Since early 2025, weekly crypto code commits have decreased by nearly 75%, from roughly 850,000 to 210,000, while the number of active developers has decreased by 56%, to about 4,600. In contrast, AI attracted over $258 billion in venture funding in 2025, while crypto startups received an estimated $18–20 billion. These data show that money has a new address, and talents are after it. However, narratives don't matter to infrastructure. The pipelines still need to be built. And whether the developers that stayed did so out of conviction or because they had no better offer is the awkward subject that no one in Web3 wants to publicly address. To get an honest answer, I spoke with Uttam Singh, Senior Developer Relations Engineer at Alchemy, who sits at the exact place where that question stops being theoretical. Uttam Singh Uttam Singh Let’s dive in! 75% of crypto developers left for AI in under a year. What does that say about the state of Web3 — and is there a way back? Uttam: To be honest, it’s true that crypto developer activity is down significantly from the peak, but I don’t think this is something new or unique to this cycle. We’ve seen this pattern before. Every time the market enters a bear phase, a lot of developers who came in during the hype cycle tend to leave. When new narratives cool off and short-term incentives disappear, only the more committed builders stay. What’s different this time is that AI is not just another narrative competing for attention — it’s a real, massive shift with strong fundamentals and an insane amount of capital flowing into it. Naturally, developers are moving toward where both the opportunity and funding are. So yes, AI is definitely competing with crypto for talent right now. At the same time, I think this is also a sign that the crypto industry is maturing. For a long time, markets and VCs rewarded protocols without real products or sustainable use cases. That phase is now being corrected. The focus is shifting toward actual utility, real users, and real businesses. Another factor is that current crypto use cases — mainly payments and DeFi — while powerful, are not equally exciting to all developers. Compared to building consumer AI products or working on cutting-edge models, these areas can feel narrower in scope. Regulatory uncertainty has also played a role. It creates friction and slows down experimentation, which discourages some developers from committing long-term. But if you zoom out, the long-term direction is still very strong. If the financial system, identity, and ownership move onchain  which I believe they will — then most developers will eventually interact with crypto rails, even if they don’t identify as “crypto developers.” Also, with AI, building has become significantly easier. A smaller number of developers can now produce much more output. So even if the number of developers decreases, it doesn’t necessarily mean innovation slows down. And again, this is cyclical. It has happened before, and it will likely happen again. You've onboarded thousands of engineers into Web3. What's the one thing the ecosystem kept telling new developers that was simply not true? Uttam: One of the biggest things that wasn’t really emphasized or sometimes was implicitly ignored — is that projects actually need a real revenue model and product-market fit before launching. For a long time, especially in the last few years, crypto created this perception that you could launch a token first and figure out everything else later. That distribution, tokenomics, and hype could compensate for the lack of a real product. In reality, that doesn’t work long-term. If you don’t have real users, real demand, and a clear way to generate value, your project is going to die sooner or later. We’re seeing that play out now. Revenue generation and sustainability were not treated as priorities in many projects during the previous cycle. But now, the ecosystem is realizing that you have to build something people actually want and are willing to use or even pay for before going public or launching a token. The mindset is shifting from “launch fast and tokenize” to “build something meaningful, validate it, and then scale.” And honestly, that’s a healthy correction for the space. AI agents making micropayments onchain sounds inevitable on paper. What's the ugly technical reality most people aren't talking about? Uttam: I do think that AI agents will end up moving a significant amount of money over time, potentially even more than humans, and they will likely rely on stablecoins, smart contracts, and onchain rails rather than traditional banking infrastructure. The current financial system was not designed for machines, while crypto is much more aligned with that future. However, the reality is more nuanced than people make it sound. One important thing that people often overlook is that stablecoins and onchain payment systems are very good at moving money, but that’s only one part of what modern financial systems do. For example, card networks don’t just move money. They: extend unsecured credit
allow pre-authorized transactions
provide fraud detection systems
enable chargebacks and dispute resolution extend unsecured credit allow pre-authorized transactions provide fraud detection systems enable chargebacks and dispute resolution Onchain payments, on the other hand, are typically irreversible. If an AI agent makes a mistake, sends funds to the wrong address, or interacts with a malicious contract, the funds are gone. There is no native recovery or dispute mechanism. So, from a user and consumer perspective, you’re giving up a lot of protections when you move to purely onchain systems. Also, there isn’t a strong, standardized fraud or safety layer at the network level yet. Each application has to build its own safeguards, which leads to fragmentation and an inconsistent user experience. Because of this, I don’t think stablecoins or onchain payments will completely replace traditional payment systems like cards. Instead, they will coexist and serve different types of transactions. Onchain payments will likely dominate machine-to-machine transactions and programmable flows, while traditional systems will still be preferred for consumer-facing use cases where protections matter. Vitalik reshared one of your technical videos. What did that moment reveal to you about what the Ethereum core community actually values? Uttam: That moment was quite meaningful for me, but more importantly, it reinforced something fundamental about Ethereum as an ecosystem. The video that got reshared wasn’t something flashy or hype-driven. It was a deep technical explanation of an Ethereum upgrade, specifically around changes at the protocol level and how they impact scalability and transaction processing. What that showed me is that the Ethereum core community, including Vitalik Buterin, genuinely values depth, research, and clarity of understanding over surface-level narratives. Ethereum has always been a research-driven ecosystem. A lot of effort goes into solving hard problems — scalability, decentralization trade-offs, execution environments — even if those solutions take time to materialize. The focus right now is very clear: improving the base layer, scaling Ethereum L1, and strengthening its role as a settlement layer for rollups and other systems. Contributions that help people understand these deeper layers, not just use them — are highly valued. That experience made it clear that if you engage with Ethereum at a fundamental level and communicate those ideas well, the ecosystem pays attention. Crypto.com just laid off 12% of its staff, and the CEO said companies that don't pivot to AI "will fail." Is that the future of Web3 companies — or just the latest excuse to cut headcount? Uttam: I don’t think this is specific to Web3 at all. This is a broader shift happening across the entire technology industry. AI is fundamentally changing how work gets done. The productivity gains are real — a small team using AI effectively can do the work that previously required a much larger team. Because of that, companies are rethinking how they structure teams and allocate resources. So when you see layoffs, it’s often less about “failure” and more about optimization. Companies are adjusting to a new baseline where efficiency expectations are higher. This isn’t just happening in crypto — it’s happening across major web2 tech companies like Google and Amazon as well. At the same time, it’s not just about reducing headcount. Companies are still hiring, but they’re looking for people who can effectively leverage AI and operate at a much higher level of output. Instead of hiring 10 engineers, a company might hire 2–3 highly effective engineers who can orchestrate AI tools and agents to achieve the same or better results. So I wouldn’t frame it as an excuse. It’s more of an adaptation to a rapidly changing environment. And going forward, the expectation is clear: if you’re not integrating AI into how you build and operate, you’re going to fall behind — regardless of whether you’re in Web2 or Web3. If a serious developer came to you today and said, "I want to build at the AI x crypto intersection" — what's the first thing you'd tell them NOT to do? Uttam: The first thing I would tell them is: don’t jump in and start building immediately. A lot of people see the AI x crypto narrative gaining traction and rush to build something just to be part of it. But in many cases, they haven’t actually understood the problem they’re trying to solve. That leads to products that technically combine AI and crypto, but don’t deliver meaningful value to users. Instead, I would encourage them to first spend time understanding where the real gaps are: What problems does AI currently have that crypto can help solve?
What problems does crypto have that AI can improve? What problems does AI currently have that crypto can help solve? What problems does crypto have that AI can improve? For example, crypto can help with payments, coordination, and ownership in AI systems, while AI can improve usability, automation, and developer experience in crypto. Another important point is not to reinvent infrastructure that already exists. There are already emerging payment rails, protocols, and tooling being developed by companies with strong distribution. Building on top of these is often a much better approach than trying to create everything from scratch. Especially in areas like payments, where companies like Coinbase(X402), Stripe(MPP), and players in the crypto ecosystem are actively pushing new standards and infrastructure, leveraging existing rails can significantly accelerate development. The biggest mistake right now is building something just because it sounds cool, “AI + crypto”  - without validating whether anyone actually needs it. At the end of the day, product-market fit matters more than the stack.

Between Two Computer Monitors: This story includes an interview between the writer and guest/interviewee. 

Ethereum

Google

Amazon

Coinbase

Stripe

Crypto.com

Why Brazil's Biggest Institutions Chose XDC, XRP Ledger, and Polygon for Tokenization

Read My Stories

Too Long; Didn't Read

Is Web3 Still for Developers - or Just for Speculators With a GitHub Account?

Is Web3 Still for Developers - or Just for Speculators With a GitHub Account?

About Author

Comments

TOPICS

THIS ARTICLE WAS FEATURED IN

Related Stories

BlackRock's Fink Joins World Economic Forum as Co-Chair—Is Global Crypto Integration Next?

Crypto Goes AI: 3 Early Projects to Keep an Eye On

Don’t Bet on a Single Technology: Blockchain and AI could bring New Outstanding Projects

The Convergence of AI and Blockchain: A Game Changer in the Tech Industry

What Happens When AI and Crypto Meet?

When AI Meets Crypto: A Deep Dive into the Confluence of Machine Learning and Digital Currencies

BlackRock's Fink Joins World Economic Forum as Co-Chair—Is Global Crypto Integration Next?

Crypto Goes AI: 3 Early Projects to Keep an Eye On

Don’t Bet on a Single Technology: Blockchain and AI could bring New Outstanding Projects

The Convergence of AI and Blockchain: A Game Changer in the Tech Industry

What Happens When AI and Crypto Meet?

When AI Meets Crypto: A Deep Dive into the Confluence of Machine Learning and Digital Currencies

Light-Mode

Classic

Newspaper

Minty

Dark-Mode

Neon Noir

Minty

HN StartUps