NewsFlash Articles Data Fundraising Skill&API

IOSG: Crypto Didn't Die After Developer Count Halved

IOSG Ventures

Read this article in 30 Minutes

Crypto Developer Halving, 'Veterans' Shift to AI Infrastructure for New Opportunities

Original Title: "IOSG Weekly Brief | After Halving Developer Count: Crypto Is Not Dead, Just Handing Talent Over to AI #326"
Original Authors: Xinyang, Ethan, IOSG Ventures

In 2026, the GitHub activity curve of the Crypto open-source community completed a stunning "reversal." The monthly active developers dropped from the peak of 45K in 2022 to around 23K. This halving of the on-paper data sparked discussions on "narrative exhaustion" on social media. However, as we dissect this curve, what we see is not an industry contraction but a profound "talent deleveraging."

▲ Data Source: Electric Capital Developer Report, based on Crypto Ecosystems Github

Who Left? Who Stayed?

Those who left were mostly newcomers. In February 2024, the single-month new developer additions reached 5462, then sharply declined, with a churn rate of 52% for those in the industry for less than a year. Most of these individuals entered during the bull market, working on NFT minting contracts, forking DeFi protocols, and front-ending for new L2s.

These roles were highly dependent on market hype. Once the hype faded, projects ceased operations, and the positions disappeared. From the data, newcomers' code contributions never exceeded 25% of the overall, indicating that these individuals were never part of the industry's core circle.

▲ Newcomers rushed in during the bull market and left during the bear market; Established devs (2+ years of experience) hit a historical high during the same period
Data Source: Electric Capital Developer Report

On the other hand, developers with over two years of experience saw an increase during the same period, hitting a historical high and contributing about 70% of the code volume. Electric Capital's GP Maria Shen's assessment was straightforward: "When we look at the established developers group, it is growing, and it looks very healthy."

They stayed not because they had no other choice.

Technically, the core work in crypto now generally requires years of accumulation to understand: infrastructure development such as protocol development, security audits, cross-chain architecture. These tasks require years of accumulation to truly master and cannot be eliminated by the market just because the hype has faded.

From an economic perspective, many veterans hold unreleased tokens, governance power in protocols, and equity relationships. Their accumulation in this industry has formed real barriers and rewards.

Looking at the ecosystem distribution, they are voting with their feet: Bitcoin developers grew by 64.3% in two years, Solana by 11.1%, while Cosmos declined by 51.1%, and Polkadot by 46.9%. Veterans are consolidating into ecosystems with real users and revenue, leaving projects that still rely on narratives.

▲ Source: Coincub Web3 Jobs Report 2025
Data Source: Web3.Career

The change in job structure also confirms the same thing. In the new Web3 positions in 2025, the highest proportion is not developers, but Project & Programme Management, exceeding 27%.

For an industry known for being technology-driven, this is counterintuitive, but the logic behind it is not complicated: the industry is transitioning from the construction phase to the execution phase, with over 100 chains needing integration. Institutional clients coming in have completely different compliance and security requirements, and DAO governance needs to balance interests among stakeholders with different demands.

This is not project management in the traditional sense but coordination and judgment in an environment where rules are still taking shape.

While the industry appears to be shrinking on the surface, the core density is actually increasing. The bear market of 2018-2019 also saw a loss of many developers, but afterwards, phenomena such as Uniswap, Aave, OpenSea emerged, defining the bull market of 2020-2021. The builders who stayed this round have a more mature infrastructure, and the AI era has given them a larger stage than the previous round.

What abilities do those who stayed possess?

What special abilities have builders in the Crypto industry developed? To answer this question, we need to go back to the underlying principle of blockchain. Between the bull and bear market cycles, this industry has always operated on the same fundamental rule: Code is Law, and Execution is Finality.

In the 2016 The DAO incident, the attacker exploited a recursive call vulnerability to siphon off $36 million. The code had no bugs, and the logic executed as expected; it was just a boundary that the designer had not anticipated. In 2021, the Poly Network cross-chain bridge was attacked, and $610 million was transferred within a few hours.

There is no platform to halt the transactions, no institution to reverse them, and no legal terms to recoup the losses. This is a structural characteristic that sets crypto apart from almost all other industries: zero room for error, with virtually no post-event intervention.

What this environment has brought about is a set of abilities rarely needed in other industries: under conditions of rule and trust gaps, to construct from scratch a functional system that strangers are willing to participate in.

This ability consists of two aspects. Firstly, establishing trust from scratch without relying on any external authority, solely depending on code and mechanisms to persuade strangers to put their real assets at stake. Secondly, making judgments under technical and economic uncertainties, designing operational systems without regulatory frameworks, historical data, or industry standards as references.

Both aspects have specific validations in crypto. Uniswap operates without a company guarantee, KYC, or customer service. Anyone can provide liquidity to the pool solely based on trust in a few hundred lines of code and an economic mechanism, achieving daily trading volumes in the billions of dollars.

MakerDAO has no central bank endorsement or deposit insurance, relying purely on on-chain governance and collateral mechanisms to maintain DAI's stability.

During DeFi Summer, it was even more extreme – no regulatory framework, audit standards, or historical data to reference. Builders designed AMMs, lending protocols, and liquidity mining, achieving from concept to billions of dollars in TVL in just a few months. This ability 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. The model's decision-making process is opaque, and the output results cannot be independently verified. AI agents are starting to autonomously execute trades and allocate funds, yet the associated rules and constraints are still nonexistent.

Large AI companies control both the models and the evaluation standards, while users lack effective means of verification. Computing power is highly concentrated among a few top players, leading to monopolistic pricing during times of high demand. These issues all point to the same core problem: the trust issue in autonomous systems, replayed on a larger scale in the AI domain.

Crypto builders have been dealing with these issues in an environment without external authoritative rules for years, albeit in the context of on-chain protocols, which has now shifted to AI. A group of people has directly brought the capabilities accumulated in crypto into AI and have produced results.

How are these capabilities being repriced in the era of AI?

In recent years, cases of transitioning from crypto to AI have been abundant, but upon closer inspection, what they took with them is not the same.

The most straightforward path involves a direct transfer of hardware and experience. CoreWeave's three founders, Michael Intrator, Brian Venturo, and Brannin McBee, started mining Ethereum with GPUs in 2017, scaling from one machine to thousands. In 2022, they shut down their mining operations, and two months later, ChatGPT was released. Their GPUs transformed directly into AI computing power. CoreWeave went public on the NASDAQ in March 2025, with an IPO valuation of around $23 billion, reaching a peak market cap of nearly $70 billion.

Alex Atallah, co-founder of OpenSea, dealt with the aggregation and routing of highly heterogeneous assets in the NFT market. He transferred the same set of experiences to AI model routing, founding OpenRouter. Within two years, they served over 5 million developers, reaching a valuation of $5 billion.

Another noteworthy type of migration is seen in NEAR Protocol. NEAR's founder, Illia Polosukhin, was one of the co-authors of the Transformer paper. After leaving Google, he initially aimed to build AI applications using natural language. However, he encountered a real-world problem during development: the need to make cross-border payments to data labelers worldwide, many of whom did not have bank accounts. Blockchain technology emerged as the optimal solution to this payment challenge.

NEAR is now transitioning into an AI infrastructure platform, with a focus on user-owned AI and decentralized confidential machine learning (DCML), allowing users to utilize AI services without exposing their data. The decentralized architecture experience accumulated in NEAR has become the most challenging starting point to replicate in this direction.

After leaving Circle, co-founder Sean Neville has founded Catena Labs, positioning it as an AI-native bank, directly migrating the understanding of stablecoin infrastructure to an AI agent financial scenario, with a $18 million seed round led by a16z crypto.

Nader Dabit, a senior developer at Aave and Lens Protocol, has transitioned to Cognition, bringing his developer ecosystem development experience from multiple crypto protocols into the AI agent tools field.

What these individuals take with them is not just GPU hardware or user networks, but rather the intuition of mechanism design, developer ecosystem development experience, and the judgment to build a trusted system from scratch in the absence of rules. These abilities happen to correspond to three structural gaps encountered in AI scalability.

Aggregation and Optimization of Computing Power

Computing power is the most direct bottleneck to AI scalability. Training and inference require a large number of GPUs, with high demand fluctuations, expensive and queuing cloud providers, and enterprises unwilling to hoard hardware. This problem has two aspects: how to aggregate and allocate computing power, and how to use the aggregated computing power more efficiently. Crypto builders have directly transferable accumulations in both of these aspects.

Hyperbolic addresses the allocation and trust issues. Founder Jasper Zhang has brought decentralized mechanism design into the AI computing power track: tokens encourage dispersed GPU holders to contribute idle computing power, but the core issue is trust.

Why trust that a calculation result provided by a stranger node is correct? The core innovation PoSP uses random sampling plus game theory, making honesty the dominant strategy for nodes, without requiring full verification, low cost, scalable, and reliable results. This mechanism is directly migrated from the logic of crypto verifying the behavior of unknown nodes.

MoonMath tackles the efficiency issue. Previously known as Ingonyama, it focused on ZK hardware acceleration, increasing ZK proof generation speed several times under extreme computational constraints.

It is now shifting towards the Physical AI performance layer, working on Sparse Attention Acceleration for video diffusion models (LiteAttention), Low-Rank Decomposition for FFN layers (LiteLinear), and Training Backpropagation Acceleration (BackLite). From ZK acceleration to AI inference acceleration, the underlying capabilities are the same: making math run faster under extreme computational constraints. Though the track has changed, the accumulation has not been wasted.

AI Governance and Incentive Mechanism Design

When multiple AI agents begin to collaborate to perform tasks, how can we ensure that they will not undermine the overall system while pursuing their individual objectives? Each participant is optimizing its own objective function, with no guarantee that their collective actions will keep the system intact, especially considering that the agents operate at speeds far beyond human intervention.

This is a type of issue that crypto builders have repeatedly dealt with in DAO governance and tokenomics design: aligning participants with vastly different incentives to operate in a predefined direction without a central authority. The crypto solution has been economic mechanisms, where any misconduct incurs a tangible economic cost, rules are encoded, and enforcement is automatic.

EigenLayer has directly applied this mechanism to the AI domain. Through a restaking mechanism, nodes are required to stake assets before engaging in collaboration. Non-compliance or misconduct triggers automatic penalties, where rules are not suggestions but rigid boundaries with real economic consequences.

EigenCloud extends this logic to the verifiable computation and collaborative governance of AI agents, ensuring that agents must operate within predefined boundaries while pursuing their own objectives. Economically constraining agents proves to be more reliable than constraining them with ethical guidelines.

AI Agent Self-Payment

There is another fundamental question: how do agents pay? Traditional payment systems are designed for humans; credit cards require accounts, and bank transfers need authorization, with each step assuming a human operator with identity who can wait. Agents do not wait; they may generate numerous requests per second, each involving micropayments, rendering traditional payment channels ineffective in this context.

Stablecoins and on-chain rules are foundational infrastructure that crypto builders have already established, supporting programmability, permissionlessness, and 24/7 operation. These three characteristics happen to be hard requirements for the agent payment scenario, with the missing piece being a protocol that integrates stablecoin payments into agent workflows.

x402, launched by Coinbase in May 2025, activates the HTTP 402 status code to embed stablecoin payments directly into HTTP requests. Agents initiate requests and concurrently complete payments, requiring no accounts and settling in around two seconds.

As of April 2026, the x402 Protocol has processed over 165 million transactions, with a total transaction volume of about $50 million and an active agent count of 69,000 (Source: x402 Foundation). Cloudflare, AWS, Stripe, and Anthropic MCP have all been onboarded. Agent payments have become a track with real traffic.

The three directions correspond to three structural gaps encountered in AI scalability: aggregation and efficiency of computing power, incentive alignment for multi-agent collaboration, and infrastructure for autonomous payments. These three issues do not have ready-made answers in traditional software architecture, but the crypto industry has corresponding experience in addressing them. The capabilities have not disappeared; they have just found new application scenarios.

New Positioning of Builders: From Contract Writers to Rule Setters for AI

The scalability of AI is creating a functional gap that did not exist before. It is not a talent gap in technology but a gap in people who can design trust mechanisms in autonomous systems. As the service recipients shift from humans to AI, the role of crypto builders is also being redefined.

The table below contrasts the dimensional shifts in specific functional paradigms:

The core difference between the two paradigms lies not in the technology stack but in the way trust is established and rules are enforced. In the Pre-AI era, crypto builders dealt with human participants, rules were encoded in contracts, there was zero margin for error, but the system boundaries were relatively clear.

In the AI-Native era, when the interacting entities become autonomously operating AI agents, the challenge is: the behavior of agents is unpredictable, the execution speed far exceeds human intervention windows, and the system boundaries themselves need to be redefined under greater uncertainty.

The role of the crypto builder is transitioning from "writing secure contracts" to "designing trust mechanisms for AI autonomous systems."

The recruitment by top institutions is already reflecting this shift:

▲ Q1 2026 Head Trading Platforms Actively Opening AI/Data Core Positions
Source: Gate Research Institute

The recruitment trend of top trading platforms and institutions in 2026 clearly reflects this shift: no longer just hiring AI engineers or crypto developers, but looking for individuals who can bridge the gap, understanding both on-chain incentive alignment and governance games, embedding AI tools deeply into crypto workflows, and designing mechanisms that align agents with regulation and long-term user incentives.

The allocation of capital has also echoed this sentiment. Paradigm is raising a new fund of up to $15 billion, expanding its investment focus from crypto to the AI and robotics fields.

Haun Ventures has closed Fund II at $10 billion, with a key focus on the financial infrastructure integrating crypto and AI, particularly supporting autonomous trading and coordination by AI agents in payment systems, stablecoins, and agent-to-agent economic ecosystems.

a16z crypto has closed its fifth fund at $22 billion (Crypto Fund V), explicitly stating that the fund will be 100% dedicated to the crypto space. In the face of the complexity and opacity of the AI era, they will concentrate on the aspects of crypto that emphasize transparency, verifiability, and applications of decentralization.

According to PitchBook data, in 2025, around 40% of U.S. crypto VC investments flowed into companies simultaneously involved in AI business, a significant increase from 2024.

Similarly, as crypto builders turn to AI, the paths chosen under different market environments show significant differences.

In the U.S., as the regulatory environment has become relatively clearer, there has been genuine space for protocol-layer innovation. The capital network density is high, the path from idea to funding is short, and there is ample room for trial and error.

Common characteristics among projects such as Hyperbolic, EigenCloud, Gensyn, and Ritual include designing new mechanisms from scratch rather than simple application integration on existing systems. Top VCs have a clear investment thesis on directions like "verifiable computation, agent coordination, and decentralized ML," and are willing to provide sufficient leeway for early-stage technical exploration.

The situation in Asia is different. Singapore and Hong Kong, China, mainly play roles in regulatory compliance and institutional fund transfer, with relatively conservative regulatory frameworks and a lower tolerance for pure protocol-layer innovation. When builders with a crypto background turn to AI, they more often choose the path of application layer and industry integration—leveraging the user base, payment capabilities, or data assets accumulated through crypto to quickly integrate into AI products and services.

This is not a gap in capability, but a difference in path choices due to different market signals and regulatory environments: the U.S. encourages bottom-up innovation of mechanisms and early-stage technological exploration, while Asia emphasizes compliance-friendliness, rapid monetization, and deep integration with traditional industries.

Returning to the GitHub curve at the beginning. Monthly active developers dropped from 45K to 23K, seemingly indicating industry contraction. However, among those who stayed, the proportion of established developers reached a historic high, moving towards ecosystems with real users, while being repriced by the AI industry in unprecedented ways.

When AI scalability encounters structural bottlenecks such as compute aggregation, agent autonomous payments, data and decision verifiability, privacy coordination, etc., these builders at the intersection of Crypto and AI, with their long-standing sensitivity to rules, incentives, and authenticity, are gradually transforming into scarce systemic capabilities in the AI era.

As an investment institution that has been deeply involved in crypto infrastructure since 2017, IOSG's assessment of this trend goes beyond observation.

We participated in investments in EigenLayer's restaking mechanism before it was widely recognized by the market, led the seed round investment in Ingonyama (now MoonMath) betting on ZK hardware acceleration migrating to AI performance layer, and in 2024 invested in Hyperbolic, optimistic about its path to solving decentralized compute trust issues through crypto-native validation mechanisms.

The common logic behind these deployments is: the trust, coordination, and validation issues faced by AI at scale will ultimately require the mechanism design capabilities accumulated by the crypto industry to solve. We believe that the intersection of crypto and AI is not just a narrative but a structural opportunity unfolding.