Hold the line.
A JPMorgan report this week reaffirmed Tencent as 'Overweight' with a HKD 690 target price, citing AI investment as the catalyst for revenue conversion by 2027. The numbers are impressive: Q2 revenue up 8% YoY, net profit surging 53%, and AI capex estimated at 105 billion yuan per quarter. Free cash flow, however, turned negative 13.8 billion yuan—a sign that even the most centralized giants are bleeding capital to stay in the AI race.
But here's the truth that institutional analysts refuse to see: the same capital that powers Tencent's AI also fuels the very system crypto is designed to replace. The question isn't whether Tencent will monetize its AI. It's whether the model of centralized AI—where compute, data, and governance are controlled by a single entity—can ever be trusted with our digital sovereignty.
I've been here before. In 2017, I translated Tezos technical documents for 50,000 Chinese readers, believing that self-amending governance could democratize code. The subsequent collapse of ICO vanity projects taught me that idealism without structural integrity is just another form of hype. Today, I see the same pattern repeating in AI: centralized players pouring billions into closed models, while the crypto community builds decentralized alternatives with a fraction of the budget.
The Blind Spot in JPMorgan's Analysis
JPMorgan's thesis rests on two assumptions: first, that Tencent's AI investment will eventually convert to revenue; second, that the balance sheet can absorb the negative free cash flow until 2027. Based on my experience auditing DeFi protocols during the 2020 SPIKE incident, I've learned that balance sheets are trust proxies, not guarantees. When a centralized entity holds all the keys—both literal and metaphorical—the risk isn't just financial; it's existential.
Consider the numbers: JPMorgan estimates Tencent's adjusted free cash flow at 37.6 billion yuan, but that figure excludes the 13.8 billion yuan negative free cash flow from AI capex. In crypto terms, this is like a DeFi protocol showing a 30% yield while its liquidity pool is being drained by a single whale. The yield is real, but the system is fragile.
The Crypto AI Paradigm: Capital Efficiency Through Sovereignty
Meanwhile, decentralized AI projects are proving that you don't need billions to build the future. Bittensor, for instance, uses a subnet architecture where miners contribute compute power and are rewarded in TAO tokens. The network's market cap is a fraction of Tencent's AI capex, yet it produces intelligence that is open, auditable, and permissionless. During the 2022 bear market, I retreated to audit decentralized identity protocols like Polygon ID, and what I found was a blueprint for how AI could be governed: not by corporate boards, but by code, community, and cryptographic proofs.
The key insight is capital efficiency. Tencent spends 105 billion yuan per quarter on AI infrastructure. A decentralized network like Render can match that compute capacity by leveraging idle GPUs from thousands of individual users, paying them in tokens. The cost is not zero, but it's distributed, and the incentives are aligned: users earn tokens for contributing resources, not for selling their data.
The Trust Crisis that Centralized AI Cannot Solve
In 2020, I helped the MakerDAO community create ethical lending guides after the SPIKE incident. We learned that trust is built through radical transparency, not through opaque balance sheets. Tencent's AI models are black boxes. We don't know what data they train on, what biases they encode, or how they make decisions. JPMorgan's report doesn't ask these questions because its model assumes that centralized authority is efficient. But efficient isn't the same as trustworthy.
Crypto AI, by contrast, is built on verifiable computation. Zero-knowledge proofs allow us to verify that an AI model's output is correct without revealing the underlying data. On-chain governance ensures that changes to the model are voted on by token holders, not dictated by a CEO. This isn't just idealism; it's a technical response to the failure modes of centralized systems.
Contrarian Angle: The Pragmatist's Reality Check
Let me be clear: I'm not saying Tencent will fail. I'm saying the narrative that centralization is the only path to AI monetization is incomplete. The bear market has taught us that survival matters more than gains. Many crypto AI projects today are bleeding LPs and facing a 40% decline in liquidity. The question is whether they can survive long enough to prove their model.
Truth decays slowly. The 2027 timeline JPMorgan cites is conveniently far enough to avoid accountability. But the crypto community has a different timeline: we build for the long game, not quarterly earnings. The 2024 ETF era showed us that institutional adoption can coexist with individual sovereignty, but only if we hold the line on principles. I've seen this before—in 2017, in 2020, in 2022. Each cycle, the centralized incumbents promise efficiency, and each cycle, they fail to deliver on trust.
The AI-Crypto Convergence: A Human-Centric Future
In 2026, I co-founded the 'Human-in-the-Loop' consortium to ensure that AI agents executing smart contracts remain accountable to human values. We designed a verification layer that requires ethical sign-offs for high-value autonomous transactions. The pilot with 500 users showed that humans are willing to pay a premium for transparency. This is the future JPMorgan's models miss: the market doesn't just want returns; it wants dignity.
The algorithm is not the enemy. The unchecked power behind the algorithm is. Tencent's AI will generate revenue, but it will also generate dependence. Crypto AI generates sovereignty, but it requires patience. The next bull run will be driven by the convergence of these two forces, but only projects that survive the bear market with real utility—not just hype—will emerge.
Takeaway
JPMorgan's target price of HKD 690 is a bet on centralized efficiency. I'm betting on decentralized resilience. The numbers don't yet favor my side, but numbers can be faked. Trust cannot. In five years, when the negative free cash flow of centralized AI converges with its governance risks, the crypto AI models that survived the bear market will be the only ones left standing.
Build anyway. Hold the line.
