At the San Francisco AI Summit this week, South Korean President Lee Jae-myung is scheduled to meet with the CEOs of Nvidia, OpenAI, Anthropic, and Broadcom. This is not just a diplomatic photo opportunity; it is a clear signal that nation-states are placing their strategic bets on centralized AI infrastructure. As an open-source evangelist who spent 600 hours auditing the Aave V2 interest rate models in 2020, I recognize the pattern: the promise of efficiency masking the consolidation of power.
Consider the context. The blockchain movement was built on a simple premise—decentralization through trust minimized by code, not by institutions. Yet here we have a democratically elected leader prioritizing private, closed-source giants. Nvidia’s CUDA lock-in, OpenAI’s proprietary model weights, Broadcom’s networking monopoly—these are the pillars of the new AI order. Even Anthropic, with its heavily marketed “constitutional AI,” remains a corporate entity where the final decisions rest in a boardroom, not a consensus protocol. Code is law, but ethics is soul—and when the source of that soul is a single company, the law becomes fragile.
From my experience in the Ethereum ecosystem, I have seen how centralization creeps in: the reliance on Infura for RPC access, the subtle coordination among a handful of validator pools, the developer dependence on a small core team. Now imagine an entire country’s AI strategy depending on a handful of US-based servers and proprietary APIs. The “compute sovereignty” anxiety that many nations feel is real—but the solution offered by this summit is not sovereignty; it is deeper dependency. In my 2024 work on the Verifiable Humanity initiative, I used zero-knowledge proofs to preserve privacy while verifying humanness. That project proved that decentralized verification is possible, even at scale. Yet here, the Korean government is likely negotiating for GPU quotas and model access, not for open protocols or verifiable computation frameworks.
The technical risk is glaring: AI becomes a utility controlled by a few, with no possibility for community governance, forkability, or on-chain audit. When the president meets Nvidia’s Jensen Huang, the conversation will likely center on securing H100 or B200 allocations. But from a blockchain perspective, the real question should be: Who verifies the compute? Who ensures that a model’s training data isn’t censored, or that the inference outcomes aren’t manipulated? Transparency isn’t the oxygen of trust—verifiability is.
Let me now offer a contrarian angle within our own community. Are we not also guilty of similar centralization? Many blockchains rely on a small group of core developers. Ethereum’s transition to proof-of-stake concentrated influence among a few large staking pools like Lido and Coinbase. The irony is that while we condemn Big Tech, we often replicate their structures in the name of efficiency. The Korean president’s move is merely a mirror of our own compromises. During the 2022 bear market, I mentored ten junior developers on building resilient systems. Our manifesto, “Code as Law, but People as Gods,” emphasized that even the most decentralized code rests on human coordination. The Korean AI summit is no different: it is a political act, not a purely technical one. The real test is not to oppose centralization outright, but to build decentralized alternatives that can actually compete—alternatives that provide verifiable, sovereign compute for AI workloads without sacrificing privacy or security.
What does this mean for the blockchain space? First, the demand for decentralized compute will only grow. Projects like Akash Network and Golem have been early movers, but they need real-world adoption from enterprises and governments. Second, the AI safety framework that Anthropic champions could be integrated with blockchain-based audit trails. Imagine a DAO that governs an open-source AI model, where every update is signed by a quorum, and every inference is accompanied by a zero-knowledge proof. That is the infrastructure we should be building, not merely reacting to centralized power moves.
Open source is not a business model; it’s a commitment. The Korean president’s courtship of the AI giants is a wake-up call. Fifty years from now, will we look back and see this summit as the moment when we chose convenience over resilience? Or will it be the catalyst for a truly decentralized AI ecosystem? The choice does not lie in Seoul’s blue house—it lies in every open-source repository, every smart contract audit, every Verifiable Humanity initiative. Guard the commons, or lose the future.