Watching the silence between the candlesticks.
On July 22, 2024, two names in Hong Kong's AI sector—MINIMAX and Zhipu—dropped with a quiet brutality: MINIMAX shed over 9%, Zhipu over 3%. The news feed offered nothing else. No product recall, no team departure, no regulatory hammer. Just a collective exhale from a market that had been holding its breath. To the macro eye, this is not noise. It is the sound of a structure settling.
I have seen this pattern before. In 2017, when I audited 40+ ICO whitepapers for a Sydney fund, the same silence preceded the collapse of tokens that were propped up by narrative alone. The ICOs with solid tokenomics survived what followed. The rest became ghosts. Now, the narrative driving AI stocks—the promise of AGI, of infinite productivity—has begun to crack under the weight of reality: high interest rates, zero revenue visibility, and a liquidity cycle that rewards only the most defensible assets.
The context here is not merely Hong Kong. It is global. The Federal Reserve's balance sheet runoff continues. Real yields remain elevated. Capital that was freely flowing into speculative tech in 2023 is now being redeployed into safer harbors. This is the same macro current that dragged down crypto in 2022, when the Luna collapse marked the end of free money. We are experiencing a second-order effect: the unwinding of the AI hype cycle. Flow follows the path of least resistance—and right now, resistance is high.
But what is actually breaking?
Let me be precise. MINIMAX and Zhipu are not frauds. They are engineering-heavy teams with legitimate technology. MINIMAX's linear attention architecture is innovative; Zhipu's GLM-4 holds its own in Chinese language processing. But innovation does not equal investability. In a bull market for attention, their stocks were bid up on the expectation of exponential growth—a growth that requires massive capital expenditure for GPU clusters, sky-high researcher salaries, and years before meaningful revenue materializes.
This is a structural mismatch. The companies are burning cash at rates that assume perpetual generosity from capital markets. When the mood shifts—as it did on July 22—the valuation compression is brutal. I call this the "liquidity mirage." During my 2020 DeFi mining days, I watched protocols with $100M+ TVL evaporate in hours when arbitrageurs withdrew. The same mechanics apply here: the exit is faster than the thesis.
Harvesting the liquidity that others overlook means recognizing that this selloff is not a random tremor. It is a signal that the market is differentiating between narrative and substance. In my work analyzing cross-chain bridges, I learned that $2.5 billion in bridge hacks did not kill the interoperability narrative—but it did force capital to flow toward more secure architectures. Similarly, this AI stock correction will force capital toward companies with real customer traction, not just demos.
Contrarian angle: The decoupling that isn't.
Conventional wisdom says AI and crypto are different asset classes. They share one common denominator: sensitivity to global liquidity. When liquidity contracts, both get sold. But the contrarian truth is that this selloff creates a rare opportunity to decouple from the herd. Just as I retreated to a cabin in the Blue Mountains after the 2022 Luna crash to rebuild my mental framework, the market now has a chance to separate signal from noise.
What is the signal? The companies that will survive this cycle are those with measurable unit economics: API pricing that recovers costs, enterprise contracts with recurring revenue, and a clear path to reduce reliance on venture capital. Both MINIMAX and Zhipu have announced enterprise partnerships, but the market wants proof. Solitude reveals the truth the crowd ignores: the drop may be driven by panic, but the underlying technology progress continues.
In March 2024, I advised a mid-tier Australian fund on hedging strategies ahead of the US Spot Bitcoin ETF approval. The lesson there was that regulatory clarity—not hype—drives sustainable institutional inflows. The same holds for AI: clear regulations on content safety and data sovereignty will eventually create a moat for compliant players. Zhipu's ties to Tsinghua and Chinese academic ecosystem give it an edge in navigating local requirements. MINIMAX's relationship with Alibaba provides cloud infrastructure at scale. These are assets that the current selloff is pricing as liabilities.
But I must be honest: the risk of further drawdown is real. The short-term technicals look weak. Volume patterns suggest institutional distribution, not accumulation. If the Fed signals another rate hike, these stocks could fall another 20-30%. That is the nature of being early in a technology transition. The question is not whether the technology will matter—it is whether the market's patience aligns with the development timeline.
Patience is the leverage that never depreciates.
The takeaway is not a buy or sell call. It is a warning to respect the structure. The silence between these candlesticks is the market telling us that the easy money phase of AI is over. What remains is the hard work of building real economic value. As a macro watcher, I am watching the liquidity flows into and out of this sector. When the next catalyst arrives—whether a Fed pivot, a breakthrough in inference efficiency, or a major enterprise deal—the capital will return. But only for those who have survived the silence.
Diving for pearls in the deep web of value means staying focused on the fundamentals while others chase the next narrative. The pattern emerges from the chaos of noise. I have seen it in crypto. I am seeing it now in AI. The question is: are you watching the candlesticks, or the silence between them?