On July 22, 2024, Hong Kong’s AI concept stocks took a hit. MINIMAX dropped over 9%, Zhipu slid 3%, and the broader sector bled. Headlines called it a “market correction.” But as someone who spent 2017 arbitraging EOS token sales and watching DeFi Summer burn itself out, I saw something else: the invisible current beneath the surface. The yield is a lie—and so is the narrative that AI stocks are simply overvalued. They’re not. They’re mispriced against a macro reality that’s already shifting beneath our feet.
Context: The Global Liquidity Map
Let’s set the stage. In mid-2024, the Federal Reserve’s interest rate policy remains a sledgehammer. Despite market whispers of a pivot, the U.S. dollar index (DXY) stays elevated, draining liquidity from risk assets globally. Hong Kong, a proxy for China’s tech exposure, feels the pinch first. But here’s the twist: the sell-off in AI stocks isn’t about AI. It’s about the same macro forces that crushed altcoins in 2022. Tracing the invisible currents beneath the market, you see a pattern: when central banks tighten, assets without earnings—whether tokens or tech stocks—get repriced first. MINIMAX and Zhipu are not profitable. Their valuations rest on future revenue, which requires cheap capital. That capital is gone.
Core: Crypto as Macro Asset Analysis
Now, let’s apply the lens I use for digital assets. In 2020, I published a white paper arguing DeFi was a liquidity transfer mechanism, not value creation. The same logic applies here. MINIMAX, with its cutting-edge linear attention architecture, and Zhipu, backed by Tsinghua’s academic clout, have real technology. But technology does not equal a business. Their stock prices are driven by narratives—just like BRC-20 tokens on Bitcoin. BRC-20 on Bitcoin is like using a Rolls-Royce to haul cargo—it insults the car and doesn’t carry much. Similarly, AI models without a revenue engine become a burden, not an asset.
I’ve seen this movie before. In 2021, I tracked NFT wash trades and realized 60% of volume was fake. Today, the AI narrative is getting washed through a similar filter. The market is demanding proof of adoption: API calls, enterprise contracts, unit economics. MINIMAX and Zhipu have none of these public. Their recent valuation rounds—$1.5 billion for MINIMAX, $2 billion for Zhipu—were set when liquidity was abundant. Now, with the Fed’s balance sheet shrinking, those valuations are being stress-tested. The bubble is audible, and it sounds like a stock dropping 9% in a day.
Contrarian: The Decoupling Thesis is a Lie
The prevailing narrative says AI and crypto are decoupled—one is “real tech,” the other is “speculation.” My 23 years of watching markets say otherwise. Both are assets that trade on future cash flows discounted by a risk-free rate. When that rate rises, both get crushed. The 2022 Terra collapse taught me that algorithmic stablecoins fail not because of bad code, but because of broken incentives. AI stocks fail not because of bad models, but because of broken liquidity assumptions.
Here’s the contrarian angle: the market is not wrong to sell. It’s wrong to think the correction is over. We are in the early innings of a structural repricing. The institutions piling into Bitcoin ETFs in 2024 are the same ones that will demand AI companies show EBITDA. They are not buyers of dreams—they are buyers of cash flows. And when they don’t see them, they leave. Watch the hands, not the charts. The hands are selling AI stocks because they see the same macro headwinds that made DeFi tokens lose 90%.
Takeaway: Cycle Positioning
So what does this mean for a crypto portfolio manager? It means we are in a global liquidity contraction that spares no asset. AI stocks are a canary in the coal mine. If you’re long on crypto, you need to hedge with macro-aware positions—short duration, high cash flow, maybe even a short against AI ETFs. The euphoria of 2023 is fading. The market is asking for receipts.
I wrote this not as a bear, but as a skeptic who’s been burned by the 2017 ICO arbitrage paradox, where I captured $150,000 risk-free and lost it to a hack because I over-optimized code instead of securing keys. Chaos is the only constant. The AI sell-off today is not about AI. It’s about the macro realizing that the party is over. Position accordingly.