The 55% Signal: Hong Kong's AI Ambition and the Architecture of Dependency

The protocol does not lie; the interface does. In Hong Kong, the interface is a policy blog post from the Financial Secretary. The protocol is the underlying economic and technological architecture. Reading Paul Chan's recent statement on AI, I see a familiar pattern: a government discovering the narrative power of a technology without yet understanding its structural demands. The numbers are impressive. AI-related IPOs raised nearly 100 billion HKD, representing 55% of all new listings. Thirty efficiency projects across thirteen government departments. High double-digit export growth. But numbers, like interfaces, can obscure the underlying mechanics. To own the chain is to own the history. To own an AI strategy is to own the compute, the talent, and the data. Hong Kong, I suspect, owns none of these. Not yet. The context here matters. Hong Kong is not Shenzhen. It is not Beijing. It has no homegrown foundation model lab. Its AI strategy, as articulated, is one of application and adoption, not creation. This is a rational choice, given the resource constraints. Building a large language model from scratch requires billions in compute, a deep pool of research talent, and years of patience. Hong Kong has capital markets, a legal system built on common law, and a position as a gateway between China and the West. The strategy, therefore, is to be the layer where AI meets finance, trade, and public services. The strategy is to be the interface. The danger is that the interface, without the protocol, is just a facade. Let me be precise about what the 55% figure actually represents. In my years auditing smart contracts and tokenomics, I learned to distinguish between genuine value creation and narrative arbitrage. The 55% concentration of AI-related IPOs is a market signal, but it is not a technology signal. It tells us that investors believe AI is the future. It does not tell us that these companies have defensible moats, proprietary technology, or even meaningful AI revenue. I have seen this movie before. In 2017, every ICO was a "protocol." In 2021, every NFT project was a "community." In 2025, every company with a chatbot is an "AI company." The label is cheap. The underlying technology is not. The core of my analysis, however, is not about the quality of these IPOs. It is about the structural dependency that Hong Kong is building. The government's AI efficiency projects, the financial sector's AI adoption, the push for SME integration—all of these require compute. And compute requires infrastructure. Hong Kong has land constraints, high energy costs, and a climate that is hostile to data centers. The article does not mention a single plan for a local AI compute cluster. This is not an oversight. It is a strategic blind spot. The city is building an AI economy on rented infrastructure. The cloud providers—Alibaba, Tencent, AWS—will be the landlords. And landlords, as any tenant knows, set the terms. I have spent the last decade working on consensus mechanisms and zero-knowledge proofs. I have learned that trust is not a feeling; it is an architecture. Hong Kong's AI strategy lacks a trust architecture. Consider the data question. The government plans to use AI across thirteen departments. This means citizen data—identity records, tax filings, public service usage—will flow through AI systems. Where will this data be processed? In Hong Kong, on local infrastructure? Or in mainland China, on cloud servers? Or in the United States, on hyperscaler infrastructure? The article is silent. This silence is not neutral. It is a signal of unresolved tension between the "one country, two systems" framework and the practical realities of AI deployment. The data governance framework is not a technical detail. It is the foundation of public trust. Without it, the AI strategy is building on sand. The contrarian angle here is not that Hong Kong will fail. The contrarian angle is that Hong Kong's success, as currently defined, will create a new form of dependency. The city is positioning itself as the "AI hub" for the region. But a hub is a node, not a source. The value flows through, but it does not originate. The 650 billion HKD in potential economic benefits from SME adoption is a compelling number. But it is a potential, not a certainty. It assumes that SMEs have the digital infrastructure, the talent, and the capital to integrate AI. It assumes that the technology is mature enough to deliver value in a Cantonese-speaking, highly regulated, cross-border business environment. These are not trivial assumptions. In my experience, the gap between a technology's potential and its realized value is where most strategies die. Let me also address the talent question, because it is the most under-discussed constraint. Hong Kong has a world-class financial sector. It does not have a world-class AI research community. The article mentions no specific plan for AI talent attraction. No visa program, no tax incentives, no university partnerships. This is a critical omission. AI is not a technology you buy; it is a capability you build. And capabilities require people. The city can import models, but it cannot import the ability to innovate on those models. The result will be a perpetual follower status—always adapting, never leading. This is a viable strategy, but it is not the strategy of a "hub." It is the strategy of a service provider. The infrastructure question is equally pressing. I have audited enough systems to know that AI is not a software problem. It is a hardware problem. The models that power the AI revolution require GPU clusters, high-speed interconnects, and massive energy supplies. Hong Kong has none of these in sufficient quantity. The article's silence on this point is telling. It suggests that the government either does not understand the infrastructure demands of AI, or it has a plan that it is not ready to disclose. Both possibilities are concerning. The first suggests a lack of technical depth. The second suggests a lack of transparency. Neither is a good foundation for a national AI strategy. There is also the question of the "pseudo-AI" problem. The 55% IPO concentration is a market signal, but it is also a warning. In a bull market, the incentive is to label everything as AI. The financial sector, which Hong Kong knows well, is particularly prone to this. A fintech company that uses a simple rule-based system is not an AI company. A logistics firm that uses a spreadsheet is not an AI company. But in a market hungry for AI exposure, these labels stick. The result is a misallocation of capital. The 650 billion HKD in SME benefits will not be realized if the capital is flowing to companies that are not actually building AI capabilities. The market is rewarding the narrative, not the technology. This is a classic bubble dynamic. I have been through enough market cycles to recognize the pattern. The 2017 ICO boom was driven by the narrative of decentralization. The 2021 NFT boom was driven by the narrative of digital ownership. The 2025 AI boom is driven by the narrative of intelligence. In each case, the narrative was not false. It was incomplete. The technology was real, but the applications were immature. The result was a period of overinvestment, followed by a correction, followed by a period of genuine value creation. Hong Kong's AI strategy is likely to follow the same arc. The question is whether the city can survive the correction and emerge with real capabilities. The answer depends on whether it is building infrastructure, talent, and data governance, or just narrative. The geopolitical dimension adds another layer of complexity. Hong Kong sits between the US and China, between two AI superpowers with competing standards and export controls. The city's AI strategy must navigate this divide. It must comply with US export controls on advanced chips. It must comply with Chinese data regulations. It must maintain its own legal framework. This is a narrow path. The article does not address it. This is a significant omission. The AI supply chain is not neutral. It is deeply political. Hong Kong's position as a "super-connector" is both an asset and a liability. It can facilitate cross-border AI trade, but it can also become a point of friction. The city needs a clear strategy for managing this tension. The article provides no such strategy. Let me return to the data. The 30 efficiency projects across 13 departments are a positive signal. They show that the government is serious about AI adoption. But they also raise questions. What are the specific use cases? What are the success metrics? How will the government measure the impact of these projects? The article is vague on these points. This vagueness is a risk. Without clear metrics, the projects will be judged by narrative, not by outcomes. And narrative, as I have learned, is a poor substitute for evidence. The government needs to publish the details of these projects. It needs to subject them to independent audit. It needs to be transparent about the data being used and the algorithms being deployed. This is not just a matter of good governance. It is a matter of public trust. And public trust, like cryptographic security, is hard to build and easy to destroy. The SME opportunity is real, but it is not automatic. The 650 billion HKD figure is an estimate, not a forecast. It assumes that SMEs will have the capacity to adopt AI. This is not a given. Many SMEs lack basic digital infrastructure. They lack the talent to implement AI solutions. They lack the capital to invest in new technology. The government's role should be to bridge these gaps. It should provide subsidies, training, and access to affordable AI tools. It should create a marketplace where SMEs can find vetted AI solutions. It should measure the adoption rate and the impact. The article does not mention any of these measures. It presents the opportunity but not the plan. This is a common pattern in policy documents. The vision is clear. The execution is vague. And execution, as any engineer knows, is where the value is created or destroyed. I am also concerned about the lack of attention to AI safety and ethics. The article does not mention bias, fairness, transparency, or accountability. These are not academic concerns. They are practical issues. A government AI system that makes biased decisions can cause real harm. A financial AI system that is not transparent can undermine market confidence. The city needs a clear framework for AI governance. It needs to define the principles that will guide AI deployment. It needs to establish mechanisms for oversight and redress. The article is silent on these issues. This silence is a risk. It suggests that the government is prioritizing speed over safety. In the long run, this is a false economy. A single high-profile AI failure could set back the entire strategy. The comparison with Singapore is inevitable. Singapore has a national AI strategy, a dedicated AI research institute, and a clear plan for talent development. It is building its own compute infrastructure. It is investing in foundational research. Hong Kong, by contrast, is focusing on application and adoption. This is a legitimate strategic choice, but it is a choice with consequences. Hong Kong will be a consumer of AI technology, not a producer. It will be a market for AI solutions, not a source of innovation. This is not necessarily a bad position. It can be profitable. But it is not a position of leadership. And in a rapidly evolving field, followers are always at risk of being disrupted by the next wave of innovation. Let me be clear about what I am not saying. I am not saying that Hong Kong's AI strategy is wrong. I am saying that it is incomplete. The city has identified the opportunity. It has not yet addressed the structural requirements. The compute, the talent, the data governance, the safety framework—these are not optional extras. They are the foundation. Without them, the AI strategy is a house of cards. The 55% IPO concentration is a signal of market enthusiasm. It is not a signal of technological readiness. The 650 billion HKD in SME benefits is a potential. It is not a plan. The 30 government projects are a start. They are not a strategy. Certainty is a bug in a stochastic world. I have learned this lesson repeatedly in my career. The market is certain that AI is the future. The government is certain that AI will transform the economy. The investors are certain that AI companies will generate returns. But certainty is not a substitute for analysis. The future is not a linear projection of the present. It is a complex system with feedback loops, tipping points, and emergent properties. The AI revolution will not unfold according to anyone's plan. It will be shaped by the interaction of technology, policy, and human behavior. Hong Kong's role in this revolution is not predetermined. It will be determined by the choices the city makes today. The question is whether those choices are based on a deep understanding of the technology or on a superficial embrace of the narrative. I have spent my career in the space between code and society. I have seen technologies that promised to change the world and delivered incremental improvements. I have seen technologies that were dismissed as toys and became the foundation of new industries. The difference is not the technology itself. It is the ecosystem that surrounds it. The talent, the infrastructure, the governance, the culture of innovation. Hong Kong has the capital. It has the legal system. It has the geographic position. What it lacks is the technical depth. The city is trying to build an AI economy without a strong AI research community. This is like trying to build a financial center without bankers. It is possible, but it is not sustainable. The takeaway from this analysis is not a prediction of failure. It is a call for depth. Hong Kong needs to move beyond the narrative and address the fundamentals. It needs to invest in compute infrastructure, either locally or through a clear partnership with the mainland. It needs to develop a talent pipeline, through education, immigration, and collaboration with universities. It needs to establish a data governance framework that protects citizens while enabling innovation. It needs to create an AI safety and ethics framework that builds public trust. These are not easy tasks. They require political will, technical expertise, and long-term commitment. But they are the tasks that will determine whether Hong Kong's AI strategy is a genuine transformation or just another bubble. The protocol does not lie; the interface does. Hong Kong's AI strategy is currently an interface. It is a set of policy statements, market signals, and government initiatives. The underlying protocol—the compute, the talent, the data, the governance—is still being built. The city has a choice. It can continue to build the interface, polishing the narrative and attracting capital. Or it can go deeper, building the protocol that will sustain the narrative. The first path is easier. The second path is more durable. I know which path I would choose. But I am not the one making the decision. The Financial Secretary is. And the clock is ticking. The AI revolution will not wait for Hong Kong to catch up. It is happening now, in real time, in the code and the compute and the data. The question is whether Hong Kong will be a participant or a spectator. The answer will be written in the infrastructure the city builds, the talent it attracts, and the governance it establishes. The answer will be written in the protocol, not the interface.