While $240 billion pours into AI… why have AI coins plummeted by up to 90%?

▲ Artificial Intelligence (AI), semiconductor equipment/AI-generated image ©

While astronomical funds are pouring into the artificial intelligence (AI) industry, AI cryptocurrencies are being left out of the rally. With many major AI tokens remaining at levels 70-90% lower than their all-time highs, investors are rapidly shifting from simple AI themes to computing, infrastructure, and AI agent businesses that generate actual revenue.

According to investment media FXStreet on September 25 (local time), the AI market size is projected to expand from $391 billion in 2025 to $540 billion in 2026, and approximately $3.5 trillion by 2033. In contrast, the total market capitalization of the AI coin market is only about $25 billion. Representative AI coins like Near Protocol (NEAR) are 77% below their all-time high, Bittensor (TAO) is 60% lower, and Internet Computer (ICP) is 99% lower. While CoinGecko classifies 1,473 AI and blockchain-related projects, the industry's growth does not seem to translate into token prices.

Analysis suggests that the destination of funds has changed. In the first quarter of this year, approximately 80% of global venture capital (VC) funds, or $240 billion, were concentrated in AI. In the cryptocurrency sector, companies combining AI and blockchain accounted for 40% of total VC investments, more than doubling from 18% a year ago. Global AI spending is also projected to increase from $1.76 trillion in 2025 to $2.52 trillion this year and $3.34 trillion by 2027, with AI infrastructure expected to account for the largest share.

Interest in AI coins has not disappeared entirely. In the first quarter of this year, AI coins accounted for 35.7% of the cryptocurrency market's interest, surpassing meme coins (27.1%). However, this interest did not translate into actual investment funds. Investors preferred infrastructure that secured actual revenue and usage over speculative tokens, with Bittensor and Render (RNDR) being cited as prime examples. Bittensor recorded $43 million in revenue in Q1 this year based on actual AI usage, and Render also generated approximately $18 million in quarterly revenue through GPU rendering.

Conversely, Virtual Protocol (VIRTUAL) was pointed out as a case where platform growth and token value do not align. While it supports the creation and monetization of AI agents on the Ethereum Layer-2 network Base, the structure allows developers to take all generated revenue, making it difficult for token holders to directly profit. VIRTUAL once exceeded $5, expanding its market capitalization to over $5 billion, but its current market capitalization is around $462 million. The media noted similar discrepancies between token investors and business value in AI16Z (AI16Z), Fartcoin (FARTCOIN), and Gamebuild (GAME).

However, this does not mean that the combination of AI and cryptocurrency is weakening. BlackRock defines AI as 'machine-based intelligence' and digital assets as 'machine-based money,' analyzing that blockchain, stablecoins, and smart contracts can serve as payment and settlement infrastructure in an environment where AI agents trade autonomously. FXStreet reported that in the future AI coin market, actual trading volume, revenue, infrastructure development, and how much profit tokens secure will become more important than simple themes, adding that the speculative-driven investment approach that worked in the 2024-2025 bull market is rapidly losing its power.

*Disclaimer: This article is for investment reference only, and we are not responsible for any investment losses based on it. The content should be interpreted for informational purposes only.*