Every week we are treated to news about new upgrades to agentic AI programs like Anthropic's Claude model, or the new products made by Austrian developer Peter Steinberger – OpenClaw – now owned by OpenAI. These products all stem from the traditional Web2 developer sphere. Their constant upgrades have overshadowed similar AI development in the Web3 universe. Within the financial press, blockchain AI startups are non-existent.

"Big Tech dominates today's AI economy with its centralized infrastructure,"said Jacob Cantele,  Co-founder & CEO of CoinFello. Cantele was a key product leader at Consensys, where he oversaw MetaMask, the leading Ethereum wallet. CoinFello is an AI agent platform launched in March.

Blockchain based AI projects exist, of course, but that corner of the Web3 market is much smaller than the overall blockchain/cryptocurrency market. 

AI projects in Web3 represent a minority, roughly between 20% to 30%, of the overall money flow into Web3. According to India-based market research firm Market.Us, there were an estimated 17,000 AI agents operating on Web3 platforms by the end of 2025. Autonomous agents now cover 19% of all Web3 activity with billions in daily transaction volume.

The market's largest AI-linked tokens mostly use blockchains for payments and identity. The strongest current examples are Cortex Labs for direct on-chain inference, and AI infrastructure plays ORA and Ritual. These are available to more sophisticated crypto traders on decentralized exchanges rather than on the big centralized ones like Kraken. 

For crypto traders who like the AI theme, Bittensor (CRYPTO: TAO) may be a way to play it without getting too focused.  Bittensor powers open-source, collective AI as a Web3 alternative to the centralized giants and can still be considered part of the AI/blockchain story. Investors who bought Bittensor in 2023 and held it saw the token value go from $0.13 to around $240 as of this writing.

"As of this moment, there hasn't been major traction for any Web3 uses of AI outside of DeFi or development-stage projects," said Dr. Ben Goertzel, CEO and Chief Scientist at SingularityNET & CEO at the Artificial Superintelligence Alliance (ASI). 

ASI was created in 2024 after a token merger between three major Web3 AI projects: SingularityNET, Fetch.ai, Ocean Protocol, and later CUDOS. They consolidated their tokens into a single token now (CRYPTO: FET) to create a unified ecosystem for decentralized AI research, infrastructure, and agent networks. 
Despite Claude and OpenClaw getting all the attention, Goertzel thinks the tide will turn during the next 12 months, "We will see some much more broadly Web3 AI tools emerge," he said. 

"A key factor is a number of recent research results in the AI space that will allow advanced AI systems to be trained more effectively on decentralized networks, without the need for huge hyperscaler server farms. As these alternate methods of training large neural models and other AI systems become a little more mature, they will enable decentralized networks to rival or exceed what the Big Tech hyperscalers are able to do," he predicted. 

AI-linked tokens and tokenization-related projects have been relatively resilient, with AI becoming one of crypto's key investment themes. 

In the first quarter, AI-linked tokens outperformed as attention on artificial intelligence accelerated more broadly among investors and the need for on-chain financial infrastructure for agents is becoming clearer.

Even though crypto has been taking it on the chin all year, with the war in the Persian Gulf not doing long-crypto investors any favors, a subset of tokens outperformed in the first three months of the year, noted Grayscale Research analysts led by Will Ogden Moore in a report on March 23. All the outperformers were concentrated in sectors tied to the big structural growth themes such as AI and decentralized financial infrastructure.

"We are some of the people that are building an alternative to Big Tech AI," said CoinFello CEO Cantele. "We want users, rather than corporate monopolies, to own and control their own agents. We're creating verifiable, user-controlled AI agents that operate autonomously in the decentralized economy." 

CoinFello users can keep their code, data, and models encrypted – essential protection against rogue or tampered agents. CoinFello recently launched a consumer-facing Web3 AI product that allows users to research, execute, or automate any on-chain action by chatting with their "self-sovereign agent" to discover DeFi investment opportunities, execute complex strategies, swap tokens, bridge assets, rebalance portfolios, and automate smart contract interactions across any Ethereum Virtual Machine blockchain. "We are in the beginning stages of a self-sovereign AI economy," Cantele said.

The nearer a startup gets to true on-chain AI, the harder the engineering and the smaller the market cap. Bigger tokens like Bittensor benefit from incentives and settlement rather than pureplay AI execution.

Some analysts are forecasting that Bittensor could reach $1,338.94 by 2030.

For now, it is only a small, but technically serious subset of Web3 projects that are trying to make AI execution, model ownership, and machine-generated outputs verifiable, tokenizable, and settleable on-chain.

"I have said for a long time that Web3 AI will not succeed just by virtue of being Web3 because there are not enough customers outside DeFi to care enough about data sovereignty, decentralization or the other virtues of the blockchain world," said Goertzel. "Web3 AI will succeed by being smarter than Web2 AI. I think we are getting closer to that point now."

Cover art created by the author using Canva.

Benzinga Disclaimer: This article is from an unpaid external contributor. It does not represent Benzinga’s reporting and has not been edited for content or accuracy.