Hong Kong's AI Ambition: A Centralized Blueprint in a Decentralized World

In the chaos of consensus, I seek the quiet truth. The Hong Kong government's recent push to integrate AI across its bureaucracy is not a story about technology. It is a story about trust, architecture, and the unspoken assumptions we make about who gets to hold the keys to our digital future. When I read the Financial Secretary's policy statement, I did not see a roadmap for innovation. I saw a covenant being written in a language that excludes the very people it claims to serve. The numbers are impressive, at first glance. AI-related IPOs have raised nearly HKD 100 billion, accounting for 55% of all new listings. Exports are growing at double-digit rates. Thirty efficiency projects are being rolled out across thirteen government departments. But as someone who spent four months in 2017 manually auditing the governance structures of early DAOs—and found that two-thirds failed to define clear decision-making rights for community members—I have learned to look past the headline numbers and ask a different question: who controls the underlying infrastructure? This is not a question of whether AI will transform Hong Kong's economy. It will. The question is whether that transformation will be built on foundations that respect individual sovereignty, or whether it will simply replicate the same centralized power structures that blockchain technology was designed to challenge. The Financial Secretary's vision is one of efficiency, speed, and economic growth. My concern is that in the rush to capture the AI dividend, Hong Kong is building a system that concentrates power rather than distributing it. Let me be clear about what I am not saying. I am not arguing that Hong Kong should abandon its AI ambitions. The city's role as a financial hub and its unique position as a bridge between China and global markets give it genuine advantages. But the path it is choosing—a top-down, government-led, application-first approach—carries structural risks that the policy statement does not acknowledge. The 55% IPO concentration, the reliance on external model providers, the absence of any mention of compute infrastructure, and the silence on data governance all point to a strategy that prioritizes short-term economic gains over long-term resilience. In this analysis, I will examine Hong Kong's AI strategy through the lens of decentralization. I will argue that the city's application-first approach, while pragmatic, creates dependencies that could undermine its competitiveness. I will explore the hidden costs of centralized AI deployment, the risks of relying on external compute and model providers, and the missed opportunity to build a more sovereign, transparent, and equitable AI ecosystem. And I will suggest that the principles of decentralization—transparency, user agency, and distributed control—offer a more sustainable path forward. Code is the new covenant, but trust is the ink. Hong Kong's AI strategy is being written with a pen that belongs to someone else. The Context: Hong Kong's AI Strategy in Brief Hong Kong's AI strategy, as articulated by Financial Secretary Paul Chan, is built on three pillars: government-led adoption, capital market enablement, and economic integration. The government has established an AI Efficiency Task Force that has identified thirty projects across thirteen departments. These projects focus on mature technologies—document processing, data analysis, public service chatbots—rather than frontier research. The strategy is explicitly application-driven, positioning Hong Kong as a user and integrator of AI rather than a creator of foundational models. The capital market component is equally significant. AI-related IPOs have raised nearly HKD 100 billion, representing 55% of all new listings in Hong Kong. The Hang Seng Index has added AI companies to its benchmark, creating a self-reinforcing narrative that positions Hong Kong as the go-to venue for AI companies seeking public listings. This is a deliberate strategy to capture the global AI investment wave and channel it through Hong Kong's financial infrastructure. The economic rationale is clear. A research report cited by the Financial Secretary estimates that if small and medium enterprises (SMEs) were to adopt AI at the same rate as large corporations by 2035, it could unlock HKD 65 billion in economic benefits—roughly 2.2% of Hong Kong's GDP. This is a significant but not transformative figure. It suggests that AI is seen as an efficiency tool rather than a fundamental economic restructuring. What is notably absent from the strategy is any mention of compute infrastructure. There is no discussion of building AI data centers, GPU clusters, or supercomputing facilities. There is no mention of data governance frameworks for government AI applications. There is no discussion of algorithmic transparency or bias mitigation. These omissions are not accidental. They reflect a strategic choice to rely on external providers for the heavy lifting while focusing on application-layer integration. This approach has a certain logic. Hong Kong lacks the research base, the talent pool, and the physical infrastructure to compete with Beijing, Shenzhen, or Hangzhou in foundational AI research. By focusing on applications, Hong Kong can leverage its strengths in finance, trade, and professional services to create value without making massive upfront investments in R&D. It is a pragmatic strategy that plays to the city's comparative advantages. But pragmatism has a cost. By positioning itself as an application-layer participant, Hong Kong is accepting a permanent dependency on external model providers and compute infrastructure. This dependency creates vulnerabilities that the policy statement does not address. What happens if the supply of models from mainland China is disrupted? What happens if cloud service providers raise prices or change their terms? What happens if the data governance requirements of the mainland conflict with Hong Kong's international obligations? These are not hypothetical questions. They are structural risks that any serious analysis of Hong Kong's AI strategy must confront. And they are risks that a decentralized approach—one that emphasizes open-source models, distributed compute, and user-controlled data—could mitigate. The Core: A Decentralization Critique of Hong Kong's AI Strategy Let me start with the most obvious problem: the concentration of AI-related IPOs. When 55% of all new listings in a major financial center are AI-related, it is reasonable to ask whether we are witnessing genuine value creation or narrative-driven speculation. I have seen this pattern before. In 2017, during the ICO boom, I rejected dozens of token sale projects that lacked whitepaper substance. Many of them raised millions of dollars based on nothing more than a compelling story and a well-designed website. The parallels to today's AI IPO market are uncomfortable. The Financial Secretary's statement celebrates the HKD 100 billion raised by AI-related IPOs. But it does not address the quality of these companies. How many of them are genuinely AI-native businesses with proprietary technology and defensible moats? How many are traditional companies that have added "AI" to their branding to capture the narrative premium? The history of technology bubbles suggests that when a sector accounts for more than half of all new listings, the average quality of the companies in that sector tends to decline. This is not just a market risk. It is a governance risk. If Hong Kong's capital markets become known as a venue where AI companies can list without rigorous scrutiny of their technical capabilities, the city's reputation as a financial center will suffer. The 55% concentration is not a sign of strength; it is a warning sign of potential froth. The second issue is the reliance on external model providers. Hong Kong does not have its own foundational AI models. It will depend on models from mainland China—such as Alibaba's Qwen or DeepSeek—or from overseas providers like OpenAI and Anthropic. This dependency has both technical and geopolitical dimensions. Technically, relying on external models means that Hong Kong's AI applications are constrained by the capabilities and limitations of those models. If a provider decides to change its pricing, alter its terms of service, or discontinue a model, Hong Kong's applications will be affected. This is a classic vendor lock-in risk, and it is particularly acute in the AI space where models are constantly evolving and providers are consolidating. Geopolitically, the dependency is even more concerning. Hong Kong sits at the intersection of two competing AI ecosystems: the US-led Western ecosystem and the China-led ecosystem. By relying on models from both sides, Hong Kong exposes itself to the risk of being caught in the crossfire of export controls, sanctions, and technology restrictions. The US has already restricted the export of advanced AI chips to China, and it is not difficult to imagine a scenario where model access becomes a tool of geopolitical leverage. A decentralized approach would mitigate these risks. Open-source models, such as Llama or Mistral, can be self-hosted and fine-tuned locally. Distributed compute networks, such as those being built by projects like Gensyn or Akash, offer alternatives to centralized cloud providers. By building on open infrastructure, Hong Kong could reduce its dependency on any single provider and maintain greater control over its AI destiny. The third issue is the absence of compute infrastructure. The policy statement makes no mention of building AI data centers, GPU clusters, or supercomputing facilities. This is a strategic blind spot. Without sovereign compute capacity, Hong Kong's AI applications will be dependent on cloud providers—primarily Alibaba Cloud, Tencent Cloud, and AWS—for their computational needs. This dependency has implications for cost, performance, and data sovereignty. For government AI applications, the issue is particularly acute. Government departments will be processing sensitive citizen data—identity information, tax records, public service usage data. If this data is processed on cloud infrastructure controlled by external providers, questions arise about data residency, access controls, and compliance with privacy regulations. The Hong Kong government has not articulated a clear framework for how it will handle these issues. A decentralized approach would involve building or accessing distributed compute infrastructure that is not controlled by any single entity. This could involve partnerships with regional data center operators, participation in decentralized compute networks, or the development of a Hong Kong-based AI compute cluster. The absence of any discussion of this issue suggests that the government has not fully thought through the implications of its AI strategy. The fourth issue is data governance. The policy statement is silent on how government AI applications will handle citizen data. This is a significant omission. Government AI systems will process vast amounts of personal data, and the potential for misuse, bias, and privacy violations is substantial. Without a clear governance framework, there is a risk that government AI applications will be deployed without adequate safeguards. A decentralized approach to data governance would involve giving citizens control over their data. This could include mechanisms for consent, transparency, and auditability. Blockchain technology offers tools for creating verifiable data trails and ensuring that data is used only for the purposes for which it was collected. By incorporating these principles into its AI strategy, Hong Kong could build a more trustworthy and accountable AI ecosystem. The fifth issue is the lack of attention to algorithmic transparency and bias. Government AI systems that make decisions about citizens—whether in the allocation of public services, the processing of applications, or the detection of fraud—must be transparent and accountable. Citizens have a right to know when AI is being used to make decisions that affect them, and they have a right to challenge those decisions. A decentralized approach would incorporate transparency by design. This could involve publishing the algorithms used in government AI systems, establishing independent audit mechanisms, and creating channels for citizen feedback and appeal. The absence of any discussion of these issues in the policy statement is concerning. The Contrarian Angle: The Case for Pragmatism I have been critical of Hong Kong's AI strategy, but I want to be fair. There is a strong case for the pragmatic, application-first approach that the government has chosen. Let me articulate that case before I offer my final assessment. First, the application-first approach is rational given Hong Kong's constraints. The city does not have the research base, the talent pool, or the physical infrastructure to compete in foundational AI research. Attempting to build a domestic AI model ecosystem would require massive investments with uncertain returns. By focusing on applications, Hong Kong can create value more quickly and with less risk. Second, the capital market strategy is working. AI-related IPOs have raised significant capital, and the Hang Seng Index's inclusion of AI companies is attracting passive investment. This is creating a virtuous cycle where AI companies come to Hong Kong to list, which attracts more AI companies, which strengthens Hong Kong's position as an AI financial hub. Third, the government's efficiency projects are a sensible starting point. By identifying thirty high-value use cases across thirteen departments, the government is demonstrating the practical benefits of AI. This can create momentum and build confidence among businesses and citizens. Fourth, the SME opportunity is real. The HKD 65 billion in potential economic benefits from SME AI adoption is a meaningful prize. If the government can help SMEs overcome the barriers to AI adoption—cost, skills, awareness—it could generate significant economic value. Fifth, Hong Kong's unique position as a bridge between China and global markets gives it advantages that other jurisdictions do not have. By leveraging its access to both mainland Chinese AI technology and international capital, Hong Kong can create a distinctive value proposition. I acknowledge these arguments. They are not without merit. The application-first approach is a reasonable strategy for a city with Hong Kong's constraints. The capital market strategy is generating real results. The efficiency projects are a sensible starting point. But I would argue that these pragmatic choices are not mutually exclusive with a more decentralized approach. Hong Kong could pursue its application-first strategy while also investing in sovereign compute infrastructure. It could promote AI adoption among SMEs while also establishing data governance frameworks that protect citizen privacy. It could leverage its position as a bridge between China and global markets while also building open, transparent AI systems that are not dependent on any single provider. The choice is not between pragmatism and decentralization. The choice is between a short-term, dependency-creating approach and a long-term, sovereignty-preserving approach. The former may deliver faster results, but the latter will build more durable value. Let me offer a concrete example from my own experience. In 2021, I partnered with a collective of indigenous artists to tokenize cultural heritage data on Polygon. Rather than focusing on speculative resale value, we implemented a smart contract mechanism that ensured 5% of all secondary sales funded local community preservation projects. The project involved 150 unique assets and demonstrated how blockchain could facilitate equitable value distribution. The key insight from that project was that the technology was not the hard part. The hard part was designing governance mechanisms that ensured the benefits flowed to the right people. We spent months working with the artists to understand their needs, their concerns, and their vision for the project. We built trust through transparency and accountability. Hong Kong's AI strategy would benefit from a similar approach. Instead of simply deploying AI applications and hoping for the best, the government should engage with citizens, businesses, and civil society to understand their needs and concerns. It should build trust through transparency and accountability. It should design governance mechanisms that ensure the benefits of AI are distributed equitably. This is not a naive call for consensus. It is a practical recognition that trust is the foundation of any successful technology deployment. Code is the new covenant, but trust is the ink. Without trust, even the most sophisticated AI systems will fail. The Takeaway: A Call for Sovereign AI Hong Kong stands at a crossroads. The path it chooses will determine whether it becomes a genuine AI hub or a dependent periphery of larger AI powers. The current strategy—application-first, capital-market-driven, and silent on infrastructure and governance—is a bet that Hong Kong can create value by integrating and applying AI technologies developed elsewhere. It is a bet that may pay off in the short term, but it carries significant long-term risks. A more decentralized approach would be different. It would involve investing in sovereign compute infrastructure, promoting open-source models, establishing transparent data governance frameworks, and building AI systems that are accountable to citizens. It would be slower and more complex, but it would create more durable value. I am not suggesting that Hong Kong should abandon its pragmatic approach. I am suggesting that it should complement that approach with investments in sovereignty and resilience. The city has the resources, the talent, and the strategic position to build a more robust AI ecosystem. The question is whether it has the vision. Ownership is not a receipt; it is a soul. Hong Kong's AI strategy should be about more than capturing economic value. It should be about building a system that respects the dignity and agency of its citizens. It should be about creating an AI ecosystem that is transparent, accountable, and equitable. It should be about ensuring that the benefits of AI are shared broadly, not concentrated in the hands of a few. In the chaos of consensus, I seek the quiet truth. The quiet truth is that Hong Kong's AI strategy, as currently conceived, is a centralized blueprint in a decentralized world. It is a strategy that concentrates power rather than distributing it. It is a strategy that creates dependencies rather than sovereignty. It is a strategy that prioritizes speed over resilience. There is still time to change course. The thirty efficiency projects are just beginning. The capital market strategy is still evolving. The infrastructure decisions have not yet been made. Hong Kong can choose a different path—a path that embraces the principles of decentralization and builds an AI ecosystem that is truly sovereign, transparent, and equitable. The choice is not easy. It requires vision, courage, and a willingness to challenge conventional wisdom. But it is a choice that will determine Hong Kong's place in the AI era. I hope the city's leaders have the wisdom to make the right one.