Authors: Xinyang & Ethan, IOSG
In 2026, the GitHub activity curve for the Crypto open-source community completed a remarkable "bottoming out." Monthly active developers fell from a peak of 45K during the 2022 bull run to around 23K, triggering discussions on social media about "narrative exhaustion." However, dissecting the cross-section of this curve reveals not an industry in decline, but a profound "talent deleveraging."
▲ Source: Electric Capital Developer Report, based on Crypto Ecosystems GitHub
I. Who Left? Who Stayed?
It was mainly newcomers who left. Monthly new developer additions peaked at 5,462 in February 2024, then fell sharply, with a 52% attrition rate for those in the industry less than a year. Most of them flooded in during the bull market, working on NFT minting contracts, forking DeFi protocols, or building front-ends for new L2s. These roles were highly dependent on market hype. Once the hype faded and projects ceased operations, the jobs vanished. Data shows that newcomers never contributed more than 25% of the overall code. This group was never in the industry's core circle to begin with.
▲ Newcomers surged in with the bull market and left with the bear; Established devs (2+ years experience) hit all-time highs during the same period
Source: Electric Capital Developer Report
On the other side, developers with over two years of experience in the industry did not decline but actually increased during the same period, reaching an all-time high and contributing about 70% of the code. Electric Capital GP Maria Shen's assessment is direct: "When we look at the established developers cohort, it's growing, and it looks very healthy."
They didn't stay because they had no other options.
Technologically, the core work in crypto today is infrastructure development—protocol layer work, security audits, cross-chain architecture—which generally requires years of accumulated knowledge to truly grasp and cannot be easily eliminated when market hype fades.
Economically, many veterans have unvested tokens, governance power within protocols, and vested interests; their accumulated contributions in this industry have formed real barriers and rewards. Looking at ecosystem distribution, they are voting with their feet: Bitcoin developer count grew 64.3% in two years, Solana +11.1%, while Cosmos fell 51.1%, Polkadot fell 46.9%. Veterans are concentrating in ecosystems with real users and revenue, leaving behind those still reliant on narratives.
▲ Source: Coincub Web3 Jobs Report 2025
Data source: Web3.Career
Changes in job structure corroborate the same trend. Among new Web3 positions in 2025, the highest share wasn't developers, but Project & Programme Management, exceeding 27%. For an industry renowned for being technology-driven, this is counterintuitive, but the underlying logic is straightforward: the industry is transitioning from a building phase to an execution phase. Over 100 chains need integration, institutional clients bring completely different demands for compliance and security, and DAO governance requires finding balance among stakeholders with diverse interests. This isn't traditional project management; it's coordination and judgment in an environment where the rules are still being formed.
While the industry appears to be shrinking on the surface, its core density is actually increasing. The 2018-2019 bear market also saw significant developer attrition, yet it was followed by the emergence of paradigm-defining projects like Uniswap, Aave, and OpenSea, which defined the 2020-2021 bull market. The builders who stayed this round have access to more mature infrastructure, and the AI era offers them an even larger stage than the previous cycle.
II. What Skills Did Those Who Stayed Bring?
What unique skills has the Crypto industry actually cultivated in builders? To answer this, we must return to blockchain's foundational principles. Across bull and bear cycles, this industry has always operated on the same core rule: code is law, and execution is finality.
The 2016 DAO hack saw an attacker steal $36 million via a recursive call vulnerability. The code had no bug; it executed exactly as designed, but the edge case wasn't anticipated by the designers. The 2021 Poly Network cross-chain bridge hack saw $610 million moved in hours. No platform could halt it, no institution could reverse it, no legal clause could seek recourse. This is the structural feature distinguishing crypto from almost every other industry: zero margin for error, almost no possibility of post-facto intervention.
This environment has forced the development of a set of skills rarely needed in other industries: Building a functional system from scratch, under conditions of absent rules and absent trust, that strangers are willing to participate in.
This capability has two layers. First, establishing trust from zero, relying on no external authority, only code and mechanisms to make strangers willing to deposit real assets. Second, making judgments under dual technological and economic uncertainty, designing functional systems without regulatory frameworks, historical data, or industry standards to reference.
Both layers have concrete validation within crypto. Uniswap has no corporate guarantee, no KYC, no customer support. Anyone depositing funds into a liquidity pool relies solely on trust in a few hundred lines of code and a set of economic mechanisms, achieving billions in daily volume. MakerDAO has no central bank backing, no deposit insurance, maintaining DAI's stability purely through on-chain governance and collateral mechanisms. During DeFi Summer it was more extreme: no regulatory framework, no audit standards, no historical data to reference. Builders designed AMMs, lending protocols, liquidity mining, going from concept to tens of billions in TVL in months. This capability manifests differently in builders at the protocol, application, and governance layers, but the underlying principle is the same.
The AI era is creating a structurally similar problem set. Model decision-making is opaque, outputs cannot be independently verified. AI agents are beginning to autonomously execute trades and allocate capital, while accompanying rule systems and constraint mechanisms don't exist yet. Large model companies control both models and evaluation standards, leaving users without effective verification methods. Compute is highly concentrated in a few top-tier firms, creating monopoly pricing during demand spikes. These issues point to the same core: the trust problem of autonomous systems is replaying at a larger scale with AI.
Crypto builders have been dealing with this class of problems in environments lacking external authoritative rules for years—the scenarios were just on-chain protocols, now it's AI. And a group has already taken the skills accumulated in crypto directly into AI, with results.
III. How Are These Skills Being Repriced in the AI Era?
Cases of transitioning from crypto to AI are increasingly common, but upon examination, what they bring differs.
The most direct path is direct hardware and experience transfer. CoreWeave's three founders, Michael Intrator, Brian Venturo, and Brannin McBee, started mining Ethereum with GPUs in 2017, scaling from one machine to thousands. They shut down mining operations in 2022. Two months later, ChatGPT launched, and their GPU inventory directly became AI compute supply. They IPO'd on Nasdaq in March 2025 at a ~$23B valuation, with a peak market cap later approaching $70B.
OpenSea co-founder Alex Atallah dealt with the aggregation and routing of highly heterogeneous assets in the NFT marketplace, applying the same experience to AI model routing by founding OpenRouter, which served over 5 million developers within two years, reaching a $500M valuation.
Another type of migration is more noteworthy. NEAR founder Ilia Polosukhin is a co-author of the Transformer paper. After leaving Google, his initial aim was to build AI applications using natural language, but a practical problem emerged: making cross-border payments to data labelers worldwide, many of whom lacked bank accounts. Blockchain technology became the best solution for this payment challenge.
NEAR is now pivoting to become an AI infrastructure platform, focusing on user-owned AI and decentralized confidential machine learning (DCML), allowing users to access AI services without exposing data. Their accumulated experience in decentralized architecture 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 AI agent financial scenarios, with a16z crypto leading an $18M seed round. Aave and Lens Protocol veteran developer Nader Dabit moved to Cognition, bringing experience in developer ecosystem building from multiple crypto protocols to the AI agent tooling space.
This group took more than just GPU hardware or user networks; they took an intuition for mechanism design, experience in building developer ecosystems, and the judgment to construct trusted systems from scratch when rules are absent. These abilities precisely correspond to three structural gaps emerging in AI's scaling.
Compute Aggregation and Optimization
Compute is the most direct bottleneck in AI scaling. Training and inference require massive GPUs, demand fluctuates wildly, cloud vendors are expensive with waitlists, and companies don't want to stockpile hardware themselves. This problem has two layers: how to aggregate and allocate compute, and how to use aggregated compute more efficiently. Crypto builders have directly transferable experience in both layers.
Hyperbolic addresses the distribution and trust problem. Founder Jasper Zhang brought decentralized mechanism design into the AI compute space: tokens incentivize distributed GPU owners to contribute idle compute, but the core issue is trust.
How can you trust a computation result from a stranger's node? The core innovation, Proof of Sampling and Proof (PoSP), uses random sampling and game theory to make honesty the dominant strategy for nodes, requiring no full verification, low overhead, scalable, and reliable results. This mechanism is directly transferred from the logic crypto uses to verify behavior of unknown nodes.
MoonMath addresses the efficiency problem. Its predecessor, Ingonyama, focused on ZK hardware acceleration, speeding up ZK proof generation severalfold under extreme computational constraints. Now pivoting to a Physical AI performance layer, it works on sparse attention acceleration for video diffusion models (LiteAttention), low-rank decomposition of FFN layers (LiteLinear), and training backward pass acceleration (BackLite). From ZK acceleration to AI inference acceleration, the underlying capability is the same: making math run faster under extreme computational constraints. The field changed, but the accumulated knowledge wasn't wasted.
AI Governance and Incentive Mechanism Design
When multiple AI agents collaborate on tasks, how to ensure they don't disrupt the overall system while pursuing their own objectives? Each participant pursues its own objective function, with no guarantee the aggregated system will function correctly, and agents operate at speeds far exceeding human intervention windows.
This is a problem type crypto builders have repeatedly tackled in DAO governance and tokenomics design: making participants with completely different interests operate according to the system's preset direction without a central authority. Crypto's answer is economic mechanisms: rule-breaking incurs real economic penalties, rules are encoded, and execution is automatic.
EigenLayer directly migrated this mechanism to the AI context. Through its restaking mechanism, nodes must stake assets before participating in collaboration. Failure to fulfill commitments or rule-breaking triggers automatic slashing. Rules aren't suggestions; they are rigid boundaries with real economic consequences. EigenCloud extends this logic to verifiable computation and collaborative governance for AI agents, ensuring agents pursuing their goals must stay within preset boundaries. Constraining agents with economic mechanisms is far more reliable than with ethical guidelines.
AI Agent Autonomous Payments
An even more fundamental problem: how do agents pay? Traditional payment systems are designed for humans: credit cards require accounts, bank transfers require authorization, each step assuming the operator is human, has identity, and will wait. Agents don't wait. They might initiate massive requests per second, each potentially involving micro-payments. Traditional payment pipes fail outright in this scenario.
Stablecoins and on-chain rules are infrastructure crypto builders have already built, natively supporting programmability, permissionlessness, and 24/7 operation. These three characteristics are exactly the hard requirements for agent payment scenarios. All that's missing is a layer of protocol 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 payment simultaneously with initiating a request, no account needed, settlement in ~2 seconds. As of April 2026, the x402 protocol has processed over 165 million transactions, cumulative volume ~$50M, with 69,000 live agents (Source: x402 Foundation). Cloudflare, AWS, Stripe, Anthropic MCP have integrated. Agent payments are already a space with real traffic.
These three directions correspond to three structural gaps in AI scaling: compute aggregation and efficiency, incentive alignment for multi-agent collaboration, and autonomous payment infrastructure. These problems have no ready-made answers in traditional software architecture, but the crypto industry has corresponding experience handling them. The capabilities haven't disappeared; they've just found a new context.
IV. Builder's New Role: From Contract Writers to Rule-Setters for AI
AI's scaling is creating a functional gap that previously didn't exist. It's not a gap for technical talent, but for people who can design trust mechanisms within autonomous systems. As the service target shifts from humans to AI agents, the role of crypto builders is being redefined.
The following table compares the dimensional shifts in specific functional paradigms:
The core difference between the two paradigms isn't the tech stack, but how trust is established and how rules are executed. In the pre-AI era, crypto builders faced human participants, rules were encoded in contracts, margin for error was zero, but the system's boundaries were relatively clear.
In the AI-Native era, when the interacting parties become autonomously running AI agents, the problems to solve become: agent behavior is unpredictable, execution speed far exceeds human intervention windows, and the system's very boundaries need redefinition under greater uncertainty. The functional positioning of crypto builders is shifting from "writing secure contracts" to "designing trustworthy mechanisms for AI autonomous systems."
Hiring at top institutions already reflects this shift:
▲ Actively opened AI/Data core positions at top exchanges in Q1 2026
Source: Gate Research Institute
2026 hiring at top exchanges and institutions clearly reflects this trend: No longer hiring purely AI engineers or crypto developers, but seeking people 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 aligning agents with regulation and users long-term.
Capital allocation already reflects this judgment. Paradigm is raising a new fund targeting up to $1.5 billion, expanding its investment scope from crypto to AI and robotics. Haun Ventures closed a $1 billion Fund II, focusing on financial infrastructure at the intersection of crypto and AI, especially payments, stablecoins, and agent-to-agent economic systems supporting AI agent autonomous transactions and coordination.
a16z crypto closed a $2.2 billion fifth fund (Crypto Fund V), explicitly stating 100% will be invested in crypto. Facing the complexity and opacity of the AI era, they will focus on applying crypto's transparency, verifiability, and decentralization. According to PitchBook data, in 2025 US crypto VC investments, roughly 40% of capital flowed to companies also involved in AI, a significant increase from 2024.
Similarly crypto builders pivoting to AI, the chosen paths under different market environments show clear differences.
In the US, following a relatively clarified regulatory environment, protocol-layer innovation gained real breathing room. High capital network density, short paths from idea to funding, greater room for error. Projects like Hyperbolic, EigenCloud, Gensyn, Ritual share a common trait: designing new mechanisms from zero, not simple application integration on existing systems. Top-tier VCs have clear theses on "verifiable compute, agent coordination, decentralized ML," willing to provide ample room for early-stage technological exploration.
The situation in Asia differs. Singapore and Hong Kong serve more as compliance landing hubs and institutional capital conduits, with relatively conservative regulatory frameworks and lower tolerance for pure protocol-layer innovation. Crypto-background builders pivoting to AI more often choose application-layer and industry integration paths—leveraging crypto-accumulated user bases, payment capabilities, or data assets to quickly integrate into AI products and services.
This isn't a capability gap, but a path divergence due to differing market signals and regulatory environments: The US encourages foundational mechanism innovation and early-stage tech exploration, while Asia emphasizes compliance-friendliness, quick monetization, and deep integration with traditional industries.
Returning to that opening GitHub curve. Monthly active developers falling from 45K to 23K superficially suggests industry contraction. But among those who stayed, the established dev share hit a record high, flocking to ecosystems with real users, while being repriced by the AI industry in unprecedented ways.
When AI scaling encounters structural bottlenecks like compute aggregation, agent autonomous payments, data & decision verifiability, and privacy coordination, these builders, at the intersection of Crypto and AI, are seeing their long-honed sensitivity to rules, incentives, and authenticity gradually transform into system-level capabilities scarce in the AI era.
As an investment firm focused on crypto infrastructure since 2017, IOSG's perspective on this line extends beyond observation. We invested in EigenLayer when its restaking mechanism was not yet widely recognized, led the seed round for Ingonyama (now MoonMath) betting on the migration from ZK hardware acceleration to AI performance layers, and invested in Hyperbolic in 2024, optimistic about its path of using crypto-native verification mechanisms to solve decentralized compute trust problems.
The shared logic behind these bets is: the trust, coordination, and verification problems emerging from AI scaling will ultimately require the mechanism design skills accumulated in the crypto industry to solve. We believe the convergence of Crypto and AI is not a narrative, but a structural opportunity already unfolding.
