Why have U.S. stocks surged for ten years while A-shares remain around 4,000 points?

Recently, a subscribed fan brother asked me, with some idle money, wanting to invest in AI-related stocks, whether to buy A-shares or U.S. stocks.
I said, your question is like asking whether to buy Tesla or Zhongtai.
He paused and said, is it that exaggerated?
Then I gave a comprehensive overview:
First, about the players in the AI track.
On the U.S. side, OpenAI, Anthropic, Google DeepMind—three major model giants—all based in the U.S.
NVIDIA alone accounts for over 80% of the global AI chip market share.
Microsoft, Amazon, Google—the three cloud infrastructure giants—spend billions annually building data centers.
On the A-shares side, there are quite a few AI concept stocks.
But if you look at their financial reports, few making money from AI, while those riding the concept are raking in profits.
I have a friend, born in the 90s, a programmer in Shenzhen.
Last year, when AI was hot, he bought a certain big model concept stock.
I won’t mention the name, but the announcement was full of grandiose claims.
He was quite excited when telling me, saying this time it’s about domestic substitution, holding long-term.
I looked at that company's financials.
R&D investment: 70 million yuan.
Marketing expenses: 600 million yuan.
I couldn’t bear to burst his bubble.
-----
Continuing on this topic, let’s talk about where the money flows.
In 2024, over 80% of global AI venture capital flows into U.S. companies.
China’s share is less than 10%. Crunchbase data shows it clearly.
Money is the smartest. It knows where things can grow, and where they can only grow PPT presentations.
Speaking of valuation, this part is even more interesting.
U.S. stocks value AI companies based on “discounted future cash flow.”
NVIDIA’s PE ratio is 40, and the market believes it can still grow because demand for AI chips is structural, not cyclical.
A-shares value AI companies based on “concept hype.” If a company announces plans to develop large models, its stock can hit daily limit-ups.
But if you look at their R&D investment, it might be less than a Silicon Valley Series A startup.
Many criticize the U.S. stock market for being bubble-prone.
But if you look at NVIDIA’s financials, quarterly revenue growth exceeds 200%, with a gross margin over 70%.
Such growth, which A-shares can’t match.
Conversely, many praise A-shares for low valuation, saying the Shanghai Composite PE is only 12.
But look at how that 12 is composed.
Banks PE around 5, white wine companies PE around 20, tech stocks PE around 50.
The undervalued parts are low-growth; the overvalued parts are bubbles.
This isn’t low valuation.
It’s valuation distortion.
Back to the market itself.
U.S. stocks are dominated by institutions—pension funds, insurance funds, index funds—these tend to “buy and hold.”
So, the U.S. market has a foundation for “slow bull” trends.
A-shares are retail-dominated, with the highest turnover rate globally.
Today chasing this hot topic, tomorrow chasing that concept.
4,000 points isn’t the bottom; it’s the median of sentiment.
I have to admit, I’ve also fallen into traps myself.
In early years, I chased a few hot topics in A-shares, each time thinking it was a long-term investment, only to realize I was just gambling.
You’re not investing in companies; you’re guessing what price the next person will buy your shares at.
The top ten weights in the S&P 500 are Apple, Microsoft, NVIDIA, Amazon, Google, Tesla, Meta.
Seven out of ten are tech giants.
The top ten weights in the Shanghai Composite are Moutai, ICBC, ABC, PetroChina, Bank of China, China Life, CCB.
Five are banks and insurers, two are white wine, one is oil.
A stock index packed with so many banks and white wine companies—do you expect it to represent the future?
-----
This reminds me of a book, “Madness, Panic, and Crashes.”
The South Sea Bubble in Britain in 1720, where people bought stocks for reasons exactly the same as today. “This time is different,” is one of the five most expensive phrases in financial history.
I know many will say, A-shares also have good companies—consumption, pharmaceuticals, new energy, that’s true. I’m not telling you to completely clear your A-share positions.
But if you want to invest in AI, want to participate in the biggest technological revolution of this era, you must admit that the soil is different, and what grows out of it is different.
It’s not anyone’s fault. Different institutional designs, different cultural genes.
The U.S. capital market has a 200-year history, forming a “long-termism” investment culture.
A-shares have a 30-year history, dominated by retail investors, with “speculation” as the main theme.
But the problem is, when you want to invest in AI— a track that requires long-term commitment—A-shares’ soil simply can’t grow big trees.
The scariest part isn’t these gaps.
It’s that many people haven’t realized it yet.
They’re still chasing hot stocks in A-shares, still researching which concept can ride the AI wave, still believing that “domestic substitution” stories can support trillion-yuan market caps.
Meanwhile, U.S. AI giants are building the next-generation infrastructure with real money.
NVIDIA’s chips, Microsoft’s cloud, OpenAI’s models—these aren’t concepts; they’re real moats.
-----
Looking back twenty years later, it might be this scene.
Global AI applications run on U.S. chips, on U.S. clouds, on U.S. models.
And all the “tolls” are being collected by those few American companies.
I’m not trying to sound pessimistic about A-shares.
I just think that in investing, understanding reality is more important than blind patriotism.
After all, your money won’t automatically grow just because you love your country.
Finally, I recommend ETFs—here’s the link!!