Hong Kong's AI Bet: A Liquidity Story Disguised as Industrial Policy

Let me start with a number that should make any macro observer pause. From December 2023 to May 2024, AI-related new listings in Hong Kong raised nearly HKD 100 billion. That is 55% of the total capital raised on the exchange during that period. The Hong Kong government, through Financial Secretary Paul Chan, is touting this as evidence of a successful AI strategy. I see it differently. This is not a story about technological advancement or industrial policy. This is a liquidity event, dressed up in the language of governance and efficiency. And I want to pull back the curtain on what is really happening here. First, we have to establish the context. Hong Kong is not building AI models. It is not training foundation models. It is not developing chips. The article confirms this without saying it directly. The government's focus is on "implementation and application." The AI Efficiency Task Force has pushed through 30 projects across 13 departments. The goal is to integrate AI into public services. There is a report claiming that if small and medium enterprises (SMEs) adopt AI at the same rate as large corporations by 2035, it could unlock HKD 65 billion in economic benefits. These are nice, clean numbers. They are also completely detached from the messy reality of what AI actually costs to deploy. Liquidity doesn't care about fundamentals. It flows to where the narrative is strongest. Right now, the narrative is AI. Hong Kong, as a global financial hub, is simply riding that wave. The 55% share of IPO proceeds is not a reflection of Hong Kong's AI ecosystem. It is a reflection of global capital's desperate search for yield in a high-interest-rate environment. Investors are not buying Hong Kong AI companies because they believe in the city's tech sector. They are buying them because they have to buy something, and AI is the only game in town. This is a classic late-cycle signal, and it should make anyone who cares about risk management extremely uncomfortable. Let me translate the protocol mechanics here, because that is where the truth lives. The article mentions that Hong Kong's exports have seen high double-digit growth for several quarters, thanks to global demand for AI-related products. On its face, this seems like a healthy sign. But let's look under the hood. What exactly is Hong Kong exporting? It is not software. It is not AI models. It is hardware. It is servers, and the associated components. Hong Kong is a transit point, a node in a global supply chain that originates in mainland China. The growth is not a sign of Hong Kong's own innovation. It is a sign of the mainland's manufacturing capacity and the global appetite for that capacity. Hong Kong is the intermediary, and intermediaries are always the most vulnerable layer in any stack. They have no control over the base layer, and they have no direct access to the end consumer. They are just the pass-through. When the cycle turns, the pass-through gets squeezed first. The government's strategy here is what I call "state-backed market making." They are not just creating a supportive policy. They are creating the demand, the supply, and the marketplace simultaneously. The AI Efficiency Task Force is a government institution designed to prove to the private sector that AI is usable. The IPO data is supposed to prove that capital believes in the sector. The export data is supposed to prove that the world is buying. This is a coordinated narrative. But when a government has to work this hard to create the narrative, it often means the underlying fundamentals are not strong enough to speak for themselves. Real revolutions don't need PR departments. They need supply and demand. This is not supply and demand. This is policy and hope. Now, let's talk about the elephant in the room: the SME problem. The government is betting on SMEs as the main driver of the HKD 65 billion economic benefit. This is a fantasy. The report assumes that if SMEs adopt AI at the same rate as large companies, they will see proportionate benefits. But this ignores the fundamental economics of AI deployment. AI is not a plug-and-play technology. It requires data infrastructure, engineering talent, and continuous model maintenance. Large companies have the balance sheets to invest in these upfront costs. SMEs do not. The average SME in Hong Kong is a trading company, a logistics firm, or a small retail operation. Their margins are thin. They are not going to hire an MLOps engineer. They are not going to build a data warehouse. The idea that they will "catch up" to large enterprises is a structural impossibility. The report's number is theoretical nonsense, and using it as a policy target is a way to mask the real challenge, which is that the bottom of the market has no money for this. This is where my contrarian angle comes in. Everyone is looking at this as a story about the government's AI strategy. I see it as a story about the government's strategy for a leveraged, access the global market. The AI narrative is the vehicle, not the destination. The real asset is the capital markets. Hong Kong needs to attract listings. It needs to attract liquidity. It needs to be the premier capital formation center for the region. AI is just the best sales pitch they have right now. It is a better pitch than real estate, which is a burden. It is a better pitch than old-school tech, which is too niche. AI has global mindshare. It has a built-in narrative of future growth. And the government is smart enough to know that they don't need to build the AI. They just need to provide the casino. The casino is the stock exchange. The chips are the AI stocks. And the house always wins. This is not a critique. It is a description. As a macro analyst, I need to describe the rules of the game. The problem is that the game has a structural flaw. The flaw is the assumption that liquidity will continue to flow. The data from the IPO market is a snapshot, not a trend. The demand for AI stocks is being driven by a global liquidity cycle that is currently in a bullish phase. But that cycle will turn. It always turns. When the market decides that the AI story is overpriced, the first place to feel the pain will be the markets with the highest concentration of AI listings. That is Hong Kong. The 55% concentration is a risk, not a strength. It is the same risk that a portfolio holds. When the market is high, you feel great. When the market crashes, you have nothing to fall back on. Hong Kong has put all its eggs in the AI basket, and the basket is being built on a foundation of accounting and potential, not a foundation of cash flow and earnings. This is the classic dot-com setup. The infrastructure is not there. The earnings are not there. The narrative is everywhere. Let me add a layer to this analysis. The real constraint for Hong Kong is not capital. It is compute. The city has physical limits. It has a limited land supply. It has high energy costs. It does not have the physical infrastructure to host massive AI data centers. And the article is completely silent on this. The government is not building a compute infrastructure. It is not planning a supercomputer. It is relying on the private sector to handle that. But the private sector will go where the energy is cheap and the land is abundant. That is not Hong Kong. It is a mainland, specifically in the Greater Bay Area. So the Hong Kong AI story is actually a dependency on mainland infrastructure. The value is created in Shenzhen. The revenue is booked in Hong Kong. The processing happens in Dongguan. And Hong Kong just gets the fee for the financial transaction. This is not a sustainable model for building a tech ecosystem. It is a model for building a financial center, but it is not a model for building an AI center. And the government's messaging, conflating the two, is a problem. I have to address the risk of the other person. The official rhetoric focuses on the benefits to the public sector. Thirty projects across 13 departments sounds impressive. But what do these projects actually do? Are they replacing core functions, or are they just automating the mundane? If the projects are just chatbots and workflow automation, they are not changing the game. They are just using AI as a better version of a rule-based system. The real value of AI is in the complex, non-linear, data-rich decisions. Those are the hardest to implement and the hardest to trust. And the government is not talking about that. It is talking about "efficiency improvements" which is a bureaucratic phrase that usually means "we are doing the same thing, but with a new tool." That is not transformation. That is a budget line. The government is doing what governments always do with new technology: they buy it, they don't adopt it. They create a task force to make it look like they are serious. But the core operations remain the same. The AI becomes a layer, not a fundamental change. Another rug? No, just a liquidity trap. This is what I am afraid of. The AI stocks are the new junk. They have a high yield narrative, but a high risk of principal loss. The government is not creating value. It is creating a market. And the market is a price discovery. The market is a vehicle for capital. It is not a vehicle for innovation. The innovation is happening elsewhere, and Hong Kong is just the place where the innovation comes to get a funding round. That is not a bad place to be. But it is a vulnerable place to be. The city is a hostage to the global tech cycle. It is a hostage to the global rate cycle. And it is a hostage to the sentiment of the international investor. If any of these hostages are, the entire narrative collapses. The city is not building a safe harbor. It is building a lighthouse on the edge of a cliff. It is a great view. But it is a precarious foundation. So what is the actual takeaway? What should a macro watcher do with this information? The macro does not lie, but the market is. The global liquidity is the tide. The AI narrative is the boat. And Hong Kong is the port. When the tide goes out, all the boats go down. The only question is how much damage they will do to the port. The 55% concentration is a red flag. The 100 billion is a red flag. The export growth is a red flag. These are all signs of a market that is expanding on leverage, not on fundamentals. It is the same pattern we saw in 2021. It is the same pattern we saw in 2018. It is the same pattern we saw in 2017. The narrative changes. The liquidity remains. The cycle does not. The question is not whether AI is real. It is. The question is whether the companies in Hong Kong are the ones that will survive the wave. Most of them will not. The AI market is a winner-take-all market. The big models will dominate. The startups will be squeezed. The ones that are listed in Hong Kong are not the base models. They are the application companies. And the application companies are the ones that will be squeezed. They do not have the scale. They do not have the data. They do not have the compute. They have a story. And the story is being sold to the public. This is the classic IPO trap. The founders get the money. The investors get the story. And the retail gets the bag. It is a beautiful machine. And it is running at full speed. I will not take the bait. I will not say that the AI is a bubble. It is not. The underlying technology is real. The question is the valuation. The question is the structure. The question is the sustainability. The Hong Kong market is a place where the value of the AI is being created. But the value is being created in a concentrated, leveraged way. It is a story of the financialization of a technology. And the financialization of a technology is not the same as the adoption of a technology. It is a game of capital. And the capital is the game. The tech is just the token. In the end, the Hong Kong story is a story about a liquidity event. The AI is the excuse. The capital is the real story. And the capital is a global story. It is a story of the Federal Reserve, the dollar, and the search for yield. It is a story of the tech cycle. It is a story of the China cycle. The Hong Kong government is not the main character. It is a supporting actor. The main character is the global liquidity pool. And that pool is a dangerous place. I would advise the reader to look at the Hong Kong AI market with a high degree of skepticism. The potential is there. But the structure is weak. And the structure is what matters.