Hook: The Signal in the Red

The numbers hit me like a static burst on a clean frequency. MiniMax's short interest at 20%. Zhipu AI's stock down 54% from its peak. Combined lock-up expiries totaling $11.5 billion in market value. These aren't just numbers on a screen — they're the market's collective verdict on an entire generation of Chinese AI companies.

I've spent nine years in this industry, watching narratives form, crystallize, and collapse. The current situation with China's "AI Four Little Dragons" feels different. It's not a typical tech correction. It's a structural re-evaluation of what an AI company's stock should actually be worth — and the market is speaking in a language that founders and early investors are struggling to hear.

Part I: The Signal in the Static

The Data That Tells a Story

Let me walk you through what's happening with MiniMax and Zhipu AI, two of China's most prominent independent AI labs that have chosen the Hong Kong stock market as their public battleground.

The facts are stark. Since the July release of Kimi K3 — a competitive model from rival Moonshot AI — Zhipu and MiniMax have dropped approximately 24% and 18% respectively. That's not just a correction; that's a signal being read loudly and clearly across the market.

But here's the detail that keeps me up at night: the market's reaction isn't about the technology itself. Kimi K3 is a genuinely impressive model. The problem is what it represents — an escalating arms race where every new release demands more capital, more computing power, and more aggressive pricing strategies. For investors, each new model release isn't a breakthrough; it's a bill.

The Valuation Gap That Can't Be Ignored

The numbers paint a troubling picture:

• Zhipu's stock remains 800% above its IPO price — but it's crashed over 50% from its peak
• MiniMax is trading 80% above its IPO price, yet has fallen even harder
• The lock-up expiry period is flooding the market with shares

When I look at these metrics through my analytical framework, the first thing that strikes me is the absence of an anchor. In the crypto world, we call this a "float inflation event." The supply of available shares is about to increase dramatically, and demand is questioning whether these companies will ever actually be profitable.

Part 2: The Hollow Core

When Technology Becomes Commodity

The seven-dimensional analysis framework reveals something fascinating: the technology advantage that once justified high valuations is eroding. Let me break down what the data shows.

Zhipu's GLM-5.3 model, for instance, has been praised by Jefferies for matching Kimi K3's performance at 19% lower cost per task. This is a genuine engineering achievement. It demonstrates that Zhipu has excellent optimization capabilities and cost control. The market's response? A collective shrug.

Why? Because in the current environment, technological superiority no longer translates directly to shareholder value. The market has entered a phase where the investment thesis has shifted from "who has the best model" to "who can make money with any model."

The Costs of the Arms Race

The new reality is brutal: for companies like MiniMax and Zhipu to remain competitive, they must continue investing heavily in research and development. But these investments have a limited return period. Here's what I mean:

• The Technology Premium is Dead: When models reach parity, the technology itself is no longer a defensible moat
• The Price War Escalation: Both companies are being forced to lower prices to attract customers, squeezing already thin margins
• The Scale Disadvantage: Unlike DeepSeek or Alibaba's Qwen, these companies don't have the massive ecosystems or cash reserves of their larger competitors

This is the "middle ground" problem. They're too big to be nimble and too small to compete on scale. They're caught in a strategic bottleneck.

Part 3: The Perfect Storm

The Short Sellers' Logic

Hedgeye's bearish report on Zhipu and MiniMax isn't just a random attack. It's a well-reasoned thesis that draws on three key observations:

1. Pricing Pressure: The price war in China's AI market is real, and it's brutal. Customers are sensitive to cost, and the market has become a buyer's market.

1. Lack of Clear Profitability Path: When will these companies actually be profitable? The short answer is: nobody knows. And that uncertainty is a cancer on the stock price.

1. The Unlock Burden: The lock-up expirations in July represent a massive overhang on the stock price. Early investors have every incentive to cash out — and that's a clear signal that they don't have long-term faith.

The Lock-Up Shock

Let me give you a concrete example of what I'm talking about. Zhipu has approximately 25.68 million shares, and MiniMax has 150 million shares that are now free to trade. At current prices, that's approximately $11.5 billion in potential selling pressure.

To put this in perspective: this is the equivalent of a massive wave of selling that could hit the market at any moment. The smart money is getting out before the waves arrive.

The Southbound Money Dilemma

Here's something I find particularly interesting from a market structure perspective. Southbound capital — money flowing from mainland China into Hong Kong — is actually buying these stocks. Zhipu's Southbound holdings are about 12% of the float, and MiniMax's about 8.1%.

But the stock prices are still falling. This is the classic "value trap" pattern. There are buyers trying to "catch a falling knife," but they're not big enough to stop the downward momentum. The short sellers are playing the liquidity, and they're winning.

Part 4: The Broken Narrative

The Death of the "Story" Trade

The AI model in the market has a narrative structure that's familiar to anyone who's watched crypto go through its own boom and bust cycles. It goes something like this:

1. The Story Phase: "AI will transform everything" — huge growth expectations, price targets based on total addressable market
2. The Hype Phase: Models get released, valuations go parabolic
3. The Reality Phase: Revenue growth slows, the market demands actual profitability
4. The Correction Phase: Stock prices crash as investors realize the gap between expectation and reality

We're now deep in Phase 4 for these Chinese AI companies. The "story" no longer works. The market is demanding actual numbers — revenue growth, gross margins, customer acquisition costs — and the companies are struggling to provide them.

The Specific Problem

The core issue is the "one-dimensional" nature of these businesses. They are primarily API call providers and cloud services. They lack the high-margin revenue streams that would give them more financial flexibility.

Let me break this down:

| Revenue Model | MiniMax | Zhipu AI | |----------------|---------|----------| | API Calls | Primary | Primary | | Cloud Services | Secondary | Secondary | | Enterprise Solutions | Emerging | Emerging | | Other | Limited | Limited |

The reality is that the API market is becoming commoditized. The market is saturated with players — DeepSeek, Alibaba's Qwen, ByteDance's Doubao — all fighting for the same customer base. The pricing power is gone.

Part 5: The Mid-Ground Trap

The Strategic Vulnerability

Hedgeye's assessment that MiniMax is "neither the smartest nor the cheapest" is a painful but accurate diagnosis. These companies are caught in a dangerous position:

• At the top: DeepSeek and Alibaba have deeper pockets, more advanced technology, and massive ecosystem advantages
• At the bottom: smaller, more nimble AI companies are undercutting them on price

This is the "middle ground" — and it's the most dangerous place to be in any market.

The Ecosystem Problem

Let me give you a concrete example. Alibaba's Qwen has access to Alibaba's entire ecosystem — cloud computing, e-commerce, logistics, entertainment. It can integrate its models into products that serve millions of users.

MiniMax and Zhipu don't have that advantage. They're building models in isolation, hoping that their technology will be enough to attract customers. But in a world where the technology is becoming commoditized, the ecosystem is the moat.

The Talent Drain Risk

There's another layer to this: if these companies' stock prices remain depressed, it could trigger a talent exodus. When stock options are underwater, and the future looks uncertain, the best engineers and researchers will start looking for more stable opportunities at larger tech companies or more promising startups.

This creates a vicious cycle. The weaker the stock price gets, the harder it becomes to attract the talent needed to stay competitive. The harder it is to attract talent, the weaker the technology becomes. The weaker the technology becomes, the weaker the stock price gets.

Part 6: The Bigger Picture

The Industry's Reckoning

What's happening to MiniMax and Zhipu isn't just about these two companies. It's a warning for the entire AI industry. We're witnessing the transition from "technology investment" to "business investment." The era of "story-driven" valuations is over.

I've seen this before — in the crypto market, in the internet bubble, in every technological revolution. There's a moment when the narrative shifts from "what this could become" to "what this actually is."

The K-Shaped Future

The AI industry is moving toward a "K-shaped" future:

• The Winners: Companies with clear paths to profitability, or those backed by large tech giants
• The Losers: The second-tier players struggling to find their place in the market

This is going to be the hardest for the "second tier" — the companies that aren't the leaders but aren't the fast followers either. They're going to be squeezed.

The Ripple Effect

The impact of this is broader than just the stock market. It's going to affect the entire AI industry:

1. Venture Capital: Investors are going to be more cautious about funding AI companies that don't have clear revenue paths
2. The compute ecosystem: If these companies reduce their training budgets, that could slow growth in GPU and cloud computing demand
3. Application layer: The lowering of model costs could actually accelerate the application layer, as it becomes cheaper to build AI-powered products

Part 7: The July 26th Moment

The Earnings Wait

The next big moment for these companies is going to be their half-year earnings reports. MiniMax's earnings are expected around August 26, and Zhipu's around August 31.

This is going to be the "Judgment Day" — the moment when we get to see whether the market's pessimistic view is justified or whether these companies have more to offer.

I'm looking for three things in these earnings:

1. Revenue Growth: Is the top line still growing? How fast?
2. Gross Margin: Are they able to maintain pricing power, or are they being squeezed?
3. Cash Runway: How much cash do they have left, and how fast are they burning it?

The Market's Expectations

The market's expectations are low. The short sellers are betting on the idea that the fundamentals are going to be worse than expected. The Southbound capital is betting on the long-term story.

This is the tension that's going to determine the direction of these stocks in the short term.

Part 8: The Contrarian View

The Other Side of the Argument

I'm not just a bear. Let me give you the other side of the argument.

The Bull Case: These companies are actually at the front of AI technology. They have some of the best models in China. The market is undervaluing them because it's looking at short-term issues — pricing pressure, lock-up expirations — rather than long-term potential.

• Zhipu's cost advantage: If Zhipu can maintain its cost advantage, it could be a key player in the price war, attracting customers from higher-priced competitors.
• MiniMax's product focus: MiniMax has been building products, not just models. If they can leverage their strengths in consumer applications, they could create a more sustainable revenue stream.

The Potential for Mergers and Acquisitions

Another possibility: these companies could become acquisition targets. If the stock prices continue to fall, they might look attractive to large tech companies — either Chinese tech giants or global players — looking to acquire technology and talent at a discount.

This would create a "valuation reset" moment. The stock price might drop even further, but then it would be acquired at a premium, creating a quick rebound for investors who got in at the bottom.

Part 9: The Human Layer

The Emotional Factor

I've spent a lot of time talking about numbers and market dynamics. But there's a human layer to this that I can't ignore.

These companies are built by people who genuinely believe in the potential of AI to transform the world. They're not just building companies — they're building what they think is the future. And they're watching their life's work being valued lower than they believe it's worth.

I've been in this position before. In 2022, during the FTX collapse, I watched the entire crypto ecosystem that I had believed in come crashing down. It was a painful experience, but it also gave me a valuable lesson: the market is not always right, but it's always the final judge.

The Post-Speculative Era

We're entering a "post-speculative" phase in AI. The market is no longer interested in stories. It's interested in the numbers. It's a hard adjustment for those of us who are believers in technology first.

But I also think there's an opportunity here. If these companies can show real numbers — real revenue, real growth, real paths to profitability — the market will reward them. The market isn't against AI. The market is against the gap between the story and the reality.

Part 10: The Next Chapter

The Signal in the Noise

What do I think is really happening here? I think we're watching the market's fundamental shift in how it values AI companies. We're moving from "market cap" to "price-to-earnings" — from "technology leadership" to "business model sustainability."

This is a painful transition, but it's a necessary one. The AI industry can't sustain itself on promises forever. At some point, the numbers have to work.

The Timeline

Here's what I'm watching:

| Time Horizon | What to Watch | Why It Matters | |---|---|---| | 1-2 Weeks | Half-year earnings reports | Revenue growth, gross margins, losses | | 1-3 Months | Lock-up selling volume | Whether early investors are selling | | 3-6 Months | Next-gen model releases | Whether the technology is still competitive | | 6-12 Months | AI regulatory policy | Whether the government is supportive or restrictive |

The Investment Thesis

For investors, the key question is: are these companies worth buying at these levels?

I can't give you a simple yes or no answer. The market is uncertain. But I can tell you what I'm looking at:

1. The earnings report will be the biggest catalyst. If the numbers are better than expected, we could see a massive short squeeze.
2. The lock-up selling is going to be a challenge. If the big shareholders hold their shares, that's a signal of confidence.
3. The technology roadmap matters. If either company can deliver a game-changing model that's clearly ahead of the competition, the market might start paying attention to the technology again.

Part 11: The Final Verdict

The Market's Message

The market has spoken: "We don't believe the story. We want to see the numbers."

This isn't the end of the AI revolution. It's the end of the "story-driven" investment phase. The AI revolution is going to continue, but the winners are going to be the companies that can build a sustainable business — not just the ones with the best models.

A New Framework

For me, this is a shift in the "investment" narrative. I've always believed that the "narrative" — the story of what a technology can become — is a powerful force in markets. But I've also learned that the narrative must eventually be backed by reality.

The companies that can bridge the gap between story and reality will be the winners. The ones that can't will be left behind.

Conclusion: Finding the Signal in the Static of the New Wave

I've watched a lot of market cycles in my career. I've seen the boom and bust of crypto, the rise and fall of countless tech startups, and the way that the market's perception of value can shift in a moment.

The situation with MiniMax and Zhipu AI is a classic "narrative shift" moment. The old narrative — "AI will change the world, and these companies are the future" — is dying. The new narrative — "show me the numbers" — is beginning to take hold.

The question is: which of these companies can adapt to the new narrative? The ones that can find a way to be profitable, to build a sustainable business, and to generate real value for their shareholders will be the ones that survive. The ones that can't, will fade into history.

I'm not sure which path they'll take. But I know that the next few months will be critical. The earnings reports will tell us a lot. The lock-up selling will tell us even more.

Finding the signal in the static of the new wave. That's what I do.

This is a narrative-driven analysis of the market events surrounding MiniMax and Zhipu AI. The information presented is based on public market data and the views of market participants. This is not investment advice — it's a story about a story, an analysis of the market's shifting narrative as it moves from the era of "story" to the era of "numbers." The truth lies somewhere in the middle, and only time will tell which side of the story the market is right.

The Next Chapter

As the August earnings reports approach, I'll be watching with intensity. The AI industry is about to enter a new phase — the phase of the "hard numbers." I want to see which companies can rise to the challenge.

The story of AI is still being written. The question is: who will be the authors, and who will be the footnote?