Peng Fu: The Key Factor Determining Global Asset Movements in 2026—Are There Real Cars Running on the AI "Highway"?

On December 20th, at the "Alpha Summit" jointly hosted by Wallstreetcn and CEIBS, well-known economist Fu Peng delivered a speech titled "Reconstruction of Order in the AI Era".

Fu Peng stated that the core contradiction in the current AI industry is "the road is built, waiting for the cars to run". The upstream computing infrastructure investment is basically complete, and 2026 will be the year to verify whether downstream enterprise-level applications can land and deliver profits.

He also said that in 2026, investors should focus on Tesla. Next year, it will face a "moment of identity verification" similar to Nvidia in those years: is it just a car company, or is it truly a carrier of enterprise-level "heavy AI applications"? Fu Peng pointed out that this is like testing "whether there are cars running after the highway is built". If Tesla can prove its value as an AI application, its market capitalization will have huge room; otherwise, judged by the logic of a car stock, its valuation is not attractive.

Fu Peng also emphasized that if AI is falsified, global stock markets will face drastic volatility. Currently, the US stock market (especially the AI sector) is the core of global "productivity"; the volatility of major global assets is highly tied to it. If AI is ultimately proven to be a bubble, not only US stocks but also global equity markets including Japan and Europe will collapse—“all grasshoppers on one rope”.

He believes that whether interest rates rise or fall is no longer important; the key is whether the asset side (AI) can generate real returns. If the asset side has a problem, adjustments to the liability side are futile.

The following is the live transcript of the speech:

The Interlinkage of Productivity, Production Relations, and Institutional Order

The underlying logic of this topic was explored in the chapter "Witnessing the Countercurrent" and also corresponds to the AJR model of 2024 Nobel Laureate Acemoglu—focusing on the interaction between productivity and production relations, especially landing on "institution and order" as this special production relation.

"Order" is mostly used between countries (e.g., trade, finance, security as referenced by Kissinger's World Order), while "institution" is commonly seen in internal corporate rules (e.g., clocking in for attendance). In essence, both are special forms of production relations. What we discuss today is the interlinkage between productivity, production relations, and institutional order. Many people mistakenly think macroeconomic indicators are the "barometer" of the stock market. But in my view, what the stock market truly reflects is total factor productivity (TFP)—the efficiency at which economic systems convert production factors into output.

This process is like a set of gears: productivity drives production relations, production relations reshape institutional order, and institutional order in turn drives productivity. The efficiency of gear rotation is TFP.

Extensive research (including Federal Reserve papers) confirms that the long-term trend of most national stock markets closely matches changes in TFP.

Take US stocks as an example. Since 1929, the core driver of their long-term upward trend has always been the improvement in economic efficiency—not short-term economic fluctuations. This improvement can come from any part of the gear: technological breakthrough, optimization of production relations, or institutional adjustments (such as corporate governance reforms).

In my commonly used "Numerator x Denominator x G" stock market model, G stands for this dimension of institution and order. The development of the US capital market also confirms: from Sarbanes-Oxley Act to shareholder activism, institutional optimization has always been the key to market health in the long run.

It must be emphasized, no part is perfect. Technology is a double-edged sword, productivity, production relations, and institutional order all have a dual nature. True "perfection" lies in forming an evolutionary mechanism where the good drives out the bad: good institutions can eliminate bad ones, and the system moves forward through error correction.

Industry Life Cycle Perspective: From Casting a Wide Net to Sifting Out the Genuine

Back to the topic of AI. 2015 and 2016 were critical nodes: not only the starting point for US stocks to break out of years-long broad-range volatility toward a trending market, but also the pivotal moment when the market realized US economic efficiency would leap forward.

At that time, Cathie Wood left her institution to establish her own firm. She is often called the "female Buffett", but her logic is entirely different—she runs a primary market growth stock investment strategy in the secondary market. This relates to Perez's "industry lifecycle" theory: real industrial investment often starts in the primary market, with the secondary market seeing what will emerge from the primary market in the future.

In the early stage of the industry, no one can predict which technical path will win. The optimal strategy is thus broad allocation—like Cathie Wood’s approach, incorporating all technical paths into the portfolio. This is the core logic of venture capital: invest in 100 projects, let 90 fail, and 10 succeed, that's success.

This strategy works well in the industry's early-stage valuation expansion, allowing one to enjoy all track benefits. But when the industry matures, the market inevitably has to sift out the genuine: money shifts from the 90 eliminated projects to the 10 real winners. Continued diversification leads to underperformance in returns.

The valuation plunge in 2022 was this very process of sifting out the genuine. Nvidia dropped 70%, Bitcoin fell from 80K to 20K, all valuation type assets saw deep corrections. The core of this round of adjustment was to force the industry to answer: for example Nvidia must prove it isn’t just a gaming GPU company but a provider of AI computing infrastructure.

And at the end of 2022 and early 2023, the emergence of ChatGPT marked that the market clarified a handful of winning tracks from numerous technical paths. Nvidia responded with subsequent earnings reports, confirming its AI era core position—"to get rich, build roads first; to build roads, buy shovels first", Nvidia's shovel became a symbol of certainty.

Volatility and Market Risk: Greater Certainty, Greater Risk

When analyzing the market, volatility is a core indicator. It is the opposite of certainty: the higher the uncertainty, the greater the volatility; the stronger the certainty, the lower the volatility.

After Nvidia dropped 70% in 2022, the market gradually confirmed that AI would bring huge capital expenditure, and its performance was gradually fulfilled. From then until 2023 and 2024, volatility kept falling—showing growing market consensus and very high certainty. But the problem lies precisely in "too much certainty": extreme certainty breeds greed, with leverage, private financing, mortgaging houses and cars to go all-in becoming more and more common.

On June 14, 2024, Fu Peng reminded in the 20th issue of Wallstreetcn "Fu Peng Talks" column: Nvidia should consider buying insurance. After volatility rose in August, Fu Peng promptly shared coping strategies.

To understand: much of the content in the "Fu Peng Talks" column is aimed at regular investors. As you are not financial institutions, you cannot attend quarterly offline exchanges via broker channels. Fu Peng's professional content is mainly in the column, not on short video platforms—those are just casual talk, the in-depth analysis and opinions are here.

As expected, Nvidia’s "flash crash" in 2024 validated this logic. Many analyses attributed it to “unwinding yen carry trades”, but as I see it, there’s only one core reason: global assets have all been tied to AI as a "productivity asset"; when asset-side certainty is overdrawn, any change on the liability side is just a trigger.

This is why I always emphasize: don't just watch the liability side, look at the asset side. If AI is proven to be a bubble, global markets will crash; at that point, hikes or cuts won't help. If AI delivers real productive value, then the market rally will be solidly based.

AI’s “Road Building” and “Cars Running”—Transmission from Productivity to Production Relations

After Nvidia’s flash crash, the market has been asking: is AI a bubble? The essence of this question is similar to what Xie Guozhong debated in 2002-2003 about China's infrastructure.

People once said building expressways was a waste, debt; but facts proved "to get rich, build roads first"—infrastructure drove urbanization and economic growth. Now, the AI industry is at a critical juncture of "highways built, but are there cars running".

In recent years, trillions of dollars have been invested in upstream AI infrastructure—the "highways" of computing power, energy, etc.—but true enterprise-level AI applications—the "cars"—have not yet rolled out on a large scale. Existing ChatGPT, text-to-image, image-to-text, etc., are just superficial applications, far from the core apps that can truly drive productive transformation.

The market’s doubts and waiting essentially are for one answer: Are these AI infrastructures assets that can drive economic growth, or debts that fail to yield returns? The answer will determine the direction of global assets.

Looking at the yield curve also reveals the Fed's "preventive operations": after Nvidia’s flash crash, the US Treasury three-month minus ten-year spread inverted rapidly, and each inversion matched a drop in volatility. Behind this, the Fed uses short-end liquidity adjustment to prevent systemic risk spread and buys time for AI application rollout.

But this operation cuts both ways: the good is that it slows market collapse, the bad is that valuations get higher. By the end of this year and start of next year, this problem can no longer be suppressed.

Next year will be the year to prove or falsify the transmission from productivity to production relations via AI.

Tesla is the key stock in this proof process. Just like Nvidia in 2021-2022 needed to prove it was a computing power provider, not a graphics card company, Tesla next year must prove: is it just a car company, or is it an enterprise-level heavy AI application platform? The answer is worlds apart in valuation.

If it’s just a car company, trillion-dollar market cap is already overdrawn; if it’s an AI application platform, a trillion is only the beginning.

Currently, US stocks (especially AI sector) are the core of global "productivity", and volatility of major global assets is highly tied to them. If AI is ultimately proven to be a bubble, not only US stocks but also global markets in Japan, Europe, etc. will crash—“all grasshoppers on one rope”.

Whether interest rates rise or fall is no longer important; the key is whether the asset side (AI) can generate real returns. If the asset side has a problem, adjustments to the liability side are futile.

Two Paths and the Opportunities of the Era

Back to the original question: Is AI a bubble? Next year’s yield curve has only two paths:

The first is the path of falsification: if upstream AI infrastructure cannot transform into downstream productive applications, investments in the past years will all become debt, global markets will crash, no asset will be spared.

The second is the path of confirmation: if AI completes the transmission from "road building" to "cars running", true productivity drives change in production relations, we will usher in a second wave—not only the wealth created by productivity, but systemic opportunities for production relation optimization and institutional order reform.

In each long cycle, there are three big opportunities: productivity improvement, production relation change, and institutional order reconstruction. In one's lifetime, catching even one gear’s cycle is already fortunate. Nvidia has proved itself as a definitive productivity asset; in the future it will become a mature growth stock. The next opportunity lies in transformation of production relations—i.e., the rollout and spread of AI applications.

This is the era node we are at: Either we witness the collapse of a productivity revolution, or experience the rise of a reconstruction of production relations. The answer is hidden in next year's market verification.

Risk Warning and DisclaimerThe market has risks, investment needs caution. This article does not constitute personal investment advice and does not consider the specific investment goals, financial status, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article fit their circumstances. Investment according to this is at your own risk.