Ray Dalio：當AI巨頭主宰美股，我選擇不賭方向，只做一件事

Original Title: Investment Principles: What Should You Do Under Existing Conditions?

Original Author: Ray Dalio, founder of Bridgewater Associates

Translation: Peggy, BlockBeats

Editor's Note: Against the backdrop of AI giants continually pushing up US stock indices and rising market concentration, Ray Dalio revisits a classic question in his latest notes: When a revolutionary technology is transforming the world, how should investors allocate their assets?

Dalio's core reminder is that technological progress itself does not make the related stocks equally attractive. Major technological cycles in history have typically gone through phases of excitement, crowding, volatility, and shakeout. Even long-term winners like Microsoft and Apple have experienced significant pullbacks during these cycles. Today's AI industry similarly faces multiple uncertainties, including overinvestment, intensifying competition, geopolitics, tax policies, anti-AI sentiment, and disruption from next-generation technologies.

The most important takeaway of the article is not judging whether AI will change the world, but discussing how investors should deal with a "highly concentrated" market structure. Dalio believes that when a handful of tech companies account for an increasingly large weight of indices, investors need to be wary of unconsciously holding highly correlated, high-risk concentrated exposure. Instead of continuing to chase a few leaders, a truly more robust approach is to build a diversified portfolio of high-quality, low-correlation assets and adjust volatility levels according to one's risk tolerance.

In his view, knowing what you don't know is just as important as knowing what you know. Facing the current AI-driven, overvalued, and concentrated market environment, investors should not directly translate their excitement about new technology into concentrated bets on a few AI stocks. Diversification, in Dalio's eyes, is the "Holy Grail of Investing" to navigate this technological cycle.

Below is the original text:

This note discusses: In the current environment, how should one participate in the investment game?

Imagine you are playing a game like bridge, poker, backgammon, or chess. It's your turn to move, and next to you is a computer that can help you assess the situation and suggest the next step. For me, the investment game is like this. Whether or not you have a computer to assist, I believe you should:

Based on the current board position, ask yourself what the next move should be. That is, you should decide how to act based on the existing characteristics of the market and the various forces currently influencing it.

I have been playing the investment game for a very long time. At this stage, my goal is to pass on how I would play the game; furthermore, I hope to create a platform where everyone can explore the investment game in their own way, learn, backtest how they would have done, and actually do it well. I believe there are right and wrong ways to handle the cards you are dealt. Therefore, when faced with a specific combination of situations like XYZ, you should ask yourself, "How should I bet in this situation?" and be able to give a good answer.

Now, I want to share with you my view of the current market characteristics, what I think should be done, and what I am actually doing.

How to Deal with the Current Set of Conditions

What are the most important environmental factors right now? How should one bet under these factors?

In my view, and likely in the view of most, the market environment we are in now is one where a very small number of companies, driven by a major new technology, primarily AI, are dominating market trends. These companies constitute a very high proportion of the overall market capitalization and are having a huge impact on the market and the economy. All such periods have a common feature: a lot of excitement, uncertainty, and volatility, concentrated in the new technology sector and transmitted through it to global stock markets. Therefore, the volatility and uncertainty surrounding this sector are very important.

Besides this, there are uncertainties related to other major driving forces. I call these the "Big Five Forces": 1) What is happening with debt and money; 2) What is happening with political and social issues, which can significantly influence taxes and other politically driven market factors; 3) The impact of geopolitical factors on the market, such as wars; 4) What is happening with natural forces; 5) What is happening with new technology. I input these situations into my investment system to ponder how to bet in this environment, while also thinking independently about what to bet on.

When thinking about how to bet in this environment, the most important question is: Which choice do you really want to make? a) Betting more heavily on the new technology compared to a broad stock index like the S&P 500, i.e., overweighting this new sector, or overweighting the few companies you think are best in it; b) Keeping your exposure roughly around the index weight; or c) Diversifying away from this concentration?

Almost everyone wants to buy the best investments and works hard to do so, and a new technology that seems to be changing almost everything has indeed emerged. But history shows that at this stage of the cycle, most people fail because they put a very large proportion of their chips on the stocks of a few leading technology companies. There is a logical reason behind this, and it has always evolved this way in the past. While this AI technology is indeed unique, there have been many equally "unique" new technologies in history that serve as analogies and references. People should study these cases; if they choose to ignore them, they must be able to explain well why this time is different.

The risks are undoubtedly high

All past cases of major new technologies have unfolded in similar ways due to the same logical reasons. High risk and huge uncertainty are inherent characteristics of these new technology companies. Looking back at the performance of these companies in similar historical environments, we find that even the best revolutionary new technology companies that prospered long-term, like Microsoft and Apple, suffered heavy losses at similar stages during their development. Moreover, it was not easy at the outset of these new technology companies, unlike in hindsight, to distinguish which companies would succeed and which would fail, like IBM. Observing all these cases, you see that major new technology companies inherently have highly uncertain futures.

For example, they either overinvest or underinvest. The reason is that if they don't invest enough to win the competition, they will definitely lose; but they cannot possibly know the future accurately enough to determine if they have overinvested. Whether it's overinvestment or underinvestment, the cost is high.

Furthermore, they cannot accurately predict all changes, including exogenous ones like monetary tightening, wars, significant tax changes, etc., all of which will affect them. Therefore, they all experience dramatic upside and downside cycles: first exciting investors, then scaring them, washing out vulnerable investors, and ultimately leading to exaggerated market volatility. Moreover, just as these new technologies and companies have disrupted their predecessors, most will ultimately be disrupted by newer technologies and newer companies in ways we cannot foresee. So, we should also consider whether the same risks apply to the current new technologies and tech companies. The impact of quantum computing is one of the known known risks. What about risks that haven't been imagined yet?

What about the risk from competitors? For instance, China is producing and distributing AI technology, and Chinese policymakers have a completely different view of the economy and AI. We are in a new technology war, and national leaders believe they must win it. Their understanding of AI and its impact on the economy and people's welfare will lead them to provide this technology for free or at low cost, due to its huge productivity gains and overall improvement in living standards. In their view, profits are less important than the overall benefits of many people using these new technologies. I believe they will compete in international markets just as they have with cars, solar panels, batteries, and many other products.

The current set of conditions closely resembles many historical cases that provide lessons. I can't help but think of how, towards the end of the Dutch Empire and the beginning of the British Empire, the British defeated the Dutch in shipbuilding and other important industries. Furthermore, there is a geopolitical conflict surrounding Taiwan, which should at least make us consider the possibility that, as a tool of geopolitical warfare, China might block the flow of chips from Taiwan. AI stocks also face other risks, such as the potential rise of wealth taxes and other taxes, which could force holders with large wealth concentrated in these stocks to sell; or the rise of anti-AI sentiment, which could limit the space for companies to advance technology.

I could list more things to worry about, and I could also list an equally long list of huge opportunities that AI will create and that I would want to bet on. I am not saying how these risks will definitely evolve, nor that one shouldn't bet on AI companies. I am merely stating that it is undeniable that there is a lot of concentrated risk in the market; and people should know how to deal with it in such an environment. Based on my study of all similar cases and the logic behind them, I am certain the risks are high, and the best way to deal with this environment is:

Embrace Diversification

As you may know, my mantra is "diversification." My "Holy Grail of Investing" is: striving to hold 15 high-quality, uncorrelated, and risk-balanced investments. In other words:

A well-diversified portfolio of good bets will outperform a single concentrated bet. It has a better risk-return ratio and can be engineered to achieve better returns for the same level of risk. The more risk is concentrated in one area of the market, the more one should diversify; this is especially true when the market is driven by revolutionary new technology, which inherently creates massive uncertainty.

This is not an opinion, but a mathematical certainty. For example, if I compare an investment with a risk-return ratio of 0.3, assume a return of 6% and a standard deviation of 18%, which is a typical assumption for stocks; then, if I hold 5, 10, or 15 uncorrelated investments, I can achieve the same 6% return, but the risk measured by standard deviation would drop to 8%, 6%, and 5% respectively. Therefore, by holding 15 high-quality, uncorrelated investments, my risk-return ratio improves 4.3 times, from 0.3 to 1.29. If you wish, you can add leverage on top of this to get much higher returns for the same level of risk. This is a fact.

I have strong confidence in this. The reasons come from my backtests, the actual returns I've delivered in over 50 years of investing, and the probabilistic logic: good bets diversified and scaled to the volatility one wishes to bear will, over the long term, produce much better returns than the concentrated bets most investors tend to hold. More specifically, through good diversification, one can get a better risk-return ratio than any concentrated bet; then, by adjusting it to one's desired risk level, one can achieve higher returns at that target risk level, superior to any other process.

Because I am passing on this method, it is now my "not so secret" way to investment success. Nevertheless, I rarely encounter investors who think about investment strategy this way. That is, I rarely meet people who truly think from a portfolio construction perspective, comparing how a well-structured, diversified set of bets would perform against holding a concentrated position in the stocks of a great new transformative industry. Most people just think about whether these stocks and this industry will perform well and how to bet on them. The performance results for those who think about portfolio construction versus those who don't will be vastly different. Therefore, I will elaborate on my thoughts on how to do this well at another time.

For all these reasons, in the current set of conditions, thinking about how to play your hand should lead one to ask: How large of a concentrated bet should I really hold? And then diversify.

Returns Appear to Be Very Low

High risk is undeniable. Next, I will give a potentially wrong opinion: future expected returns are low. My judgment on future expected returns comes from valuation-related analysis and my bubble indicator readings: real returns for stocks over the next 5 to 10 years look to be around -5% to -10%, although these numbers have considerable uncertainty. In my view, these stocks are long-duration assets, carry high risk because the distant future is hard to see reliably; they appear expensive, and their holder base is not solid.

A Question from My Research Team on This Topic

In a recent meeting, a member of my research team asked me: Why do you think the market's current allocation is wrong? How do you know that the current lack of diversification in the market isn't for a valid reason? For example, some investors believe the expected returns for AI stocks are very high; or, when an industry accounts for such a high proportion of total market cap, this index concentration occurs naturally; or, when an industry is met with great enthusiasm, many investors buy these stocks without making intelligent and reliable calculations about what future earnings will be and how they should be reflected in stock prices.

My Answer

There are various reasons for price increases, and not all of them are good. Some investors think about prices and push them higher because they believe prices are still attractive relative to fundamentals; some hold these stocks long-term because they recognize it's a great new technology and see rising stock prices as confirmation that they are good stocks; others have index exposure, which passive