What are asymmetric returns? 5 anti-consensus investment ideas promoted by Silicon Valley in 2026

Starting from the second half of 2025, the term "asymmetric returns" will appear repeatedly in the Silicon Valley venture capital circle. It was originally a concept in the financial field, meaning an opportunity with small downside risks and large upside returns. In 2025, it was moved to the context of AI and entrepreneurship and evolved into an anti-consensus decision-making framework. This article uses 5 specific directions and 3 judgment formulas to tell you what asymmetric returns are, why Silicon Valley advocates it, and how ordinary people use it to make investments and career choices.

The definition of asymmetric returns

The simplest definition: the downside loss of a decision is locked in a small amount, while the upside gain can be infinitely magnified. This is asymmetric gain.

Give an intuitive example. Buy lottery tickets for 10 yuan each and win up to 10 million. Downside is 10 yuan, up is 10 million, which is extremely asymmetrical. But the expected value of the lottery is negative, and the average loss is. What asymmetric income requires is not the "probability of winning" but the "return on winning" that is large enough and the loss can be tolerated.

Early-stage investments in Silicon Valley are typical examples of asymmetric returns. Every time you invest millions of dollars, you will only lose a limited amount. Winning a unicorn will bring you dozens or even hundreds of times in return. Even if you make one out of ten investments, you can still get your money back.

Why Silicon Valley advocates this idea

Three reasons. First, the uncertainty in the AI ​​industry is extremely high, and the traditional expected value model is invalid. Only targets with controllable downside and unlimited upside can survive the chaos. Second, entrepreneurship itself is the execution of asymmetric returns, and when the pattern is enlarged, it is power law investment. Third, Nassim Taleb's "antifragile" concept has been repeatedly recommended by Silicon Valley bosses, giving birth to a generation of asymmetric thinking believers.

Leading VCs such as a16z and Sequoia have long talked about "finding asymmetric opportunities", because venture capital itself is the ultimate version of this game - most projects return zero, and a few projects return the entire fund.

5 anti-consensus asymmetric investment ideas

The following five directions are perspectives that have been mentioned repeatedly in Silicon Valley over the past year and are of reference significance to ordinary people.

Idea 1: Buy niche rather than mainstream AI companies

The mainstream consensus is to invest in leading model companies: OpenAI, Anthropic, xAI, Mistral and the like. Anti-consensus is invested in the downstream infrastructure and tool chains of these companies, such as vector databases, Agent frameworks, training/inference acceleration, AI Ops, and enterprise implementation middleware.

Why is it asymmetrical? The valuation of leading companies is already in the order of hundreds of billions of dollars, and the downside space is not small, and the upside space is relatively convergent; the valuation of downstream companies is much lower, the downside is more controllable, and there is room for upside multiples. The specific valuation figures fluctuate greatly and are subject to public reports in the primary market.

Idea 2: Bet on undervalued "boring" businesses

The mainstream consensus is to vote for cool, sexy tracks with stories. Anti-consensus is to invest in "boring but urgently needed" businesses, such as B2B SaaS, compliance tools, supply chain software, and insurance technology.

Paul Graham has mentioned an observation many times: the greatest wealth is often born in areas that you hear nothing about. SAP, Oracle, and Salesforce are all examples of this. Versions in the AI ​​era are scattered in corporate finance, legal, customer service, sales automation and other scenarios. These companies are rarely discussed at Silicon Valley conferences, but their revenue growth tends to be more stable than that of star AI companies.

Idea 3: Hold cash and wait for valuation adjustment

The mainstream consensus is that the AI ​​bull market will last for many years, and if you don’t invest now, you will miss out. The counter-consensus is that valuations will be adjusted in the next one or two years, and waiting in cash is a potential asymmetric opportunity.

No specific year or data is cited. The logic is: the current early valuations are generally on the high side, and if there is a general correction in the future, players with cash can buy high-quality companies with current valuations at a discount. The opportunity cost of holding cash is missing out on some of the upside, but the return on buying a high-quality company on a valuation-adjusted basis can be multiples. This idea also holds true for individual investors - there is no need to hold a full position on the AI ​​theme, and a part of the liquidity can be reserved to wait for a better entry point.

Idea 4: Invest in early talent rather than specific projects

The mainstream consensus is to invest heavily in projects that are promising. Anti-consensus is to invest more energy and resources in early founders rather than on a specific project.

The logic is this: Excellent founders have a high probability of failure in their first project, but the probability of success in their second and third projects increases exponentially. Therefore, the long-term return of investing in the "person" of the founder itself is often higher than that of the investment project. The core strategy of institutions such as a16z and YC in the early stages is basically "invest in people but not in projects." For individuals, this idea is reflected in maintaining in-depth cooperative relationships with friends and colleagues who you are optimistic about in the long term, rather than just focusing on the current trend.

Idea 5: Invest in “seemingly boring” long content

The mainstream consensus is that social media content has become fast food and the quality of content is getting lower and lower. The anti-consensus is that long-form in-depth content is becoming increasingly scarce, and the long-term returns are huge.

"Long content" channels such as Substack's long newsletters, podcasts, in-depth documentaries, and professional research reports have begun to be rediscovered by users who are seriously looking for depth after algorithm fatigue. The investment logic is that publishing a piece of high-quality content on long content can be found in search engines and recommendation pools for many years; the life cycle of short videos is usually only a few days. This kind of time leverage is a typical asymmetric return.

Asymmetric judgment formula

Three formulas help you identify whether an opportunity is truly asymmetrical.

The first formula: the expected downside loss divided by the maximum upside gain should be much less than 1. A rough threshold is below 1/10.

The second formula: the downside trigger probability multiplied by the downside loss must be less than a psychologically tolerable threshold, usually a single-digit percentage of your net worth.

The third formula: The conditions for upward triggering are clear and quantifiable, not vague "maybe".

If all three are satisfied at the same time, it is an asymmetric opportunity worth pursuing.

How do ordinary people use asymmetric thinking to make decisions?

Three daily application scenarios.

First, career choice. When changing jobs, don't just look at annual salary, but also look at the maximum upside for long-term equity or options. Although working as a backbone in a small company means less cash, the asymmetry in the upward equity growth is often greater than working as a cog in a large company.

Second, learn to invest. The cost of learning a new skill is a few months, and the upside may be an increase in the career ceiling for many years to come. AI tools, programming, data analysis, and English are all areas with positive asymmetric returns.

Third, asset allocation. Put most assets into low-risk targets such as index ETFs or treasury bonds, and put a small portion into targets with high potential but controllable downside. This "barbell configuration" is an anti-fragile strategy advocated by Taleb. The core idea is "most of the time it is stable, and a very small amount of time it makes a big upward move."

Common misuses of asymmetric thinking

Three misuses to avoid.

The first misuse is to package "high risk" into "asymmetry". Buying garbage coins, betting on a single unlisted company, and shorting with leverage are not asymmetric returns, they are just gambling.

The second misuse is forgetting that the downside may actually be huge. If your downside losses will make you bankrupt or your health will collapse, this opportunity cannot be entered no matter how asymmetric it is.

The third misuse is to frequently try asymmetric opportunities. The core of asymmetric returns is "less but better". Finding a few truly asymmetric opportunities is enough for a lifetime. Frequent pursuit of new opportunities is itself anti-asymmetry.

FAQ

How do ordinary people apply asymmetric thinking when they don’t have much money?

Not having much money is the best scenario for practicing asymmetric thinking, because you can afford to lose. Three specific methods. The first is to put most of your daily savings into low-cost index ETFs for long-term holdings. The second is to buy several subscriptions to high-quality industry content. The difference in information is the upside. The third is to invest in learning a skill at zero cost and amplify the time leverage. Asymmetric thinking does not require you to gamble your fortune, but requires you to pursue excess returns within an acceptable range.

Are Asymmetric Returns and Power Law the same thing?

Similar but not quite. Power Law is a statistical characteristic of the outcome distribution that describes a small number of winners who take away most of the profits. Asymmetric returns are decision-making criteria that describe how to choose targets so that you are likely to become a winner. Power Law explains why most attempts fail, and asymmetric returns tell you how to reduce losses before failure while retaining the possibility of winning.

Can asymmetric thinking work in the Chinese market?

Works but with localization tweaks. The first is that compliance risks must be front-loaded. China’s regulatory pace is fast, and asymmetric opportunities are sometimes eliminated by a piece of paper. The second is that the liquidity premium should be higher, it is more difficult to exit the primary market, and you need to give yourself longer patience. Third, relationship costs must be taken into account. Certain asymmetric opportunities require key resource connections, which is a hidden cost. After comprehensive adjustments, there are still a large number of asymmetric opportunities in the Chinese market, but the judgment framework is more complicated than in Silicon Valley.

What is the relationship between antifragility and asymmetric returns by Nassim Taleb?

Antifragility is a philosophy, and asymmetric returns are a strategy. Taleb proposed in "Antifragility" that the system should not pursue stability but should pursue becoming stronger under pressure. Asymmetric return is a specific investment strategy to achieve anti-fragility. Through barbell allocation, your wealth will not only survive but also benefit from system shocks. Understanding Taleb's ideas can make your asymmetric investments more robust because you're not just looking at individual returns, but at the resilience of the entire system.

What are the new asymmetric opportunities in the AI ​​era?

Three directions are emerging. The first is the integrated services of AI and traditional industries, such as AI paralegals, AI medical imaging, and AI teaching and tutoring. The downside is small but the upside is big. The second is AI open source ecological contribution. Continuously maintaining high-quality GitHub projects or writing widely read technical articles can lift the career ceiling. The third is segmented content creators in the AI ​​era, such as reviews of a certain vertical AI tool, weekly in-depth analysis of a certain track, small but beautiful long-term returns. These directions do not require large funds and are suitable for ordinary people to enter.

Source of inspiration: Issue 380 of Ruan Yifeng's "Technology Enthusiasts Weekly" https://www.ruanyifeng.com/blog/2026/01/weekly-issue-380.html