Over the past year, AI-driven projects have taken center stage in the crypto space. Currently, this sector includes 20 tokens with a collective market cap nearing $20 billion, which makes it the smallest crypto segment at just 0.67% of the overall market.

To put this into perspective, the Financials sector holds a market value of about $519 billion.

Grayscale attributes the AI sectors modest size to the early stages of project development but sees significant growth potential ahead, both in absolute terms and as a share of the broader crypto ecosystem.

AI Sector is Catching Up Fast

Back in 2023, the total market capitalization of tokens in the AI Crypto Sector was only $4.5 billion. Since then, its value has grown over fourfold in a span of two years. Year-to-date, TAO has been the top performer among these assets with a modest gain of 2%, while ElizaOS has been the weakest, plunging by 80%.

According to the latest report by Grayscales Managing Director of Research, Zach Pandl, and Grayscale Investments Research Analyst Will Ogden Moore, stablecoins could soon play a pivotal role in the crypto ecosystem, particularly as tools for powering AI agents, due to their programmable and efficient payment capabilities.

Institutional players like Stripe, Meta, and major banks are entering the space amid improving regulatory prospects. Coinbase has also introduced a stablecoin-ready payments standard for AI. Upcoming regulatory developments, such as the crypto market structure bill and the GENIUS stablecoin bill, could further accelerate adoption.

Key Trends in Decentralized AI

As decentralized AI technologies rapidly evolve, Grayscale is turning its attention to other key developments, such as Bittensors upcoming halving and the expansion of its subnet activity. TAO, the platforms native token and the largest AI token by circulating market cap, is modeled after Bitcoin with a hard cap of 21 million and a halving every four years, the first expected later this year.

Since the dTAO upgrade launched in February, which allowed subnets to be investible, participation has surged. The report revealed that more than 7% of the circulating TAO supply is now allocated to subnets.

Grayscale also identified distributed training as a particularly promising area within the AI crypto sector. Prime Intellect, for one, has shown it can train massive models and has surpassed 30 billion parameters, using a global network of idle GPUs, while moving away from reliance on centralized infrastructure. If widely adopted, this model could reduce AI training expenses and broaden accessibility.

Meanwhile, projects like Prime, Gensyn, and Nous Research may introduce tokens later in the year. Grass, too, has emerged as a standout in the AI crypto space, reportedly generating tens of millions in annualized revenue from selling web-scraped data to AI labs.

This positions it as a rare example of a non-financial, revenue-producing project. The network has steadily increased its data output this year, and a consumer product launch may be on the horizon. Additionally, Virtuals is also seeing strong traction, as it raked in $30 million annually from trading fees on AI agent tokens.