Record Short Bets on Chinese AI Giants Signal a Deeper Crisis of Value

The numbers surged, but the room felt quiet. That is the sensation I get when I read about record short bets piling up against Zhipu AI and MiniMax. These are two of China's most prominent AI startups, companies that raised billions on the promise of building the next generation of large language models. And yet, the market is now screaming that their valuations are hollow. As someone who has spent years inside the machinery of decentralized protocols, watching the churn of speculative capital and the collapse of projects that mistook hype for substance, this story feels hauntingly familiar. The graph of short interest is spiking, but the underlying business model remains silent. This is not merely a story about two companies. It is a story about what happens when an entire industry mistakes a price war for a competitive strategy, and when investors begin to realize that the emperor, despite all his fine clothes, is standing in the cold. Let me set the stage for those who have not been following the Chinese AI landscape. Zhipu AI, backed by the Beijing government and tech giants, has been a darling of the national AI push. Its GLM series of models has been positioned as a domestic alternative to OpenAI's offerings. MiniMax, on the other hand, has carved a niche in consumer-facing applications, particularly in short-form video and interactive entertainment. Both companies have been engaged in an aggressive, public price war with behemoths like Baidu, Alibaba, and ByteDance. The strategy is simple: slash API prices to capture market share, starve out the competition, and worry about profitability later. For a while, this seemed to work. Adoption numbers rose, and the narrative of a booming Chinese AI sector persisted. But the shorts are now calling the bluff. They are betting that this strategy is unsustainable, that the unit economics are fundamentally broken, and that the revenue generated by these companies will never justify the valuations assigned to them. This is not a technical assessment; it is a cold, hard look at the balance sheet. And it is a look that the market is failing. Now, let me apply the lens I have developed over years of auditing decentralized finance protocols, because the parallels are striking. In the DeFi world, we had a phenomenon called liquidity mining. Projects would print their own tokens and distribute them to users who provided liquidity, creating a temporary surge in total value locked (TVL). The graphs looked beautiful. The metrics were dazzling. But the moment the incentives were reduced or stopped, the users vanished, and the TVL collapsed, revealing the project to be a hollow shell. The APY was not a sign of value creation; it was a subsidy paid to rent superficial growth. I see the exact same dynamics at play in the Chinese AI price war. These companies are effectively subsidizing their API usage. They are paying, through forgone revenue, to attract developers and enterprises. They are renting their user numbers. The question that the short sellers are asking, and that I believe is the right one, is this: what happens when the subsidies run out? What happens when investors demand profitability? The answer, based on my experience, is that the real users will disappear, and the underlying value proposition will be exposed as thin. The price war is not a sign of strength; it is a sign of desperation. It is the death rattle of a commoditized market where no one has built a true moat. The core of my concern lies in the conflation of technology with value. During my time at Gitcoin, I audited smart contracts designed for quadratic voting and public goods funding. I saw firsthand how code could be structured to enforce fairness and empower communities. But I also saw how easily that code could be perverted when the incentives were misaligned. The technology was never the problem; the economic model was. The same is true here. Both Zhipu and MiniMax have, by most accounts, competent engineering teams. They have released models that perform admirably on benchmarks. But a good model is not a business. A good model that costs more to serve than it can ever charge is a liability. The shorts are not betting against the technology; they are betting against the economics. They are betting that the cost of training and serving these models, coupled with the aggressive price cuts, will bleed these companies dry. They are betting that the capital required to keep pace with Baidu and Alibaba, who have massive cloud businesses to subsidize their AI efforts, is simply not available to these startups. And they are betting that the market will eventually wake up to this reality. The record short interest is not a blip; it is a signal that the market is starting to price in a future that these companies have not yet acknowledged. But let me take a contrarian stance, as I often must. Is it possible that the shorts are wrong? Is it possible that they are missing something fundamental? In the crypto world, we saw a similar phenomenon with Bitcoin. For years, the shorts piled on, calling it a bubble, a Ponzi scheme, a tool for criminals. And yet, it survived and thrived, not because of the speculation, but because of the underlying infrastructure and the community that believed in its core value proposition. Could Zhipu or MiniMax have a similar resilience? Could they be building something that the market does not yet understand? It is possible. Perhaps their work in verticalized models for government or enterprise applications will create a sticky revenue stream that the public API price war does not reflect. Perhaps their data flywheels will give them an edge that is not visible in the headline numbers. Perhaps the Chinese government will step in to protect these national champions, providing subsidies or preferential access to compute that will change the cost structure. These are all plausible scenarios. However, in my experience, betting on government intervention or hidden, unquantifiable advantages is a risky proposition. It is the same logic that led many to believe in the stability of Terra's algorithmic stablecoin, a belief that was shattered when the market tested the underlying assumptions. When the market tests Zhipu and MiniMax, I worry that the assumptions will similarly fail. The shorts are not betting on the technology failing; they are betting on the business failing. And in the current environment, that is a bet I would not be quick to dismiss. The more profound issue here is what this says about the broader AI industry. The record short bets are not just a verdict on Zhipu and MiniMax; they are a verdict on the entire "scorched earth" approach to market dominance. We are witnessing a classic race to the bottom, where companies are destroying value in the pursuit of market share. This is a strategy that only works if there is a clear winner who can eventually raise prices and recoup the losses. But in a market with so many well-funded players, including the hyperscalers with their massive cloud profits, the path to victory is unclear. The risk is that the price war persists, margins stay compressed, and innovation is stifled because all capital is diverted to subsidizing usage rather than building new capabilities. This is the tragedy of the commons playing out in real time. When I look at the DeFi summer of 2020, I see the same pattern. Projects were handing out tokens like candy to attract liquidity, and the result was a bubble that burst, leaving a trail of wreckage. The projects that survived were the ones that focused on sustainable value creation, not on subsidized growth. The ones that are surviving now are the ones that built real infrastructure, not just marketing campaigns. The Chinese AI market is now facing the same reckoning. The question is not whether Zhipu and MiniMax can survive the price war; it is whether the industry itself can survive the collateral damage. So, where do we go from here? As someone who has lived through multiple market cycles, I have learned that the loudest signals are often the most deceptive. The record short interest is a loud signal, but it is not the final word. It is a reflection of current market sentiment, not a prophecy of the future. The companies that will ultimately win are the ones that can navigate this period of turbulence and emerge with a sustainable business model. This will require a shift in strategy, away from pure price competition and toward differentiation. It will require a focus on unit economics, on understanding the true cost of serving a customer and the true value that customer brings. It will require a commitment to building moats, whether through proprietary technology, exclusive data, or deep customer relationships. The shorts have provided a valuable service by forcing this conversation into the open. They have forced the market to ask the hard questions about value creation. And they have forced the companies themselves to confront the uncomfortable truth that their current trajectory is not sustainable. When the graph spikes, the soul remains quiet. The spike in short interest is a symptom of a deeper disease, a disease of unsustainable economics and misplaced priorities. The cure is not to silence the shorts; it is to build a better business. That is the challenge that Zhipu, MiniMax, and every other AI company now faces. And it is a challenge that will separate the builders from the pretenders. I have watched this story play out before. I have seen the graphs spike and the souls remain quiet. I have seen the liquidity vanish when the incentives stopped. I have seen the projects collapse when the market tested their assumptions. The lessons are always the same. Value is not created by subsidies; it is created by solving real problems for real customers. Trust is not built by marketing; it is built by delivering on promises. Sustainability is not achieved by growth at all costs; it is achieved by disciplined execution. The Chinese AI market is now learning these lessons the hard way. The record short bets are a warning, but they are also an opportunity. They are an opportunity for the companies to reassess their strategies, to pivot toward sustainable value creation, and to build businesses that can withstand the scrutiny of the market. Whether they will take that opportunity remains to be seen. But the market is watching, and it is not patient. The quiet in the room is deafening, and it is asking for a response. The only question is whether these companies have the wisdom to answer it.