Last week, the Chengdu municipal government released its "AI+" Action Plan, targeting a core AI industry scale of 260 billion yuan by 2027 and a penetration rate of over 70% for "next-generation intelligent terminals and agents." The plan promises 100 innovative products, 100 demonstration scenarios, and 20 annual flagship projects. It sounds like a classic government-led industrial surge. But as someone who has spent four years auditing smart contracts and building decentralized communities, I see a glaring omission: not a single mention of blockchain, decentralized identity, or data sovereignty. This is a missed opportunity of epic proportions.
Let me be clear. The "AI+" plan is not wrong. It's just incomplete. By ignoring the foundational infrastructure of trust that blockchain provides, Chengdu risks building an AI ecosystem that is efficient but fragile, scalable but opaque, and profitable at the cost of user autonomy. From my experience auditing the Telegram Open Network in 2017, I learned that technical correctness without social empathy leads to community fragmentation. The same principle applies here: an AI strategy without decentralized accountability is a house built on sand.
Context: What the Plan Actually Says
The plan's core metrics are ambitious: by 2027, the AI industry scale should exceed 260 billion yuan, with smart terminal penetration above 70%. By 2030, penetration should surpass 90%. The focus is on "next-generation" terminals and agents—think edge AI, IoT devices, and autonomous software agents. The government will fund 20 demonstration scenarios annually, covering manufacturing, finance, healthcare, and cultural tourism. The underlying assumption is that existing AI models (likely from Huawei, Baidu, or local startups) will be integrated into every vertical, driven by subsidies and procurement orders.
But the plan says nothing about how these AI systems will handle data ownership, algorithmic transparency, or user consent. It does not mention decentralized identifiers (DIDs), verifiable credentials, or on-chain audit trails. It treats AI as a purely technical upgrade, ignoring the sociological and ethical layers that blockchain was designed to solve. This is exactly the kind of blind spot I encountered during the 2020 DeFi Summer, when I translated complex Aave upgrade proposals into empathetic guides for Mumbai's retail investors. Back then, trust was the missing ingredient. Today, it still is.
Core Analysis: Where Blockchain Should Have Been Embedded
Let me walk through five critical dimensions where blockchain could transform Chengdu's AI plan from a top-down mandate into a bottom-up trust network.
1. Data Sovereignty and Consent
The plan envisions smart terminals with 90% penetration. Imagine tens of millions of AI-powered cameras, voice assistants, and health monitors collecting data from citizens every second. Without a decentralized identity layer, all that data flows into centralized corporate servers—exactly the model that has led to surveillance capitalism. During my work with the Mumbai Chain Guardians in 2020, we saw how centralized oracle systems could be manipulated. Similarly, centralized AI data silos are vulnerable to abuse. By implementing DIDs on a public blockchain, citizens could grant granular permissions for data use, revocable at any time. The city could issue verifiable credentials for access to public services, while keeping biometric data encrypted on personal devices. The plan's lack of this layer is a security hole that will only grow wider as deployment scales.
2. Algorithmic Transparency and Auditing
When AI systems make decisions about loans, job applications, or insurance claims, how can a citizen verify that the algorithm wasn't biased? In traditional systems, you trust the auditor. In blockchain, you audit the code itself. During my forensic audit of the TON whitepaper, I found a game-theory flaw that would have excluded small holders—a problem that could have been caught earlier with on-chain governance logs. For Chengdu's AI, every model update and inference request could be hashed onto a public ledger. Smart contracts could enforce rules like "no discrimination based on age or zip code" directly at the execution layer. The plan currently relies on government oversight, but history shows that regulatory capture is a real risk. On-chain transparency is not just nice-to-have; it's the only way to build trust at scale.
3. Incentive Design for Data Sharing
The plan assumes that AI companies will voluntarily share data to train better models. That's naive. In every industry I've consulted—from finance to heritage preservation—the bottleneck is data access. During my 2021 Heritage on Chain project, we tokenized endangered textile patterns as NFTs, giving artisans direct ownership and a revenue stream. The same principle applies here: citizens could be rewarded with tokens for contributing health or mobility data, creating a decentralized marketplace for AI training sets. Chengdu could issue a "Data Dividend Coin" that appreciates as the AI ecosystem grows. This aligns incentives between the city, companies, and individuals, rather than relying on top-down mandates that often breed resentment and evasion.
4. Decentralized Compute and Verifiable Inference
The plan highlights the Tianfu Intelligent Computing Center and Chengdu Supercomputing Center as infrastructure pillars. But centralized compute nodes are single points of failure—both technically and politically. In my experience with the 2022 bear market counseling circles, I saw how centralized platforms (like FTX) collapsed overnight, leaving users helpless. For AI, if the central compute cluster goes down or is compromised, the entire smart city stalls. A better approach is to build a peer-to-peer compute network where idle devices (smartphones, edge servers) contribute processing power to run AI inferences, verified by zero-knowledge proofs. This not only improves resilience but also lowers costs. The plan's silence on distributed ledger technology is a gap that will become critical as the number of AI agents explodes.
5. Ethical AI Governance Through DAOs
The plan includes zero discussion of ethics or safety—a dangerous omission given that Europe's AI Act and China's own generative AI regulations demand accountability. I led the drafting of the Decentralized AI Bill of Rights in 2026, a consensus document signed by 500 Web3 organizations. The core insight is that ethical rules must be encoded, not just stated. Chengdu could launch a Decentralized Autonomous Organization (DAO) for AI governance, where citizens, developers, and regulators collectively vote on model updates, data usage policies, and dispute resolution. The 20 annual demonstration scenarios could be governed by smart contracts that release funds only after passing automated ethical audits. This turns compliance from a checkbox into a living process—exactly what I've advocated for since 2017.
Contrarian Angle: Is Decentralization Even Necessary Here?
Some will argue that China's centralized governance model is precisely what enables rapid AI deployment—that in Shenzhen, Hangzhou, and Beijing, similar plans have succeeded without blockchain. They'll point to the fact that 99% of rollups don't generate enough data to need dedicated DA (my own opinion on Layer2 hype). Why should Chengdu add unnecessary complexity? After all, the plan's goal is economic growth, not digital freedom.
I understand this perspective. I've sat in meetings where executives said, "Trust is not a protocol, it is a practice"—and then proceeded to ignore the protocol anyway. But here's the contrarian truth: history shows that the most successful decentralized systems don't compete on speed; they compete on resilience. The 2008 financial crisis was caused by opaque, centralized risk management. The 2022 crypto winter was partly caused by centralized lending platforms that hid their liabilities. Chengdu's AI plan, if fully centralized, creates a systemic risk where one bad model update or one data breach could undermine trust in the entire ecosystem. It's not about adding blockchain because it's trendy; it's about building a foundation that can survive shocks.
Moreover, the plan's own metrics—like the 70% penetration target for smart terminals—will generate massive amounts of sensitive data. Without decentralized identity, that data becomes a honeypot for hackers and a surveillance tool for overreach. The Chinese government itself has recognized the need for data privacy with the Personal Information Protection Law (PIPL). Blockchain is the only technical framework that can implement PIPL's consent and deletion requirements at scale, without relying on a single company's promise. From code audits to community heartbeats, I've learned that technology without accountability is just marketing. Chengdu's plan is currently a marketing document for AI, not a blueprint for sustainable digital civilization.
Takeaway: From Centralized Ambition to Decentralized Action
Chengdu's "AI+" plan has the right ambition: 260 billion yuan, 90% penetration, 100 demonstration scenarios. But ambition without architectural integrity creates castles that crumble. My experience in Web3 has taught me one thing unequivocally: Trust is not a protocol, it is a practice. The protocols—public blockchains, zero-knowledge proofs, DIDs—are tools to enable that practice. By ignoring them, Chengdu is building an 8-lane highway without guardrails. The speed will be impressive, but the first accident will be catastrophic.
I urge the policymakers and AI entrepreneurs in Chengdu to add a Section 8 to their plan: "Decentralized Infrastructure for Ethical AI." Embed verifiable credentials into every smart terminal. Create a data dividend token for citizens who contribute to training models. Establish a DAO to govern the annual demonstration scenarios. Audit not just the code but the soul behind the smart contract. The plan's current missing piece is not a technical oversight—it's a philosophical one. Building bridges where DeFi once built walls means extending that bridge-building to AI. The city has a chance to lead not just in scale, but in soul. I hope they take it.
As I write this from my desk in Mumbai, where I've seen both the power of centralized systems (India's Aadhaar) and their dangers (data leaks), I remain optimistic. The Web3 community—with its 500 organizations behind the Decentralized AI Bill of Rights—stands ready to help. The question is whether Chengdu will invite us in or build its own wall. The audit was just the beginning of the bond; the bond is the foundation of trust. And trust, I've learned, is the only yield that compounds forever.