Crypto's open-source developer base is contracting at a pace not seen since the 2022 bear market, according to data from Artemis analytics and GitHub's Octoverse report. Weekly code commits to cryptocurrency repositories have fallen approximately 75% since early 2025 — from roughly 850,000 to 210...
"AI agents will make 1 million times more payments than humans, and they will use crypto." — Changpeng Zhao, Founder, Binance
Crypto's open-source developer base is contracting at a pace not seen since the 2022 bear market, according to data from Artemis analytics and GitHub's Octoverse report. Weekly code commits to cryptocurrency repositories have fallen approximately 75% since early 2025 — from roughly 850,000 to 210,000. The number of active developers has declined 56%, dropping from 8,700 to approximately 4,600.
The cause is not a market crash. Bitcoin trades between $60,000 and $70,000. Ethereum holds above $2,000. The capital is there. The builders are not. They have moved to artificial intelligence, where global venture funding reached $211 billion in 2025 — more than ten times the $19.7 billion deployed into crypto over the same period, according to Crunchbase and PitchBook data.
One metric offers partial reassurance: experienced developers with more than two years of tenure in crypto grew 27% year-over-year and now produce roughly 70% of all commits. The exodus is concentrated among part-time contributors and newcomers, who declined 58%. Crypto is not dying. It is consolidating around its veterans — but the pipeline of new talent has effectively collapsed.
The raw data is unambiguous. According to Artemis analytics, weekly code commits across cryptocurrency projects peaked at approximately 850,000 in early 2025. By March 2026, that figure had fallen to 210,000 — a 75% decline in roughly 14 months.
Active developer counts tell the same story. Monthly active developers working on crypto projects have fallen to their lowest level since at least 2022. The current count of approximately 4,600 active developers represents a 56% decline from the 8,700 recorded at the start of the measurement period.
For historical context, crypto monthly active developers peaked at approximately 31,000 in 2022, fell to 23,600 in 2024, and reached an estimated 18,000 by mid-2025 before the current acceleration downward. The decline is not cyclical in the traditional sense — previous drawdowns correlated with price crashes. This one does not.
Meanwhile, GitHub's broader platform tells the opposite story. The platform added 36 million new developers in 2025, bringing the global base above 180 million. Platform-wide commits rose approximately 25% year-over-year. Crypto is shrinking inside a growing ecosystem, which makes the contraction structural rather than environmental.
The decline is not evenly distributed. Some ecosystems have been hit far harder than others.
| Chain | Developer Change (3-month) | Current Weekly Active Devs | |-------|---------------------------|---------------------------| | Ethereum | -34% | 2,811 | | Solana | -40% | 942 | | Base (Coinbase L2) | -52% | 378 | | Celo | -52% | Not disclosed | | Aptos | -60% | Not disclosed | | BNB Chain | -85% (commits) | Not disclosed |
Ethereum retains the largest absolute developer base at 2,811 weekly active developers, but a 34% decline over three months is severe for any ecosystem. Solana, which was the biggest draw for new crypto developers in 2024 according to Electric Capital's annual report, has shed 40% of its developer base.
The most dramatic losses are at the margins. Base, Coinbase's Layer 2 that was among 2024's fastest-growing ecosystems, has lost more than half its developers. BNB Chain's commits have plunged 85% — a figure that suggests near-abandonment of open-source development on the chain. Aptos, which attracted significant venture funding and developer attention in 2023-2024, has lost approximately 60% of its contributors.
The destination is clear: artificial intelligence infrastructure. GitHub's Octoverse report documents the shift in granular detail.
AI-related repositories on GitHub now exceed 4.3 million. LLM SDK imports surged approximately 178% to over 1.1 million. Generative AI projects attract over 1 million monthly contributors — roughly 217 times the entire remaining crypto developer base.
The tooling shift is equally telling. Jupyter Notebooks, the standard environment for AI/ML development, grew approximately 75%. Dockerfile repositories, which indicate production-grade deployment work, jumped 120%. TypeScript, the language of choice for both Web3 frontends and AI agent frameworks, gained over 1 million contributors in a single year.
The migration is rational from an individual developer's perspective. AI engineering roles at major technology companies offer compensation packages ranging from $150,000 to $700,000+ annually, according to Glassdoor and Levels.fyi data. Crypto developer salaries average $140,000-$180,000, with less variance and fewer employers competing for talent.
More critically, the opportunity space is different. AI is in its infrastructure-building phase, analogous to where crypto was in 2017-2018. The problems are novel, the tooling is immature, and the upside is perceived as larger. Crypto, by contrast, is increasingly perceived as solved infrastructure — boring but functional, which is precisely what repels early-career engineers seeking greenfield challenges.
Capital flows explain the migration better than any cultural narrative. Crunchbase reports that AI attracted approximately $211 billion in global venture funding in 2025, accounting for roughly half of all venture capital deployed worldwide. PitchBook pegs crypto venture capital at $19.7 billion over the same period — a ratio of approximately 11:1.
In Q1 2026, the disparity has widened. Total crypto VC funding has stayed below $1 billion for the year so far, with $135 million raised in one recent week representing one of the slowest totals of 2026. AI accounted for 58% of all fintech-adjacent venture investments in 2025, and that share appears to be growing.
The funding drought creates a self-reinforcing cycle. Fewer funded projects means fewer job openings for developers. Fewer developers means less innovation. Less innovation means less reason for venture capital to deploy into the sector. According to reports from The Block, institutional investors have shown what one source described as "zero interest" in non-AI deals — a sentiment that has bled into crypto venture allocations.
The one bright spot in the data is the composition of who remains. Electric Capital's latest developer report found that established developers — those with more than two years of experience working in crypto — grew approximately 27% year-over-year. These veterans now produce roughly 70% of all commits in the ecosystem.
This suggests the departure is concentrated among:
The veterans who remain tend to work on core infrastructure: consensus mechanisms, virtual machines, cryptographic libraries, and protocol-level code. This is the work that requires deep domain expertise and cannot be easily replaced by general-purpose software engineers.
The implication is a smaller but more focused ecosystem. The projects that survive the talent drain will be those with strong core teams and real usage. The long tail of speculative, lightly-maintained protocols is being pruned by attrition rather than market forces.
A 75% decline in code contributions carries concrete risks beyond slower feature development.
Security auditing capacity is thinning. Fewer developers means fewer eyes on code, fewer internal reviewers, and longer audit queues. The Solv Protocol hack in March 2026 — a $2.7 million loss from a reentrancy vulnerability in an ERC-3525 implementation — illustrates the type of exploit that proliferates when codebases are understaffed. The attacker triggered a double-minting flaw 22 times, converting 135 BRO tokens into approximately 567 million. This is the kind of vulnerability that experienced code review catches.
Protocol upgrade velocity is slowing. Ethereum's Glamsterdam fork, while significant, required extended timelines. Smaller chains face the prospect of multi-quarter delays on routine upgrades simply because the teams are smaller.
Innovation is narrowing. The most active areas of development are increasingly those with direct commercial applications — stablecoins, real-world asset tokenization, and AI-agent infrastructure. Experimental work on novel consensus mechanisms, privacy technologies, and alternative execution environments is receiving fewer commits and less attention.
The paradox is that crypto needs more developers precisely when it has fewer. The industry's response — exemplified by Kraken's March 2026 launch of a CLI tool with built-in MCP server support for AI agents — is to make the remaining infrastructure more accessible to AI-powered development tools. Whether AI coding assistants can offset the human developer shortage remains unproven at scale.
The crypto developer exodus is not a crisis of conviction. It is a crisis of opportunity cost. When AI engineering offers higher compensation, larger funding pools, more novel problems, and broader career optionality, the rational developer migrates. The 75% decline in commits is the aggregate expression of thousands of individual career calculations.
What remains is a leaner, more experienced ecosystem. The 27% growth in veteran developers suggests that crypto's core infrastructure maintainers view this as their domain — they are not leaving. But the feeder system that produces the next generation of protocol engineers, auditors, and tooling developers has effectively dried up.
The industry's bet — implicit in launches like Kraken's AI-agent CLI and the broader push toward autonomous on-chain agents — is that artificial intelligence will both consume and produce crypto development. AI agents may become the fastest-growing user base of blockchain infrastructure. Whether they can also maintain and secure that infrastructure is the open question that the next 12 months will begin to answer.