Hook
What if the city that wants to embed AI into 90% of its devices by 2030 has no way to prove that the AI hasn't been tampered with? Chengdu's “AI+” action plan, targeting 260 billion yuan in core industry scale, reads like a Silicon Valley dream—except it’s missing a critical layer: a trust machine. I’ve seen this pattern before. In 2017, my own DAO experiment in Cape Town collapsed not because the idea was wrong, but because we had no infrastructure to verify our intentions on-chain. Now, as I look at Chengdu’s numbers, I can feel the same tension: ambition without transparency is just a promise waiting to be broken.
Context
Last month, the Chengdu municipal government released a sweeping “AI+” action plan. The headline numbers are huge: 260 billion yuan ($36B) in core AI industry revenue by 2030, with “new-generation smart terminals and agents” reaching over 70% penetration by 2027 and 90% by 2030. They promise 100 innovative products, 100 demonstration scenarios, and 20 benchmark scenes each year. It sounds like a typical government push, but the details reveal a dangerous gap. The policy is all about scale and speed—zero about how to audit, trace, or secure the AI systems that will run half the city’s devices. As a Web3 founder who has spent years building communities around cryptographic truth, I see this as a classic “code is law, but people are truth” moment. Chengdu needs a blockchain layer, and it needs it now.
Core: The Seven Dimensions Through a Lens of Decentralization
Let me break down the plan using the same seven dimensions I use when evaluating protocols—because a city’s AI strategy is just a protocol for resource allocation and trust.
1. Technical Route – The Missing Verifiability
The plan mentions “new-generation smart terminals and agents” but never defines how their decisions will be logged or verified. In blockchain terms, this is like launching a DeFi protocol without a public ledger. Without on-chain provenance, how do we know if an AI agent that approves a loan or diagnoses a patient has been corrupted by biased training data? I remember when we were building TruthChain in 2026—a project to authenticate AI-generated content—the hardest part wasn’t the AI, it was getting people to believe the provenance. Chengdu is repeating the same mistake: focusing on scale without building the verification rails. The real technical challenge isn’t hitting 90% penetration; it’s ensuring that 90% of those terminals can prove their integrity.
2. Commercialization – The Subsidy Trap
The plan relies heavily on government procurement and subsidies—100 demonstration scenes, 20 benchmark projects per year. This is classic “dumb money” strategy, akin to the yield farming frenzy I experienced in 2020. Back then, chasing 100% APYs taught me that subsidies mask true commercial viability. Chengdu’s 260 billion target will be built on a mix of genuine revenue and inflated “AI-washing” from traditional electronics. Without a transparent on-chain record of which projects actually deliver value, it’s easy for firms to double-count or fake metrics. I’ve seen this before in the NFT space: projects that launched with a bang and then went silent once the hype subsidies ended. The city needs a verifiable mechanism—like a DAO-based grant tracker or an on-chain reputation system—to ensure that every subsidized AI project produces measurable outcomes.
3. Industry Impact – A Real Opportunity for Web3
Chengdu’s strengths—electronics manufacturing, finance, tourism, education—are perfect for Web3 integrations. Imagine a smart city where each AI terminal’s decisions are hashed onto a public chain, allowing citizens to verify that the traffic light’s timing wasn’t manipulated, or that a bank’s AI credit score wasn’t biased. This is not just a nice-to-have; it’s a regulatory necessity in a world where the EU AI Act and China’s own AI regulations demand auditability. During my AfricanCode project, I saw how artists embraced NFTs as a way to prove originality and ownership. The same logic applies to AI—let the output be rooted on-chain. The 100 demonstration scenes could become the world’s largest showcase of “proof-of-inference” if they adopt blockchain.
4. Competition – Differentiation Through Trust
Chengdu positions itself as an “AI application first city,” competing with Beijing (research), Shenzhen (hardware), and Hangzhou (cloud). But none of them have yet claimed “most trustworthy AI city.” By embedding a public blockchain into the AI infrastructure—like requiring all deployed models to log their inference hashes on a national blockchain—Chengdu could leapfrog. I’ve seen this in the DeFi space: projects that emphasize transparency (like Uniswap’s open-source code) attract more liquidity than those with opaque, centralized risks. The window to claim the trust narrative is maybe two years. After that, every city will copy.
5. Ethics & Security – The Vacuum That Blockchain Fills
The plan has zero mention of AI safety, ethical review, or data privacy. This is a red flag. In Web3, we’ve learned the hard way that “code is law” isn’t enough—you need social consensus and transparent governance. When I led the Cape Town DAO, our failure partly stemmed from not having a clear mechanism for dispute resolution. Chengdu’s high-risk AI applications (e.g., in healthcare, finance, autonomous driving) will inevitably face incidents. Without an immutable audit trail, blame and liability become a political mess. A blockchain-based audit log—think of it as a “black box” for AI—could provide the forensic evidence needed to hold both companies and government accountable. The city should adopt the same principle we used in TruthChain: every AI decision that affects a human should be timestamped and signed.
6. Investment – The On-Chain Alpha
Local concept stocks (like Jiafa Education, Creative Information) will pump on this news—that’s short-term noise. But long-term value lies in companies that integrate verifiable AI. I’ve seen this in the Layer2 space: after the Dencun upgrade, rollup gas fees dropped, but only those with transparent sequencing and data availability won real adoption. Similarly, AI firms that tokenize their model’s inference logs or issue on-chain attestations will attract premium valuations. The 260 billion target implies a 30% CAGR—optimistic but possible if you count the multiplied effect of on-chain transparency. Investors should look for startups that use blockchain to prove their AI’s integrity, not just its speed.
7. Infrastructure – The Decentralized Compute Angle
Chengdu has two major compute centers (Tianfu Smart Computing Center, Chengdu Supercomputing Center) with plans to reach 1000P by 2025. But as we saw with the gas fee crisis after Dencun, centralized compute nodes are bottlenecks. During the 2022 bear market, I realized that decentralized compute networks (like Akash or io.net) could offer cheaper, censorship-resistant alternatives. Chengdu could partner with these networks to run their AI inference—especially for privacy-sensitive applications. If they lock all their AI compute to a single vendor, they face a single point of failure; if they use a mix of on-premise and decentralized cloud, they gain resilience. The real question: can Chengdu’s AI plan scale without creating a centralized infrastructure monopoly? Based on my experience, no—unless they embrace Web3.
Contrarian Angle – The Risk of Over-Engineering Trust
But here’s the counter-intuitive truth: adding blockchain to every AI system could kill speed and increase costs. I’ve watched many projects waste months building token-gated demos that nobody used. Chengdu’s plan is already ambitious; adding on-chain verification for every terminal could throttle the deployment pace. The contrarian view is that the city should start small—only require blockchain provenance for high-stakes applications (health, finance, autonomous driving) and let consumer devices run without it. The best is the enemy of the good. Don’t let decentralization idealism slow down the AI rollout in low-risk areas. In 2021, during the NFT cultural renaissance, I learned that not everything needs to be on-chain—utility tokens died, but art thrived because it chose its battles. Chengdu needs to choose its battles too.
Takeaway
Chengdu stands at a crossroads. It can build the most powerful centralized AI hub in China, vulnerable to bias, regulation, and single points of failure. Or it can become the world’s first “Decentralized AI City”—using blockchain to ensure every smart terminal is auditable, every model is verifiable, and every citizen’s data is under their control. Embrace the volatility of building truly transparent systems; the signal is that trust is the scarcest resource in the AI era. I’ll be watching which path they choose. And if they decide to go the Web3 way, I know exactly where my next community meetup will be: Cape Town meets Chengdu.
Vibes > Algorithms | Code is law, but people are truth | Embrace the volatility, find the signal | Build in public, live in truth