We are told that the future of AI belongs to the giants—the ones with the deepest pockets and the most GPUs. When Keling AI, a spinout from Kuaishou, raised $3 billion in a single round last week, the narrative wrote itself: another Chinese AI unicorn, another leap toward AGI, another 7.56% pop in the parent company's stock. But what if I told you this funding is actually a glaring signal that the centralized AI model is broken? What if the only real path to trustworthy, scalable intelligence runs through decentralized infrastructure?
This isn't a hot take. It’s a cold analysis from someone who has spent the last three years bridging the gap between finance and protocol design. I’ve watched bear markets kill hype and bull markets mask flaws. And right now, the market is euphoric about AI—but it’s ignoring the fundamental asymmetry of power embedded in these funding rounds.
The Context: Keling AI and the Centralization Gambit
Keling AI is not a blockchain project. It’s an AI video generation company incubated by Kuaishou, the Chinese short-video giant. Its $3 billion raise (reportedly at a pre-money valuation north of $15 billion) places it in the same echelon as Zhipu AI, MiniMax, and Moonshot AI. The stated use case: generate high-quality short videos using diffusion transformers and multimodal models. The implied use case: turn Kuaishou’s billions of user-generated videos into a proprietary training dataset that no competitor can replicate.
The market loved it. Kuaishou’s stock jumped 7.56% on heavy volume. But here’s the part that makes my decentralized heart both excited and uneasy: not a single line in the news coverage mentions how that training data is sourced, who owns it, or what happens when the model starts generating deepfakes. The entire narrative is about valuation and hype.
I’ve seen this movie before. In 2020, during DeFi Summer, every project that raised $50 million was promising to “disrupt finance.” Most of them were just forks with better marketing. Keling AI is different—it has real products, real users, and real revenue potential. But its centralization creates vulnerabilities that no amount of capital can fix.
The Core Insight: Data Sovereignty Is the Missing Layer
Based on my audit experience in protocol design, I can tell you that the most valuable asset of any AI company is not its model weights—it’s the data pipeline. Kuaishou’s users are generating thousands of hours of video daily. Their faces, their voices, their creative expressions are being fed into Keling AI’s training set. In exchange, they get better video filters. That’s a terrible deal.
Decentralization is a verb, not a noun. It’s not about where you host your model; it’s about who controls the value generated from user contributions. Keling AI, despite its $3 billion war chest, is essentially repeating the Web2 playbook: extract user data, train a proprietary model, sell access back to users. The users have no stake, no governance, no ability to audit the model’s behavior.
Contrast this with the emerging decentralized AI stack. Projects like Render Network, Akash, and Bittensor are building markets where compute, data, and inference are all permissionless. A creator could tokenize their video dataset, sell access to training algorithms, and earn royalties every time a model is used commercially. Keling AI could do this—but it won’t, because its business model relies on owning the data, not enabling the community.
This is where the contrarian in me gets loud. Most analysts will tell you Keling AI’s $3 billion validates the AI video sector. I say it validates the thesis that centralized AI will inevitably hit a trust wall. The next generation of users—especially Gen Z creators—will demand transparency and ownership. Keling AI’s valuation already prices in a future of regulatory backlash and data rights battles.
The Contrarian Angle: Why Keling AI’s Funding Accelerates Decentralization
Here’s the counter-intuitive truth: every massive centralization event creates the conditions for its opposite. The dot-com bubble gave us open-source protocols. The data leashes of Facebook and Google gave us zero-knowledge proofs and self-sovereign identity. Keling AI’s funding will do the same for decentralized AI.
Think about it. $3 billion is enough to buy 150,000 H100 GPUs. That’s a lot of compute. But it’s also a lot of centralized control over the world’s AI capacity. If Keling AI becomes a gatekeeper for video generation, it will bottleneck innovation in creative industries. The only way to break that bottleneck is a decentralized compute network that allows anyone to train and run models without asking permission.
Based on my work at a Layer-2 scaling solution, I’ve seen how permissioned systems inevitably leak value. In 2024, I led a project called “Ethical Bridge” that mapped DeFi concepts to traditional corporate governance. The pain point was always the same: central parties extract rents because they control the pipe. Keling AI is building a pipe. The decentralized AI community is building a mesh.
And yes, I know the objections: “Orderbook DEXs will never beat CEXs because market makers won’t leave quotes on-chain to be front-run.” That’s true—for finance. But for AI, the front-running risk is different. A decentralized inference market doesn’t need to match quote latency. It needs to match trust and cost. A farmer in Vietnam shouldn’t have to call an API that costs $0.10 per second and is controlled by a Chinese corporation. They should run their own local model, trained on community-validated data, settled on a public ledger.
But here’s the trap I see many fall into: assuming that because Keling AI is centralized, all AI must be centralized. That’s survivorship bias. The $3 billion is a signal of market confidence, not technical superiority. It’s the same signal we saw in 2022 when Ethereum rollups raised billions—only to discover that most of those projects were just Ethereum rebrands for hype. 90% of so-called “Bitcoin Layer2s” are Ethereum projects rebranding for hype. The real Bitcoin community doesn’t acknowledge them. Similarly, 90% of the value in AI is being captured by incumbents, but the remaining 10%—the actual innovation in data ownership, model transparency, and decentralized governance—will grow faster than the incumbent’s valuation multiple.
The Takeaway: The Vision Forward
Keling AI’s $3 billion round is not the end of the story. It’s the beginning of a new chapter in the battle between open and closed intelligence. As someone who has spent the last year exploring AI-crypto symbiosis, I’m convinced that the next Sora-level breakthrough won’t come from a company with a $30 billion valuation. It will come from a decentralized network where the creators of data are also the owners of the model.
Decentralization is a verb, not a noun. It’s not a feature you add; it’s a principle you live. Keling AI is building a castle. We need to build the common.
The market is euphoric, but euphoria masks technical flaws. Look at the code. Look at the data pipeline. Ask who controls the key. That’s where the real value—and the real risk—lies.
Adapted from an internal note circulated on October 13, 2025, during a protocol strategy meeting in Seattle. The video generation race is just getting started.