is being done somewhat awkwardly. They have to act, and that’s one of the reasons why we’re also optimistic about healthcare. I think it’s becoming a common belief, but I really think that labs should start trying to make some progress in healthcare and drug development, because it could even change public opinion.

10. Favorite sectors

They need to change their rhetoric from “we will destroy all jobs” to “we will cure cancer.” This would be a big win for their image. It seems like such an obvious move for them to change their vector to this type of signal for their own products and marketing.

Warren, I think my takeaway from what we’ve been talking about is—you mentioned healthcare. Would it be fair to say that semiconductors is your favorite sector?

I think we—I’m more of a hyperscaler right now. I like big tech owners of computing power. If you want to add spice, you can look into some neo-cloud services, but that’s not exactly my approach. I want to have long-term positions in computing power.

I think everything we see points in that direction, and it makes sense to have a stake in semiconductors as well. I think they will move together in the next wave. If you look at a sector outside of tech, we like healthcare for the reasons I briefly outlined here. I think this is a good sector for investment outside of the tech sector.

11. 3Fourteen Research

We’ve always liked to keep energy with a weight advantage. It’s not a huge market share, but it’s a great hedge. So technology, healthcare, and energy are all perfectly decent ways to structure a portfolio if you want to look at high-level sector diversification.

I really like it. Warren, where can we direct people to find more of your work?

Yes, you can…if you…we offer institutional research, so if you represent an institution, you can submit an application on our website. It’s the number three, and then the name 14research.com—314research.com, and we’ll contact you to see if we’re a good fit. You can follow me on Twitter: Warren Pies. There are two funds that track our models that we have been using for clients for years. You can familiarize yourself with them: RAA and FCTE tickers. So, those are two different ways you can find us.

Incredible. Warren, I really appreciate your time. You are truly one of the best at this, and I appreciate your ideas. I always learn from you.

Thank you, Phil. It was nice talking to you.