Silence is the first vote in a true consensus.
When news broke that Zhongji Innolight, the Chinese optical module giant that rode the AI wave to become the largest weighting in the CSI 300, was preparing a Hong Kong IPO of up to $8 billion with BlackRock, Hillhouse, and Temasek as cornerstone investors, the blockchain community did not erupt. There was no Twitter storm, no DAO proposal, no gnashing of digital teeth. Just silence. And in that silence, a vote was cast—a vote for the old world, for the familiar altar of Wall Street capital, for the very system Satoshi Nakamoto asked us to transcend.
I have spent the last eight years auditing the ethical architecture of decentralized systems. From the smouldering wreckage of The DAO in 2017—where I pored over Etherscan logs to find not just reentrancy bugs but a moral vacuum—to the governance design of MakerDAO in 2020, where I sat in twelve virtual town halls listening to small holders plead for a voice, I have believed that decentralization is not merely a technical upgrade but a covenant of human empowerment. Yet as I read the term sheet for Zhongji Innolight, I felt a familiar chill. This was not a disruption; it was an absorption.
Context: The Machine in the Background
Zhongji Innolight is not a household name outside tech finance circles, but it sits at the throat of the AI revolution. It manufactures the high-speed optical modules—800G, 1.6T—that connect the GPUs inside data centres. Without them, Google's Gemini, Microsoft's Copilot, and Meta's Llama choke on bandwidth. The company became the largest weight in the CSI 300, overtaking CATL, the battery behemoth, in a symbolic handover from the electric vehicle era to the artificial intelligence era. Its Hong Kong IPO, rumoured to be the largest in seven years, is designed to raise up to $8 billion to expand production and push into next-generation silicon photonics.
The cornerstone investors are a who's-who of global capital. BlackRock alone manages $10 trillion. Hillhouse and Temasek are legendary allocators of patient money to China's tech winners. This is not speculative retail money; this is the deep, cold ocean of institutional conviction. The message is clear: AI infrastructure is the new oil, and the most efficient drillers are in China.
But for those of us who believe that Web3 was supposed to offer an alternative model—a peer-to-peer fabric for compute, bandwidth, and storage—this IPO feels like a defeat. We have spent years building decentralized compute networks like Golem, iExec, and Akash. We have argued that AI training could be outsourced to a global mesh of idle GPUs, that the profits of the intelligence age should flow to the many, not the few. And yet here is the market voting with $8 billion that the centralized, vertically integrated model wins.
Core: When Code Is Not Law, Capital Is
My own journey through the cryptosphere began with an ethical audit. After The DAO hack in 2017, I wrote a 30-page whitepaper titled Code is Not Law: The Moral Vacuum in Smart Contracts. I argued that technical efficiency without ethical governance leads to societal harm. The reentrancy bug was not the real flaw; the flaw was that the protocol had no mechanism for human override, no forum for moral deliberation. Code executed, and people lost everything.
Now, in 2025, I see the same pattern emerging in AI. The underlying infrastructure—optical modules, GPUs, data centres—is being built by centralized entities financed by Wall Street. The profits will accrue to the shareholders of Zhongji Innolight, Nvidia, and the hyperscalers. The workers, the users, the broader society will receive what is left. This is not a bug; it is a feature of the current financial system. And the blockchain community, for all its talk of sovereignty, is largely cheering from the sidelines, building meme coins and layer-2 bridges while the real means of production are locked behind corporate walls.
In 2020, I consulted for a mid-sized DAO that wanted to redesign its governance tokenomics. I proposed quadratic voting to prevent whale dominance. We ran twelve town halls. We listened. The proposal passed, and unique voter participation increased by 40% over six months. I learned that true decentralization requires emotional inclusion—not just algorithmic fairness, but a sense that every participant is seen and heard. Zhongji Innolight has no such mechanism. Its shareholders vote with dollars, not with voice. Its governance is a product of boardrooms, not of communities.
Yet the market loves it. The IPO is oversubscribed. The analysts call it a “must-have” for any portfolio exposed to AI. The silence from the blockchain camp is deafening. We have been so busy building our own sandcastles that we forgot to look up and see the empire rising.
Contrarian: The Pragmatist's Heresy
Let me play the devil’s advocate for a moment. Perhaps this IPO is precisely what the decentralized AI movement needs. The $8 billion will flow into R&D for silicon photonics, which could lower the cost of optical interconnects, making it easier for smaller players to build their own compute clusters. Scale has a way of commoditizing yesterday’s luxury. The same happened with cloud computing: AWS made compute cheap, and then we built decentralized storage on top of it. Maybe Zhongji Innolight is the on-ramp to a more distributed future.
But I reject that comfort. I retreated to a cabin on Hiiumaa island in 2022, after FTX collapsed, and spent six weeks disconnected. I wrote a manifesto, The Hollow Promise of Yield, that went viral for its raw honesty. I realized that much of what we called “innovation” was financial engineering wearing a mask of disruption. The same is true here. The IPO is not an on-ramp; it is a moat. It deepens the dependence of the AI supply chain on a single company, on a single country, on a single financial system. If the goal of decentralization is resilience, then this is its antithesis.
Moreover, the foundation of Zhongji Innolight’s growth is the AI capital expenditure cycle of the hyperscalers. If that cycle turns—if Google or Microsoft decide to pull back—the stock will crater. But more importantly, the entire centralized AI edifice will tremble. A decentralized web of compute nodes, by contrast, would not have a single point of failure. The market, in its wisdom, has chosen fragility over antifragility.
Consensus is not a transaction; it is a covenant. That covenant requires participants to share in the risks and rewards. Zhongji Innolight’s IPO is a transaction. It extracts capital from the world and concentrates it in a single entity. The covenant of blockchain—the promise that we can build systems where value flows to those who contribute—is broken every time we applaud such concentration.
Takeaway: The Human in the Loop
In 2026, I worked on a decentralized identity protocol for Tallinn’s AI startup hub. We integrated ZK-proofs into AI agent wallets, ensuring autonomous agents could prove their origin without revealing proprietary data. The pilot involved 100 AI agents and $5 million in secure transactions. I wrote a column called “The Human in the Loop,” arguing that privacy is a fundamental right in an automated age.
Zhongji Innolight’s IPO is not just a financial event; it is a moral referendum on the direction of AI infrastructure. Will we allow the most critical layer of the intelligence age to be controlled by a handful of corporations and their Wall Street patrons? Or will we build a parallel, decentralized stack that is owned by its users?
The silence of the blockchain community today tells me we have already made our choice. But silence is also the first vote in a true consensus. If we can break that silence, if we can articulate an alternative vision with the same clarity and conviction that BlackRock brings to its pitch books, then perhaps we can still build the world we promised each other in 2017.
Until then, I will be in my cabin on Hiiumaa, listening to the wind, and wondering if the price of progress is the soul of the revolution.
In the absence of a covenant, code becomes a cage. Winter teaches what spring forgets.