The numbers coming out of Hong Kong are breathtaking, if you read them at face value. AI-related new listings have raised nearly HK$100 billion since December, representing 55% of all IPO proceeds on the exchange. The Financial Secretary, Paul Chan, tells a story of an economy riding the wave of an AI boom, a government pushing 30 efficiency projects across 13 departments, and exports growing at double-digit rates. It sounds like a city that has found its future. But I have spent years in this industry, watching narratives form, bubble, and burst. And I can tell you: the most important signal in this narrative is not the $100 billion raised; it is the infrastructure that is not being built.
We are looking at a city-state that is making a massive, coordinated bet on the application layer of AI, without any visible strategy for the substrate. It is a bet on assembling the car without owning the engine, the road, or even the fuel. The chaos of this chain of thought is where the signal lies.
Let's start with what is obvious. The Financial Secretary's message is a clear statement of intent: Hong Kong wants to be the world's most AI-friendly place to do business. The 55% figure for AI-related IPO fundraising is a staggering number, likely surpassing any other major exchange in the world. It is a testament to the power of the city's capital markets and the hunger of investors for anything with an AI label. The 30 efficiency projects across 13 government departments are a classic top-down move to force adoption and set a precedent for the private sector.
This is a strategy of "application-led, efficiency-first" adoption. It is a choice to be a fast follower, not a pioneer. It is a rational choice, given Hong Kong's profile. It does not have the deep well of foundational AI research or the giant model labs of Beijing, Shenzhen, or Hangzhou. It is not a city-state that can spend billions on training a GPT-4 competitor. So, it is saying, "We will not build the engines; we will be the best pilots in the world." This is a classic, pragmatic, and even wise move for a small, wealthy, and highly specialized economy.
The problem is not the strategy's logic; it is the strategy's execution. The pilot's logic is sound, but the plane is missing its wings.
The core of my concern is the complete silence on computational infrastructure. The report I am looking at is a deep-dive analysis of Hong Kong's AI push, and it does not mention a single plan for a government-backed computing cluster, a supercomputer, or a dedicated AI data center. This is a glaring omission. In my experience, in the world of blockchain, we talk about the "trilemma" of security, scalability, and decentralization. In the world of AI, there is a similar trilemma: You can have cheap, you can have fast, or you can have sovereign. You can only have two of them.
Hong Kong is choosing cheap and fast, by renting compute from cloud providers, and potentially from the mainland. But it is giving up sovereign compute. This is a profound choice with long-term consequences. If your AI application is the core of your economic strategy, and you do not control the underlying compute, you are not just renting a service; you are renting your destiny. You are a tenant in the digital ecosystem, not a landlord.
Let's look at the economic figures more closely. The analysis suggests that if small and medium enterprises (SMEs) adopt AI at the same rate as large companies by 2035, it could release a HKD 65 billion economic benefit. That sounds like a massive prize. But as someone who has audited a few financial models in my time, let's look at the assumptions. This figure is a potential, not a certainty. It depends on SMEs having the digital foundation, the talent, and the financial firepower to integrate AI. In Hong Kong, the SME sector is the backbone of the economy, but it is also dominated by traditional retail, trade, and real estate services. The path to AI adoption is not a simple one; it is a complex journey of cultural change, workflow redesign, and skills uplift. The 65 billion is a mountain of value, but it is locked behind a door that requires a key, and the key is not just software.
Culture is the new consensus mechanism. This is a truth I have learned in the crypto world, and it applies here. You cannot just mandate AI adoption. You need a culture that is curious, data-driven, and tolerant of failure. Hong Kong has a culture of speed, efficiency, and commerce, which is a great base. But it also has a culture of hierarchy and risk aversion, which can be a barrier to experimentation. The government's 30 projects are a good start, but they are also a drop in the bucket. They are a top-down signal. The real test is whether the bottom-up organic adoption takes root.
The report also highlights the export boom, with double-digit growth in high-tech exports. This is a positive signal, but it is worth asking what is being exported. In Hong Kong, this is likely not about high-value AI software. It is more likely about the physical components of the AI boom: the GPUs, the storage chips, and the server components that are being re-exported from China to the world. This is the story of the shipping hub and the trading hub, not the innovator. The value is in the logistics and the trade finance, not in the intellectual property. This is a low-margin, high-volume game, and it is subject to the whims of global supply chains and trade policies. It is a good boost, but it is not the foundation for a long-term competitive advantage.
The report also highlights the growth of the financial market, with the Hang Seng Index incorporating AI companies. This is a positive step, but it is also a potential trap. When an index starts to chase a narrative, it can create a self-fulfilling prophecy that attracts passive capital, which does not do the fundamental analysis. It just flows in because the AI is in the index. This creates a momentum-driven market that can easily turn into a bubble. I have seen this in the crypto world with tokens that get added to major exchanges. The price pump comes, but the utility is often left behind. The question is not whether the AI companies are in the index, but whether their earnings and cash flows justify their valuations. Truth is not mined; it is remembered. And the market will remember the truth of a company's fundamentals in the long run.
The report's analysis of the competitive landscape is clear. Hong Kong is positioning itself as a "hub player," a bridge between the AI supply from China and the capital demand from the world. This is a classic hub strategy, and it can work. It is a strategy of "借力打力" (borrowing force to defeat force). It is using the mainland's technology and the world's capital to create a middle layer. But the sustainability of this strategy depends on the stability of the mainland's tech progress and the continued attractiveness of Hong Kong's capital markets. Both are fragile. Singapore is also a major competitor, and it is not just resting on its laurels. It is building its own national AI strategy, investing in talent, and building out its compute infrastructure. The race is not won by the starting position; it is won by the speed and the endurance of the runner.
The report also has a very high level of confidence in the ability of the government to act as a catalyst. The 30 projects are a signal, but they are a signal of intent, not a signal of impact. The impact will be measured by the quality of these projects. Are they truly innovative, or are they just automating the paperwork? Are they changing the way the government interacts with citizens, or are they just making it faster to fill out a form? In my experience, true innovation is not about efficiency; it is about a change in the process. It is about asking a new question, not just answering an old one faster.
I want to bring this back to a core principle. In the chaos of the chain, find the signal. The signal here is not the $100 billion. The signal is the lack of a plan for a data center. The signal is the lack of a plan for a talent pipeline. The signal is the lack of a clear regulatory framework for data privacy and security. These are the foundational elements that are missing.
Let's talk about the talent. Hong Kong's universities are good, but they are not producing enough AI researchers and engineers. The report mentions the need for talent, but it does not offer a plan. It does not have a clear, aggressive talent visa policy, like Singapore's. It does not have the housing and tax incentives to attract top-tier talent from the Bay Area. Without a deep pool of local talent, the application layer will always be shallow. It will be a layer of consultants and integrators, not a layer of innovators.
And let's talk about the data. Hong Kong is a special administrative region. It has a different legal system, a different data privacy regime, but it is also a part of the Chinese economy. Its AI applications will have to handle a cross-border data flow, which is a complex regulatory issue. The government's AI projects will be processing sensitive citizen data. Where will that data be stored? In Hong Kong, or on the mainland? What is the oversight? What is the transparency? The report is a completely silent on this. In the crypto world, we talk about code is law. In the AI world, we need to talk about data is the foundation. Ideas have no gas fees, only gravity. The gravity of data is the regulatory and ethical weight of the data.
Now, let me pivot to the contrarian angle. Is the lack of infrastructure actually a problem? Or is it a calculated risk? In the world of AI, it is possible to be an application layer without owning the base. Look at the rise of the SaaS companies in the past decades. They did not own the cloud; they built on top of it. Salesforce did not build a data center, but it built a trillion-dollar company. So, maybe Hong Kong does not need to build a GPU cluster. Maybe it can just be a super user.
The problem with this analogy is that the SaaS companies built their own intellectual property on top of the cloud. They wrote the code that solved the specific problems. They had the engineers who understood the business problem and the technology. The Hong Kong government is not building proprietary algorithms; it is using off-the-shelf models. This is a big difference. It is like a company that uses an ERP system without customizing it. It will get the standard efficiency, but it will not get the competitive advantage. The AI application layer requires a lot of custom work, data cleaning, and integration. If you do not have the talent to do that, you will be stuck with the generic solution.
There is also the issue of the data moat. The real value in AI is not the algorithm; it is the data. The government has a massive amount of data. It has data on the population, the economy, the health. If it can use this data to train a model that is specific to the city, that is a moat. But this requires a robust data governance framework, which is not present. And the private sector, the data is fragmented. The SMEs do not have the data or the talent. The large banks have the data, but they are often using global models, not local ones.
I also see a potential for a, a real blind spot. The report is all about the upside: the capital, the exports, the efficiency. There is no mention of the downside: the job displacement. The government's 30 efficiency projects, which are likely to replace some administrative roles. What is the plan for those workers? Is there a social safety net? Is there a re-training program? The report is silent. In a city-state with a high cost of living, the unemployment is a political issue. The narrative of the AI-driven growth could be a source of social unrest if the benefits are not widely shared.
My takeaway is a sense of a urgency and a sense of a hope. The Hong Kong government is making a move. The decision is to be a fast follower, not a slow leader. It is a bet. But the bet is incomplete. It is a bet on the application without a bet on the infrastructure. It is a bet on the demand without a bet on the supply. It is a bet on the future without a plan for the present.
We do not build walls; we build bridges for value. But a bridge needs to be anchored on both sides. The first anchor is the capital, and it is strong. The second anchor is the foundation, and it is weak. I would like to see a plan for the foundation. I would like to see a plan for a compute cluster. I would like to see a plan for a data governance framework. I would like to see a plan for a talent pipeline. I would like to see a plan for the societal impact. Without these, the bridge will be built, but it will not be safe for crossing.
I am a believer in the power of decentralized systems. I believe in the power of a shared truth. But this truth must be built. It must be built with the right tools. And the most important tool is not the algorithm, the code, but the ecosystem. The ecosystem is the soil. If the soil is not fertile, the seeds of AI will not grow.
The future is written in code, but it is felt in spirit. The spirit of Hong Kong is the spirit of the entrepreneur, the hustler, the bridge. This spirit is perfect for the application layer. But the spirit needs to be supported by the physical. The physical is the compute, the data center, and the talent. The spirit is the will. The physical is the way. The way is not just the code. The way is the process of building.
So, I will watch the space. I will watch the 30 projects. I will watch the next earnings season. I will watch the data on the SME adoption. And I will watch the political landscape. But I will also watch the corner for the announcement of a new data center. I will watch the government's budget. And I will watch the flow of talent. The financial data is important, but the infrastructure is the signal. I will look at the infrastructure. And I will ask the question: What is the plan for the foundation? The answer to this question will determine if the $100 billion is a foundation for the future or a monument to a missed opportunity. The future is written in code, but felt in spirit.