开发者数量腰斩之后：Crypto没有死，只是把人才让给了AI
Original authors: Xinyang, Ethan, IOSG Ventures
In 2026, the GitHub activity curve of the Crypto open-source community completed an astonishing "bottoming out." After declining from a peak of 45K monthly active developers in 2022 to approximately 23K, this apparent halving in on-paper data sparked discussions about "narrative exhaustion" on social media. However, when we deconstruct the segments of this curve, what we see is not an industry contraction, but a profound "deleveraging of talent."
▲ Data Source: Electric Capital Developer Report, based on Crypto Ecosystems Github
Who Left? Who Stayed?
Those who left were primarily newcomers. In February 2024, new developers added in a single month reached 5,462, followed by a sharp decline, with a churn rate of 52% for those with less than one year in the industry. Most of these individuals flooded in during the bull market, working on NFT minting contracts, forking DeFi protocols, or building frontends for new L2s.
These roles are highly dependent on market hype. Once the hype fades, projects cease operations, and the positions disappear along with them. Data shows that newcomers' code contributions never exceeded 25% of the total; from the beginning, this group was never at the core of the industry.
▲ Newcomers surge in with the bull market and exit during the bear; Established devs (with 2+ years of experience) hit an all-time high in the same period.
Data Source: Electric Capital Developer Report
On the other side, developers with over two years of experience not only held steady but increased during the same period, reaching an all-time high and contributing approximately 70% of the code. Maria Shen, GP at Electric Capital, was direct in her assessment: "When we look at the established developers group, it is growing and appears very healthy."
They stay not because they have no other options.
Technologically, the core work in crypto today typically involves infrastructure development that requires years of accumulated understanding: protocol layer development, security auditing, cross-chain architecture. These tasks require deep, long-term experience to master and are not the kind that market corrections can easily eliminate.
Economically, many veterans hold tokens that have yet to vest, governance powers within protocols, and equity relationships. Their accumulation within this industry has formed real barriers to entry and substantial returns.
From an ecosystem distribution perspective, they are voting with their feet: Bitcoin developers grew by 64.3% over two years, Solana by +11.1%, while Cosmos decreased by 51.1% and Polkadot by 46.9%. Veterans are converging towards ecosystems with real users and revenue, leaving behind projects sustained merely by narratives.
▲ Source: Coincub Web3 Jobs Report 2025
Data Source: Web3.Career
The changing composition of job roles confirms the same trend. In 2025, the highest share of new Web3 positions added was not for developers, but for Project & Programme Management, exceeding 27%.
This is counterintuitive for an industry renowned for being technology-driven, but the underlying logic is not complex: the industry is transitioning from a construction phase to an execution phase. Over 100 chains need integration; institutional clients entering the space impose entirely different requirements for compliance and security; DAO governance must find balance among stakeholders with diverse interests.
This is not traditional project management; it is about coordination and judgment within an environment where rules are still being formed.
While the industry appears to be shrinking on the surface, its core density is actually increasing. The bear market of 2018-2019 also saw significant developer attrition, but it was followed by the emergence of phenomenal projects like Uniswap, Aave, and OpenSea, which defined the 2020-2021 bull run. The builders who remain this cycle have access to more mature infrastructure, and the AI era provides them with a stage larger than the last one.
What Capabilities Do Those Who Stay Possess?
What specific capabilities has the crypto industry actually honed in its builders? To answer this, we must return to the foundational principles of blockchain. Amidst the cycles of bull and bear markets, this industry has always operated on the same underlying rule: Code is Law, Execution is Finality.
In the 2016 DAO incident, an attacker exploited a recursive call vulnerability to siphon off $36 million. The code had no bug; the logic executed exactly as programmed, but the edge case was not anticipated by the designers. In 2021, the Poly Network cross-chain bridge was attacked, with $610 million transferred away within hours.
No platform could halt it, no institution could reverse it, no legal clause could recover it. This is a structural characteristic of crypto that distinguishes it from almost every other industry: zero room for error, and virtually no post-hoc intervention.
This environment has forced the development of a capability rarely needed in other industries: the ability to build a functioning system from scratch, under conditions of missing rules and absent trust, that convinces strangers to participate.
This capability comprises two levels. First, establishing trust from zero, independent of any external authority, relying solely on code and mechanisms to make strangers willing to place real assets within it. Second, making judgments under dual technological and economic uncertainty, designing systems that can operate without a regulatory framework, historical data, or industry standards to reference.
Both levels have specific verification within crypto. Uniswap, with no company guarantee, no KYC, no customer service, achieved hundreds of billions of dollars in daily trading volume based solely on trust in a few hundred lines of code and a set of economic mechanisms.
MakerDAO, without central bank backing or deposit insurance, maintains the stability of DAI purely through on-chain governance and collateral mechanisms.
The DeFi Summer was even more extreme. Without regulatory frameworks, audit standards, or historical data to reference, builders designed AMMs, lending protocols, and liquidity mining, going from concept to billions of dollars in TVL in just a few months. While this capability manifests differently in builders at the protocol, application, and governance layers, the underlying principle remains the same.
The AI era is creating a structurally similar problem. Model decision-making processes are opaque, and output results cannot be independently verified. AI agents are beginning to autonomously execute transactions and allocate capital, yet the accompanying rule systems and constraint mechanisms do not yet exist.
Large model companies control both the models and the evaluation criteria, leaving users without effective means of verification. Computing power is highly concentrated among a few top-tier companies, creating monopolistic pricing when demand spikes. These issues all point to the same core problem: the trust issue in autonomous systems is repeating itself on the larger scale of AI.
Crypto builders have been dealing with this type of problem for years within environments lacking external authority rules. Previously, the context was on-chain protocols; now, it has shifted to AI. And a group of people has already directly brought the capabilities accumulated in crypto into AI, yielding results.
How Are These Capabilities Being Repriced in the AI Era?
Cases of transitioning from crypto to AI are not uncommon in recent years, but upon closer inspection, what they bring with them differs.
The most straightforward path is the direct transfer of hardware and experience. The three co-founders of CoreWeave, Michael Intrator, Brian Venturo, and Brannin McBee, started using GPUs to mine Ethereum in 2017, scaling from one machine to thousands. They shut down their mining operation in 2022. Two months later, ChatGPT was launched, and their GPUs directly became supply for AI computing. In March 2025, they went public on Nasdaq with an IPO valuation of approximately $23 billion, with its market cap peak later reaching nearly $70 billion.
OpenSea co-founder Alex Atallah dealt with the aggregation and routing of highly heterogeneous assets in the NFT marketplace. He transferred the exact same expertise to AI model routing by founding OpenRouter, serving over 5 million developers within two years and reaching a valuation of $500 million.
Another type of migration is more noteworthy. NEAR founder Illia Polosukhin is a co-author of the Transformer paper. After leaving Google, his initial idea was to build AI applications using natural language. However, during development, they encountered a real-world problem: needing to make cross-border payments to data labeling workers worldwide, many of whom lacked bank accounts. Blockchain technology became the optimal solution to this payment challenge.
Now, NEAR is pivoting into an AI infrastructure platform, with core directions being user-owned AI and Decentralized Confidential Machine Learning (DCML), allowing users to use AI services without exposing their data. The experience in decentralized architecture accumulated at NEAR has become the most difficult-to-replicate starting point in this direction.
Circle co-founder Sean Neville left to found Catena Labs, positioning it as an AI-native bank, directly transferring his understanding of stablecoin infrastructure to the AI agent financial scenario. a16z crypto led an $18 million seed round.
Nader Dabit, a senior developer for Aave and Lens Protocol, moved to Cognition, bringing his experience in building developer ecosystems across multiple crypto protocols into the AI agent tools space.
What this group takes with them is not just GPU hardware or user networks, but an intuition for mechanism design, experience in building developer ecosystems, and the judgment to construct trustworthy systems from scratch when rules are absent. These capabilities precisely correspond to three structural gaps encountered in AI scaling.
Aggregation and Optimization of Computing Power
Computing power is the most direct bottleneck for AI scaling. Training and inference require vast amounts of GPUs, demand is volatile, cloud providers are expensive and have queues, and enterprises don't want to stockpile hardware themselves. This problem has two facets: how to aggregate and allocate computing power, and how to use the aggregated power more efficiently. Crypto builders have direct, transferable experience in both areas.
Hyperbolic solves the distribution and trust problem. Founder Jasper Zhang brought decentralized mechanism design into the AI computing track: tokens incentivize dispersed GPU holders to contribute idle computing power, but the more core issue is trust.
Why should you trust that the computation result from an unknown node is correct? Its core innovation, PoSP (Probabilistic State Proofs), uses random sampling combined with game theory to make honesty the dominant strategy for nodes, requiring no full verification, being low overhead, scalable, and producing reliable results. This mechanism is directly migrated from crypto's logic of verifying the behavior of unknown nodes.
MoonMath addresses the efficiency problem. Its predecessor, Ingonyama, focused on ZK hardware acceleration, multiplying the speed of ZK proof generation under extreme computational constraints.
Now, its direction has shifted to a performance layer for Physical AI, working on sparse attention acceleration for video diffusion models (LiteAttention), low-rank decomposition of FFN layers (LiteLinear), and training backpropagation acceleration (BackLite). From ZK acceleration to AI inference acceleration, the underlying capability is the same: making math run faster under extreme computational constraints. The track changed, but the accumulated expertise was not wasted.
AI Governance and Incentive Mechanism Design
When multiple AI agents begin to collaborate on tasks, how do you ensure they don't undermine the overall system while pursuing their individual goals? Each participant is optimizing its own objective function, and there's no guarantee that the sum of their actions keeps the system functioning properly. Moreover, the execution speed of agents far exceeds the window for human intervention.
This is precisely the type of problem crypto builders have tackled repeatedly in DAO governance and tokenomics design: getting participants with completely different interests to operate according to the system's intended direction without a central authority. Crypto's answer is economic mechanisms. Violations carry real economic costs; the rules are written into code and executed automatically.
EigenLayer has directly migrated this mechanism to the AI context. Through its restaking mechanism, nodes must pledge assets before participating in collaboration. Failure to perform or rule violations trigger automated penalties. The rules are not suggestions; they are hard boundaries with real economic consequences.
EigenCloud extends this logic to verifiable computation and collaborative governance for AI agents, constraining agents to operate within preset bounds while pursuing their goals. Using economic mechanisms to constrain agents is far more reliable than using ethical guidelines.
Autonomous Payments for AI Agents
There is an even more fundamental problem: how do agents pay? Traditional payment systems are designed for humans. Credit cards require opening an account, bank transfers require authorization, and every step assumes the operator is human, has an identity, and will wait. Agents don't wait. An agent might initiate a large number of requests per second, each potentially involving micropayments. Traditional payment pipelines fail directly in this scenario.
Stablecoins and on-chain rules are the infrastructure already built by crypto builders. They natively support programmability, permissionlessness, and 24/7 operation. These three characteristics are precisely the hard requirements for agent payment scenarios. All that's missing is a protocol layer connecting stablecoins to agent workflows.
x402, launched by Coinbase in May 2025, activates the HTTP 402 status code, embedding stablecoin payments directly into HTTP requests. An agent completes a payment simultaneously with initiating a request. No account needed, settlement takes about two seconds.
As of April 2026, the x402 protocol has processed over 165 million transactions, with a cumulative volume of approximately $50 million and 69,000 active agents (source: x402 Foundation). Cloudflare, AWS, Stripe, and Anthropic MCP have all integrated. Agent payments are already a track with real traffic.
These three directions correspond to three structural gaps encountered by AI scaling: the aggregation and efficiency of computing power, incentive alignment for multi-agent collaboration, and infrastructure for autonomous payments. These problems have no ready-made answers in traditional software architecture, but the crypto industry has corresponding processing experience for each. Capabilities haven't disappeared; they've simply found new application scenarios.
The Builder's New Positioning: From Writing Contracts to Setting Rules for AI
The scaling of AI is creating a functional gap that didn't exist before. It's not a gap in technical talent, but a gap in people who can design trust mechanisms within autonomous systems. As the service target shifts from humans to AI, the role of the crypto builder is also being redefined.
The following table compares the dimensional changes in specific functional paradigms:
The core difference between the two paradigms lies not in the tech stack, but in the method of establishing trust and the logic of rule enforcement. In the Pre-AI era, crypto builders faced human participants. Rules were written into contracts, the margin for error was zero, but the system's boundaries were relatively clear.
In the AI-Native era, when the interacting entities become autonomously operating AI agents, the problem to be solved is: agent behavior is unpredictable, execution speed far exceeds the human intervention window, and the very boundaries of the system need redefinition under greater uncertainty.
The functional positioning of the crypto builder is shifting from "writing secure contracts" to "designing trustworthy mechanisms for autonomous AI systems."
Recruitment by leading institutions is already reflecting this shift:
▲ Active AI/Data Core Positions Actively Opened by Top Trading Platforms in Q1 2026
Source: Gate Research Institute
Hiring trends at top trading platforms and institutions in 2026 clearly reflect this trend: they are no longer simply hiring AI engineers or crypto developers, but seeking individuals who can bridge the two – those who understand on-chain incentive distortions and governance games, can deeply embed AI tools into crypto workflows, and design mechanisms that keep agents aligned with regulators and users over the long term.
Capital allocation directions have already reflected this judgment. Paradigm is raising a new fund of up to $1.5 billion, expanding its investment scope from crypto to AI and robotics.
Haun Ventures completed its $1 billion Fund II, focusing on the intersection of crypto and AI within financial infrastructure, particularly payment, stablecoin, and agent-to-agent economic systems that support autonomous AI agent transactions and coordination.
a16z crypto completed its $2.2 billion fifth fund (Crypto Fund V), explicitly stating that the fund will invest 100% in the crypto space. In response to the complexity and opacity of the AI era, they will focus on applications leveraging crypto's transparency, verifiability, and decentralization.
Furthermore, according to PitchBook data, in 2025, approximately 40% of VC investment in the US crypto space flowed into companies also involved in AI operations, a significant increase from 2024.
When it comes to crypto builders transitioning to AI, the chosen paths show clear differences across market environments.
In the US, as the regulatory environment becomes relatively clearer, protocol-layer innovation has gained real room to survive. The capital network density is high, the path from idea to financing is short, and there is a larger margin for error.
<
