Last week, the National Development and Reform Commission published a 5,000-word AI Cooperation Action Plan. Most read it as a tech policy. I read the four pillars—data circulation, compute inclusiveness, open-source collaboration, and green decarbonization—and saw the architectural skeleton for a state-backed compute and data fabric that will directly compete with Ethereum’s rollup roadmap and challenge the economic axioms of permissionless blockchains.
We do not predict the future; we hedge against it.
Context – The Four Pillars as Blockchain Infrastructure
The plan’s core is not about training better models; it is about building the underlying resource layers: a unified high-quality multilingual corpus, an interconnected intelligent compute pool with universal access, a co-developed open-source ecosystem with compliance standards, and a requirement for energy efficiency. These are exactly the primitives that permissionless networks have been trying to bootstrap through tokens: data sovereignty (Filecoin, Arweave), compute sharing (io.net, Render), open-source governance (Ethereum’s EIP process), and green PoS.
But here is the twist: the Chinese version will be permissioned, subsidized, and regulated. The ‘trusted cross-border data spaces’ mentioned in the document are essentially enterprise-grade, licensed blockchain frameworks. The ‘high-quality corpus’ will be owned by the state-led consortium. The compute interconnectivity will be managed by state-owned operators.
Core – The Mechanics of a State-Run DePIN
Let’s stress-test the compute pillar. The plan calls for ‘interconnection of intelligent computing facilities’ and provision of affordable computing services to developing countries. As someone who has audited MEV bots and EigenLayer’s restaking contracts, I immediately saw the engineering challenge: you need a unified scheduling protocol, a settlement layer, and trustless verification across heterogeneous hardware (NVIDIA vs. Huawei Ascend).
Private blockchains can solve coordination but not trust. The Chinese solution will likely use a permissioned ledger operated by the ‘Big Three’ telcos—China Mobile, China Unicom, China Telecom—which already control China’s telecommunications infrastructure. This ledger would track compute usage, settle payments in fiat, and enforce compliance rules via smart contracts written in a custom EVM variant that embeds content moderation filters.
Based on my EigenLayer audit in 2023, I discovered that slashing mechanisms in permissioned systems have very different incentive structures. EigenLayer uses economic stakes; a Chinese version would use identity and legal risks. This changes the security model entirely. It means the network does not need to incentivize decentralization—it can trust the state backstop. But this also makes it opaque. If you are a yield farmer, you cannot verify the compute usage you paid for. The chain is a black box.
Contrarian – The Real Attack on Permissionless Crypto
Everyone is cheering this plan as an AI catalyst. But from a DeFi perspective, it is the most dangerous competitor to Ethereum’s rollup-centric roadmap. While the West debates L2 fragmentation and liquidity, China is building a single, unified, state-funded compute fabric that will be far more efficient for regulated applications like trade finance, carbon credits, and stablecoins. Retail investors in Southeast Asia will flock to this because it is cheaper and officially endorsed.
Structure defines value; chaos destroys it.
The plan’s open-source community building will also attract developers from the Global South who are priced out of Ethereum gas fees and daunted by Solana’s latency. The ‘co-developed compliance system’ could become the de facto standard for African and Southeast Asian fintechs. The result: permissionless networks lose mindshare and network effects in emerging markets.
Moreover, the ‘green low-carbon’ requirement will impose energy disclosure standards that make Proof-of-Work and even Proof-of-Stake nodes appear wasteful compared to a permissioned state-run cluster using liquid cooling. This shifts the regulatory narrative against permissionless mining.
Takeaway – Three Actionable Levels
We do not predict the future; we hedge against it.
- Short-term (0–6 months): Monitor for pilot projects between China and ASEAN on ‘trusted data spaces.’ Any partnership announcement involving a blockchain platform (e.g., Hyperledger or a local variant) will signal the start of the bifurcation. Buy ETH? No. Buy the token of any DePIN project that has secured a partnership with a Chinese state-owned enterprise.
- Mid-term (6–18 months): Short the yield-bearing stablecoin protocols that rely on cross-border data flows between the West and Asia—they will face compliance catch-22. Long the native token of the Chinese state compute network if it issues one.
- Long-term (18+ months): Expect the emergence of a two-tier crypto world: one permissioned, subsidized, and liquid; the other permissionless, trust-minimized, but capital-constrained. The alpha lies in protocols that can bridge these two worlds with zero-knowledge proofs and compliant identity frameworks.
Survival in DeFi is not about being right on narrative; it is about being the last one holding the hedge.