Qiu Dongrong's Comeback Roadshow: Doubts About Capital Returns on AI Investments, Not the Same Thing He Discussed With Jensen Huang

By Company Research Lab, Author: Zhang Yang

"This kind of business, this kind of investment, I think is very bad and very dangerous."

"If you're conservative, don't invest in it; if you're aggressive, just find an opportunity to short it."

The person saying this is not some short-selling institution, but one of the most famous standard-bearers of A-share "low-valuation value investing" over the past decade.

What's even more surprising is that he also rejected his own former specialty.

Qiu Dongrong said that low-valuation things are usually problematic—if he were to short with his eyes closed, he would short low valuations, "because it is a landmine."

After leaving Zhonggeng Fund in July 2024, he almost disappeared from public view. This nearly two-hour high-intensity roadshow was his first systematic public statement in two years.

Beyond the comeback, what's truly worth hearing are his counter-consensus judgments.

From AI and the internet to banks, coal, and oil, and then to A-shares and Hong Kong stocks—almost every one of them runs contrary to the current market consensus.

One says the money put in ultimately won't be earned back.

One says compute power still isn't enough to sell.

But what they're talking about is actually not the same thing at all. And that is the question this roadshow truly aims to answer.

Low valuations are usually "landmines"

Qiu Dongrong first clarified his difference from traditional low-valuation investors.

"Low-valuation things are usually problematic—so we don't short high valuations; we'd rather short low valuations. If you made me short with my eyes closed, I'd short low-valuation companies; it's easier, because it is a landmine."

In his view, low-valuation investing has two common pitfalls.

One is believing in mean reversion—thinking that once an asset falls, it will definitely rise back in the future; the other is over-believing in cycles—thinking that a company that has fallen enough and is cheap enough is worth buying.

But Qiu Dongrong believes that the vast majority of companies and assets do not have such cyclical recurrence, "their fate is to go to zero." The logic of "buy it because it's cheap and has fallen a lot" does not hold in itself.

What one should truly look for is the combination of low valuation, low risk, and high-quality assets.

"If certainty is upward, you're willing to give it a 15x or 20x valuation, even if it's currently at 10x; but if its certainty is downward, that's another matter." This is not a simple valuation question, but a risk-pricing question.

His attitude is very clear: "This kind of business, this kind of investment, I think is very bad and very dangerous."

First, capital investment is "very, very large"; second, the return business model is still fragile, and at least for now it's unknown how to make money; third, asset prices are already very high, yet financing continues.

Qiu Dongrong believes the problem is not whether the valuation is 100x, 40x, or 20x, but whether this business model itself is worth investing in.

Qiu Dongrong said bluntly: "If you're conservative, don't invest in it; if you're aggressive, just find an opportunity to short these things."

As for specific companies, among domestic internet companies Qiu Dongrong is more optimistic about Pinduoduo.

In his view, Pinduoduo has not invested in AI on a large scale, its core business is limited in its direct exposure to AI's impact, and more importantly, it is essentially not a tech company but a retail company, whose real competitiveness remains supply-chain efficiency, cost, and the ability to provide consumers with quality goods at low prices.

New technologies may ultimately "kill both growth and value"

Qiu Dongrong believes that technological success does not naturally mean investment success. "The market will severely overestimate the short-term impact of new technologies, but the market will severely underestimate the long-term impact of new technologies."

He took new energy vehicles as an example. Today's development of new energy vehicles has actually surpassed the most optimistic stories many companies told when they were founded. Back then, few companies dared to say they could reach 1 million vehicles in the future; now some companies have delivered on it.

But industrial and technological success does not mean all investors have earned corresponding returns. Today, the market value of some new energy vehicle companies is even only one-tenth of what it was when they were "boasting."

At the same time, the truly enormous destructive power of new technologies may fall on traditional industries.

Five years ago, coal, oil, and natural gas were once among Qiu Dongrong's heaviest positions, but now he has turned bearish, and some companies have even entered what he considers "shortable" territory.

Qiu Dongrong believes that growth industries may fail to make money because of excessive capital investment and fierce competition, but new technologies will at the same time destroy the original profit pools of traditional industries through efficiency gains.

Growth stocks don't necessarily make money, and low-valuation traditional value stocks are likewise not necessarily safe.

USD-priced Chinese assets are a value depression

If there are risks in tech, new energy, and even traditional energy, what is still worth investing in now?

Qiu Dongrong's answer is: "USD-priced Chinese assets." He even calls them the world's only "depression."

This opportunity must meet two conditions at the same time: first, Chinese assets, with corporate supply-demand and profits mainly coming from China's domestic demand; second, they must be priced in USD.

Qiu Dongrong believes that from 2021 to now, Chinese companies have gone through years of contraction, including multiple factors such as the macroeconomic downturn, the industry's own adjustments, and broad-based regulation.

The result of the contraction is that future supply may decrease, competition will no longer be so brutal, and companies' ROIC may instead have a chance to rise again.

But the problem is that the RMB system has low interest rates and ample liquidity, so many A-share companies, even if their stock prices have fallen quite a bit, still don't have valuations that are truly cheap.

By contrast, USD interest rates are higher, and overseas investors can also choose a large number of US assets such as Nvidia and Tesla, so capital continues to flow out of some Chinese assets, which instead creates genuine undervaluation.

Qiu Dongrong also specifically emphasized that what they "don't do is predict stock prices."

Buying an asset is not because of a judgment that "it will rise"; what they truly care about are only two questions: first, can this asset continuously create cash flow and earn more and more money; second, is the purchase price cheap enough, and is the implied return high enough.

English text is being prepared.

AI may of course continue to develop rapidly and may even ultimately change the world. But if the entire industry needs to keep investing huge amounts of capital and competition keeps intensifying, then of the money invested today, how much can ultimately be converted into corporate profit, and how much can truly become the returns shareholders receive?

Qiu Dongrong and Jensen Huang appear to be sharply at odds, but in fact they are not discussing the same question.

Qiu Dongrong is answering an investment question: even if demand is real and the technology succeeds, after investing such enormous capital, can shareholders ultimately still obtain a sufficiently high capital return?

AI may not have a bubble, compute demand may continue to grow rapidly, and AI may truly change the global economy; but at the same time, some AI assets may still, because of excessive capital investment, overly fierce competition, or excessively high valuations, ultimately fail to bring investors returns commensurate with industry growth.