The Hook
Over the past week, a Chinese-language article titled “Liang Wenfeng Has No Life, Yang Zhiping Has No Way Back” circulated widely in AI circles, scoring millions of views and elevating two founders to near-mythical status. Liang, the quiet force behind DeepSeek’s cost-efficient open-source models, was portrayed as a monk who traded all personal time for technical perfection; Yang, the charismatic CEO of Moonshot AI (Kimi), was framed as a gambler with no fallback, betting the company on a single product strategy.
But in blockchain, we have seen this movie before. The same narrative architecture—extreme founder sacrifice, no retreat, all-in commitment—was deployed by Do Kwon before Terra’s collapse, by Su Zhu before Three Arrows Capital’s liquidation, and by countless DeFi “founders” who disappeared after their contracts were drained. The difference is that blockchain leaves a permanent, transparent record. Code does not lie, only the architecture of intent.
I spent 29 years in financial engineering and seven years auditing Layer2 protocols. Based on my experience reverse-engineering the PlexCoin ICO in 2017 and modeling the LUNA death spiral in 2022, I can tell you that a founder’s personal sacrifice is a poor proxy for protocol security. In fact, the “no life, no way back” narrative often obscures the very risks that should concern every LP and token holder.
This article applies the same decompilation methodology to the AI founder narrative, using blockchain-specific stress tests: on-chain data, gas cost analysis, and liquidity depth charts. The goal is not to critique AI founders—they are building remarkable technology—but to warn the crypto community against importing a narrative that has already proven toxic in our own ecosystem.
Context: The AI Narrative and Its Crypto Parallels
The original analysis of the Liang/Yang piece deconstructed seven dimensions: technical route, commercialization, industry impact, competitive landscape, ethics, investment/valuation, and infrastructure. Each dimension was rated with low to medium confidence because the article itself offered no verifiable data—only narrative framing.
That is the red flag. In crypto, we demand empirical evidence: open-sourced code, audited smart contracts, measurable TPS, MEV extraction rates, and liquidation thresholds. When a founder story substitutes for these metrics, the project is already in danger.
Consider the parallels:
- Liang Wenfeng “has no life” → analogous to an anonymous or reclusive founder (e.g., Satoshi, but more relevant: the founders of Tornado Cash, who prioritized privacy over personal visibility). The risk is that a team with no visible life is harder to hold accountable. If the protocol fails, who do you sue? Who releases the upgrade? The mystique can hide centralization.
- Yang Zhiping “has no way back” → analogous to a founder who has bet everything on a single chain, a single bridge, or a single liquidity pool. The LUNA collapse was driven by a “no way back” commitment to algorithmic stability. Do Kwon famously said, “I don’t have a plan B.” The market punished that arrogance.
Core: Applying Technical Due Diligence to the Founder Narrative
Let me take you through my real audit workflow, applied hypothetically to the AI companies—but with the same rigor I use for Layer2 rollups.
1. The “No Life” Claim: Gas Cost Analysis
When I audit a protocol, the first thing I check is the gas consumption pattern. High gas costs often point to inefficient architecture. Similarly, when a founder claims to have “no life,” I want to see whether that sacrifice maps to technical efficiency.
For DeepSeek, the claim of low API cost (1/100th of GPT-4) is a strong positive signal. But does it come from genuine architectural innovation (MoE, sparse activations) or from subsidized compute? During the 2022 bear market, I wrote a minimalist report on L2 scaling solutions and noted that “cheap” often means “centralized” until proven otherwise. The same applies here: an open-source model that is too cheap may hide data-quality shortcuts or security risks.
2. The “No Way Back” Claim: Liquidity Depth and Leverage
During the Compound Finance incident in 2020, I identified a liquidation cascade vulnerability in its interest rate model. The leverage was too concentrated. When Yang Zhiping says there is “no way back,” I interpret that as a leverage-on-leverage bet on a single product—Kimi’s long-context capability. In blockchain terms, this is equivalent to a DEX that relies on a single liquidity pool for 80% of its volume. When the competition catches up (as Alibaba and Baidu have with their own long-context models), the “no way back” founder is left holding the bag.
I modeled this scenario in 2024 while analyzing Optimism’s OP Stack throughput: a bottleneck in state commitment processing could be exploited under high congestion. The same dynamic applies to AI: if a single product feature (long context) becomes commoditized, the founder’s lack of a fallback becomes a systemic risk.
3. The Industry Impact: Ethereum vs. Solana, DeepSeek vs. Moonshot
The original analysis correctly noted that the narrative creates a “martyr” and “gambler” dichotomy. In crypto, we see this as Ethereum (the methodical, consensus-driven “monk”) vs. Solana (the high-speed, single-minded “gambler”). Both have succeeded, but the narrative often distorts risk perception.
Ethereum’s slow, deliberate upgrade path is closer to Liang’s “no life” patience. Solana’s aggressive “just ship” attitude resembles Yang’s “no way back” urgency. Yet after the FTX collapse, Solana’s narrative nearly killed it—despite the protocol being technically sound. The same could happen to either AI company if a narrative shift occurs.
Contrarian: When the Narrative Hides the Real Blind Spots
The most dangerous assumption in both AI and crypto is that a founder’s personal sacrifice compensates for architectural weaknesses. Let me give you a concrete example from my 2017 ICO audit.
PlexCoin promised 10% daily returns. The founder was in the news constantly, claiming he had “no life” outside the project. I reverse-engineered the Solidity codebase and found a logical fallacy in the compound interest algorithm within hours. The narrative was irresistible; the code was fraudulent.
Today, the same pattern repeats. A founder who tweets about sleepless nights while ignoring basic reentrancy guards is a liability. Truth is found in the gas, not the press release.
Blind spot 1: The “No Life” founder may be a single point of failure. In blockchain, we mitigate this through formal verification, multi-sig, and DAO governance. In AI, a single brilliant mind who works 20-hour days creates a knowledge silo. When that person burns out (and they will), the project stalls. I have seen this happen with at least three DeFi protocols that relied on a “genius” developer with no documentation.
Blind spot 2: The “No Way Back” bet is rarely symmetric. The founder gains fame and funding; the investors gain exposure; the users gain a product. But if the bet fails, the founder often walks away with a parachute (acqui-hire, advisory role). The investors with “no way back” are the LPs who bought the token at the top. In blockchain, we have learned to check the vesting schedules and the team’s lockup periods. The comparable for AI is the founder’s equity dilution and liquidation preferences. The original analysis did not provide this data, which lowers its confidence to E-level.
Blind spot 3: Narratives are optimized for fundraising, not for security. The best protocols I have audited (e.g., Optimism’s bedrock upgrade, Uniswap v4) have founders who are boring. They write clean code, test extensively, and rarely give interviews about their “sacrifice.” The worst protocols have founders who are always on stage, updating their LinkedIn profile to read “Building the Future.”
Takeaway: Architecture Outlasts Algorithms—and Narratives
The AI founder story is compelling. It sells clicks and attracts capital. But for the blockchain community, the lesson is clear: narratives are datasets we have already optimized. We learned in 2022 that no amount of founder charisma can sustain a protocol with bad tokenomics. We learned in 2024 that a “no life” developer who ignores formal verification will produce a contract that drains under high gas.
As Layer2 Research Lead, I now prescribe a simple rubric: before you invest time or capital in any project—AI, crypto, or hybrid—run the narrative through a five-point stress test.
- Dimensional analysis: Break the founder story into the seven dimensions (tech, commercial, competition, etc.) and demand evidence for each.
- Code-first requirement: If the project has no verifiable code or data source, treat it as a meme token.
- Liquidity depth check: What happens if the market turns? Can the project survive a 90% drawdown?
- Contrarian audit: Look for at least three blind spots that the narrative is hiding.
- Forecast commitment: Ask the founder, “If your bet fails, what is your plan?” If the answer is “there is no plan,” walk away.
Simplicity is the final form of security. The most robust Layer2 designs are those with minimal governance overhead, transparent sequencers, and escape hatches. The most robust AI companies will be those that treat their founder narrative as a bug, not a feature.
I will continue to follow DeepSeek and Moonshot AI because their technology is genuinely impressive. But I will not evaluate them through the lens of “no life” or “no way back.” In the words of my 2022 report on LUNA’s death spiral: Hedging is not fear; it is mathematical discipline.
The blockchain industry has already paid the tuition for this lesson. Let us not repeat it on behalf of AI.
