Weekly code commits to public blockchain repositories have fallen approximately 75% since early 2025, from roughly 850,000 to 210,000, according to Artemis Analytics data tracked across 141 ecosystems. Active weekly developers have declined 56% to approximately 4,600. The contraction is not an in...
"Still healthy nominally, but crypto clearly not the cool kid on the block these days." — Omar Kanji, Partner, Dragonfly Capital
Weekly code commits to public blockchain repositories have fallen approximately 75% since early 2025, from roughly 850,000 to 210,000, according to Artemis Analytics data tracked across 141 ecosystems. Active weekly developers have declined 56% to approximately 4,600. The contraction is not an industry-wide GitHub phenomenon — platform-wide commits rose 25% year-over-year and GitHub added 36 million new accounts in 2025 alone, bringing total users past 180 million.
The talent is not disappearing. It is migrating. Repositories importing large language model software development kits surged 178% year-over-year to more than 1.1 million. Generative AI projects now attract over one million monthly contributors. Crypto is losing a labor market competition to AI, and the consequences are measurable across security, protocol maintenance, and ecosystem depth.
The decline is concentrated among newcomers and part-time contributors with under 12 months of experience, a cohort that shrank 58%. Developers with more than two years of tenure grew 27% and now produce roughly 70% of all commits. What remains is a smaller, more experienced core — but one with less capacity for expansion, experimentation, and redundancy.
Artemis Analytics and Electric Capital data paint a consistent picture. Weekly code commits to blockchain repositories peaked near 850,000 in early 2025. As of March 12, 2026, the figure stood at 217,000 — a 75% decline over roughly 14 months. Weekly active developers dropped from approximately 9,000 to 4,600, a 50-56% reduction depending on the measurement window.
Electric Capital's longer-term tracking shows monthly active crypto developers peaked at approximately 31,000 in 2022, declined to 23,600 in 2024, and are projected at roughly 18,000 by mid-2025 estimates. The trajectory has not reversed.
This is not a function of a broader tech downturn. GitHub's Octoverse 2025 report recorded 1.12 billion contributions from over 180 million developers globally, with new registrations averaging one per second throughout the year. Platform-wide commits rose 25.1% year-over-year. Monthly pull request merges averaged 43.2 million, up 23%. Developers created more than 230 new repositories every minute.
Crypto's share of this activity is shrinking in both absolute and relative terms.
The contraction is broad-based but uneven. According to Artemis data measured over a recent three-month window:
| Ecosystem | Developer Decline | Current Weekly Active Devs | |-----------|-------------------|---------------------------| | Ethereum | -34% | 2,811 | | Solana | -40% | 942 | | Base | -52% | 378 | | Celo | -52% | — | | Aptos | -60% | — | | BNB Chain | -85% (commits) | — |
Ethereum retains the largest absolute developer base but is not immune. Solana, which invested heavily in developer relations and ecosystem growth programs, lost 40% of its weekly contributors. Base, the Coinbase-incubated Layer 2 positioned as a developer-friendly on-ramp, saw its pool cut in half.
The sole bright spot: wallet infrastructure development grew approximately 6% to 308 weekly active developers — a category likely buoyed by the persistence of consumer-facing demand for custody and transaction tooling.
BNB Chain's 85% commit decline stands out as an extreme case, suggesting that chains reliant on incentive-driven developer programs may experience the sharpest reversals when external pull factors intensify.
The data points to a single dominant destination: artificial intelligence.
GitHub's platform statistics tell the story. AI-related repositories now exceed 4.3 million, nearly doubling in under two years. More than 1.1 million public repositories import an LLM SDK, with 693,867 of those created in the past 12 months alone — a 178% year-over-year increase. Jupyter Notebook repositories, commonly used for machine learning experimentation, grew approximately 75%. Dockerfile repositories used to deploy AI applications increased roughly 120%.
Generative AI projects attracted a median of approximately 160,000 monthly contributors from September 2024 through August 2025, rising from 86,000 in January 2025 to a peak of 206,830 in May — a 132% year-over-year increase.
Individual departures illustrate the pattern. Nader Dabit, formerly Director of Developer Relations at Eigen Labs, joined Cognition, the company behind AI coding tools Devin and Windsurf. In a December 2024 post, Dabit wrote: "I've been working in software for 13+ years, the two biggest opportunities in my lifetime are crypto and AI and the time is now." He chose AI.
Compensation data provides partial context. According to Built In and CryptocurrencyJobs.co, blockchain developer salaries in the US average approximately $140,000 annually, with a common range of $105,000 to $146,000. AI professionals command a median of $160,000, with machine learning engineers at top firms earning $212,928 annually. Specialized AI skills carry 25-45% premiums over base compensation, according to the Rise AI Talent Salary Report.
The gap exists but is not dramatic enough to explain a 75% commit decline on its own. The more significant factors appear to be opportunity scale and perceived trajectory. AI-related venture funding dwarfs blockchain investment. GitHub added 36 million developers in a single year — a migration that creates network effects, tooling improvements, and career optionality that compound over time. Crypto, by contrast, offers a contracting developer community and reduced venture activity outside of stablecoins and infrastructure.
As Bloomberg reported on March 26, 2026, crypto VCs are increasingly abandoning Web3 application bets for the perceived dependability of stablecoin-related investments — further narrowing the innovation surface that would attract new builders.
The developer exodus has corporate parallels. On March 19, 2026, Crypto.com announced 12% workforce reductions — approximately 180 employees from a 1,500-person staff. CEO Kris Marszalek stated that "companies that do not make this pivot immediately will fail," referring to enterprise-wide AI integration. Layoffs concentrated in growth and customer relationship management departments.
The announcement followed Crypto.com's $70 million acquisition of the ai.com domain, signaling that even firms with "crypto" in their name view AI as the primary operational frontier.
This pattern — crypto-native companies pivoting internal resources toward AI while shrinking blockchain-specific teams — suggests the developer talent drain is both a cause and effect of shifting institutional priorities.
The developer contraction carries direct security consequences. According to industry estimates cited by CoinLaw and Sherlock, only approximately 2,000 security specialists globally focus on blockchain — a figure that has not kept pace with the growing complexity of DeFi protocols, cross-chain bridges, and zero-knowledge systems.
Smart contract audit costs reflect the scarcity. Comprehensive audits in 2025-2026 range from $20,000 to $500,000. Rust-based Solana program audits carry a 25-40% premium due to the smaller reviewer pool. Cairo and Move language audits run 30-45% above EVM equivalents. ZK circuit audits command 80-120% premiums above EVM baseline pricing.
The financial stakes are substantial. Over $6 billion has been stolen through DeFi hacks since 2020, with the majority targeting protocols that either skipped audits or used low-quality reviews. The first quarter of 2026 alone saw $137 million in crypto exploits, with infrastructure attacks — not smart contract bugs — emerging as the dominant vector, according to separate webthreepedia research.
A shrinking developer base means fewer eyes on code, longer audit queues, and increased reliance on automated tooling that cannot yet replicate human judgment on novel attack surfaces.
The optimistic interpretation of the data rests on a consolidation thesis: the developers who remain are more experienced, more committed, and more productive. The 27% growth in developers with two-plus years of tenure, who now generate 70% of commits, supports this reading.
However, consolidation carries risks. A smaller, more senior developer base reduces redundancy. Ecosystem experimentation — the speculative projects that occasionally produce category-defining protocols — depends on a broad base of newcomers willing to build without guaranteed returns. The 58% decline in sub-12-month contributors represents a collapse in this experimentation layer.
The question is whether AI-assisted development tools can compensate for headcount losses. If experienced developers using AI copilots can match the output of larger teams, the commit decline may overstate the actual productivity loss. But this remains an untested hypothesis at ecosystem scale, and commit counts — while imperfect — have historically correlated with protocol maintenance quality and security responsiveness.
The 75% decline in blockchain code commits is not a temporary market-cycle dip. It is a structural reallocation of engineering talent toward artificial intelligence, driven by larger funding pools, faster career growth, and platform-level network effects that compound with each departing contributor. The remaining crypto developer base is more experienced but less diverse, less experimental, and potentially less resilient to the security and maintenance demands of an ecosystem managing trillions in value. Whether AI-augmented productivity can offset the headcount loss is the open question — one that the market will answer in the coming quarters through protocol uptime, audit coverage, and the pace of meaningful new deployments.