Last week, JPMorgan slapped a 690 HKD target on Tencent, pointing a shiny finger at the WeChat AI Agent. Their logic was clean: uncertainty is dropping, beta is rolling, and the valuation multiple expands as risk premium shrinks. I read the report three times. And I felt that familiar itch—the one I got back in 2017 when everyone was bullish on centralized ICO platforms while my CapeHorizon DAO was burning gas fees on Ethereum.
JPMorgan is right about one thing: AI agents will reshape digital economies. But they’re looking at the wrong architecture. They’re betting on a walled garden, a super-app that controls every node of the agent stack—model, data, payments, and supply chain. That’s a bet on the old playbook. I’ve lived through enough cycles to know that the real signal is in the cracks of centralized narratives.
Context: What JPMorgan Actually Described
The report paints WeChat AI Agent as a step change from “AI option with no timeline” to “phased rollout with beta testing.” They identify three core uncertainties: integration into the WeChat platform as a super entry point, the scope of transaction permissions (payments, mini-programs, service calls), and the construction of a supply system (service/content provisioning). Their valuation logic shifts from earnings-per-share to multiple expansion—meaning the stock price moves on narrative and reduced risk discount, not actual revenue.
This is textbook for early-stage technology pricing. I saw the same pattern during DeFi Summer in 2020, when I chased triple-digit APYs across three protocols simultaneously. The market priced the promise, not the execution. JPMorgan’s confidence stems from Tencent’s engineering track record and user stickiness. But that confidence is a double-edged sword.
Core: The Decentralized Agent Thesis JPMorgan Ignored
Let me break down why the centralized agent model is fragile, and how Web3 protocols are building the better alternative—right now, in this bear market.
1. Integration vs. Interoperability
WeChat’s integration is a feature, but it’s a bug in disguise. It means the agent lives inside a single app, controlled by one company. In Web3, we’re building interoperable agent frameworks—like Autonolas’ off-chain service stack or Fetch.ai’s autonomous economic agents. These agents can move across chains, access DeFi protocols, and execute multi-step operations without permission. I saw this firsthand during my DeFi liquidity trap: when you’re dependent on one platform, you’re exposed to its downtime, censorship, or governance shifts. Decentralized agents reduce that correlation risk.
2. Transaction Permissions: Trustless vs. Trustful
JPMorgan worries about how much “transaction scope” WeChat Agent will have. They’re asking: will it be allowed to spend money on my behalf? That’s a trust question. In Web3, agents can operate with smart contract-based permissions—pre-defined, auditable, revocable. No one entity can change the rules overnight. I learned this lesson during AfricanCode NFT project: when we minted art through a centralized platform, the platform could freeze our collections. With on-chain agents, ownership and control stay with the user. Code is law, but people are truth—and truth requires transparency.
3. Supply System: Open vs. Fee-extractive
WeChat’s supply system will likely be curated and taxed. Developers and merchants will pay commission or ad fees to get agent visibility. Compare that to protocols like Olas (formerly Autonolas) or Bittensor, where agents register on a decentralized registry, stake token, and compete for tasks based on reputation and performance. The fee is algorithmically determined, not extracted by a gatekeeper. In my TruthChain project, we built an on-chain provenance system for AI-generated content. The key insight: when the protocol is open, anyone can contribute services, and the value flows back to the network, not a single company.
Contrarian: Why the Bull Case on WeChat Agent Is Overblown
JPMorgan’s report signals “an increase in certainty,” but data tells a different story. Look at the decentralized agent space. The total value locked in agent-focused protocols grew 340% in Q3 2024, even as the broader market contracted. Meanwhile, centralized AI apps face rising regulatory scrutiny—China’s new AI rules require content filtering, data localization, and political alignment checks. WeChat Agent will be compliant, but that comes at the cost of openness and user sovereignty.
Furthermore, the report avoids a critical question: what happens when the underlying model (Tencent Hunyuan) lags behind open-source alternatives? Llama 3.1, DeepSeek, and Mistral are closing the gap. If WeChat’s agent intelligence plateaus, users will seek alternatives. And those alternatives will be decentralized, portable, and composable.
I’ve seen this pattern before. In 2022, during the bear market, my portfolio dropped 70%. But I dug into ZK-rollup research and realized that privacy-preserving computation was the next frontier. The same thing is happening now with agents: the signal is in the obscure protocols that prioritize user control over convenience. JPMorgan sees a beta test; I see a temporary window before decentralized agents eat the lunch of the super-app.
Takeaway: The Real Investment Is in Open Agent Protocols
The JPMorgan report is a useful piece of narrative analysis, but it’s not a guide to where value will accrue. The 690 HKD target is a bet on centralized execution. The contrarian bet—and the one aligned with Web3 values—is on protocols that enable permissionless agent creation, trustless transactions, and open supply systems.
As I tell my community in Cape Town: build in public, live in truth. The AI agent revolution will not be orchestrated by a single company. It will be a stochastic, composable, multi-chain network of agents serving humans directly. We’re early, but the code is already being written. The question is whether you’re betting on the walled garden or the open plains.
Embrace the volatility, find the signal. The signal is decentralized.
