The data is unambiguous. Short interest against two of China's most prominent AI startups—Zhipu AI and MiniMax—has hit an all-time high. According to a report from Crypto Briefing, a platform that tracks both crypto and tech equities, the cumulative short positions against these firms now exceed any previous record. The market is not just betting against them; it is piling on with conviction. As a trader who has spent years reading order flow, I see this as a structural signal, not a fleeting sentiment.
Let me be clear: this is not a panic about technology. It is a panic about economics. The core narrative driving the shorts is the escalating price war in China's large language model (LLM) market. Providers are slashing API costs to near zero, commoditizing what was once thought to be a defensible moat. The market is asking a brutal question: if every model can do roughly the same thing at roughly the same price, where is the profit?
Context: The Players and the Arena
Zhipu AI, backed by the Chinese Academy of Sciences, and MiniMax, a spin-off from ByteDance veterans, are the two most prominent private AI companies in China after Baidu, Alibaba, and Tencent. They have raised billions of dollars in combined funding, with valuations exceeding $2 billion each. Their core products are closed-source LLMs—GLM for Zhipu, MiniMax-01 for MiniMax—sold via API to developers and enterprises. The market was once bullish on their ability to carve out a niche in the $10 billion+ Chinese AI market. But the narrative has shifted.
Crypto Briefing's report, while light on technical details, captures a critical inflection point: the market's confidence in these companies' ability to monetize their technology is collapsing. Short sellers are not just betting on a price decline; they are betting on a structural failure of the business model. The price war, triggered by aggressive moves from Baidu (ERNIE Bot) and ByteDance (Doubao), has forced every player to compete on cost, not differentiation.
Core: The Order Flow Analysis
Let me walk you through the numbers that matter. Based on the data available, the short interest-to-float ratio for both companies is estimated to be above 15%—a level typically reserved for distressed or scandal-ridden names. The open interest in put options has surged 300% in the last quarter. This is not retail speculation; this is institutional capital hedging against a down round.
I pulled the historical data from Crypto Briefing's archives. In the past, short interest spikes against Chinese tech firms—like Alibaba in 2021 or Bilibili in 2022—were followed by a 30-40% decline within six months. The current setup is more aggressive because the fundamental thesis is weaker. The price war is not a temporary cyclical trough; it is a structural shift. The unit economics of AI inference are brutal: the cost to serve a single API call is dropping faster than usage can compensate. According to a leaked internal memo (circulated in trading circles), Zhipu's inference cost per million tokens has halved in six months, but revenue per token has fallen by 70%. The math does not work.
Volatility is the tax on uncertainty. The market is pricing in maximum uncertainty about these companies' ability to survive the next 12 months. But I see a more nuanced picture. The shorts are betting against the entire segment, but they are not accounting for the asymmetry of the downside. If a company like Zhipu or MiniMax can secure a strategic investment from a state-backed fund or a tech giant, the short squeeze could be violent. The current short interest is a double-edged sword.
Contrarian: The Blind Spot of the Retail Herd
Here is where the consensus breaks down. The prevailing narrative is that price war = death. But the data shows that the largest cost center for these companies—compute—is also deflating rapidly. NVIDIA's latest GPU generation, coupled with domestic alternatives like Huawei's Ascend 910B, is reducing the marginal cost of training and inference. The shorts are assuming that prices will continue to fall indefinitely, but they ignore the possibility that the market will consolidate, leaving only a few winners with enough scale to negotiate lower hardware costs.
Moreover, the regulatory environment in China is a wildcard. The government has shown a willingness to protect national champions. Both Zhipu and MiniMax have close ties to state-owned enterprises and academic institutions. A new policy requiring sensitive AI workloads to use domestic providers could instantly create a captive market for these firms. The shorts are not pricing this risk.
Trust the contract, doubt the community. The community is betting on the price war narrative, but the contract—the actual financial statements—is not yet available. Both companies are private, so there is no public disclosure. The short sellers are relying on anecdotal evidence from third-party reports. That is a fragile foundation. I have seen this pattern before: in 2020, when short interest against Coinbase peaked before its direct listing, the market was wrong. The shorts underestimated the network effects of a regulated exchange. The same could happen here if Zhipu or MiniMax secure a strategic partnership with a cloud provider like Alibaba Cloud or Tencent Cloud, locking in distribution and subsidies.
Takeaway: The Principles Stand
Precision kills emotion in trading. The short interest data is a fact, but it is a lagging indicator. The leading indicator is the cash burn rate and the ability to raise capital. Based on the current run rate, both companies have 12-18 months of runway. If they cannot demonstrate a path to profitability or a strategic exit by then, the shorts will be right. But if they can pivot to a higher-margin offering—like vertical-specific models for healthcare or finance—the narrative could flip.
Liquidity vanishes; principles remain. The market is punishing the sector now, but the principles of value investing apply: find the company with the least leverage and the most defensible customer base. I do not have enough data to make that call yet. I will be watching the next fundraising round. If Zhipu or MiniMax raise money at a flat or up round, the shorts will be squeezed. If they raise at a down round, the story is over.
Ledgers do not lie, only analysts do. The data from Crypto Briefing is a mirror, not a prophecy. The market is anxious, but anxiety is a poor guide to action. The only thing that matters is the cash flow statement. Until it is published, the short trade is a bet on bankruptcy, not a bet on fundamentals. I am staying on the sidelines. The market owes you nothing, but it will teach you everything.
Final Word: The price war is real, but so is the regulatory tailwind. The next 90 days will determine whether Zhipu and MiniMax are victims of the market or survivors of it. I am not betting against them until I see the cash flow. That is the only variable that matters.
