After a recent flash crash under $99,000, Bitcoin (BTC) caused an even deeper drawdown for other assets. Previously hot narratives like AI agents and DeFAI were the hardest hit by the correction.
The AI agent narrative and especially DeFAI were the hardest hit during the recent market correction. As Bitcoin (BTC) dipped under $99,000 briefly, recently hot narratives erased even more value.
The two narratives lost their momentum in January, despite expectations for becoming the leading source of gains in 2025. The two narratives outperformed all assets with a year-end rally in 2024, but the momentum was not sustainable.
The AI token space as a whole had a correction, affecting even legacy AI coins and tokens. The market for infrastructure AI and older shared computing projects lost more than 13% to a market capitalization of $36B.
DeFAI is one of the newest categories of tokens, currently building up its market capitalization. Top DeFAI tokens are led by AIXBT and GRIFFAIN, which dragged down the whole sector. DeFAI lost more than 21% of its market capitalization in 24 hours, down to $2.1B.
Based on other metrics, the sector lost more than 29% of its value, with accelerated selling. DeFAI is still a small-scale sector of tokens, similar to DeSCI, offering both bigger gains and sharper corrections.
The loss of enthusiasm for DeFAI follows general weakness in the AI agent sector. After becoming over-saturated, the market for AI agents could not hold the initial inflows. Most of the value accumulated in several top AI agent tokens, while less liquid assets crashed even faster.
The AI agent space corrected from peak valuations above $17B, down to $9.59B toward the end of January. Agents on Ethereum, Solana, and Base were all mostly in the red, despite aiming to be cult tokens with minimal selling.
AIXBT, the leading token in the space, lost nearly 50% of its value within days, down to $0.50 from a peak close to $1. GRIFFAIN posted a similar loss from $0.60 to $0.30. Most other tokens lost between 20% and 40% in 24 hours.
Even the cult token, FARTCOIN, continued to slide, down to $0.94. FARTCOIN accelerated its slide after the news that Truth Terminal, its AI agent owner, started divesting its holdings through an OTC deal. FARTCOIN was down 25.9% in 24 hours, as the asset continues to see frequent trading from whales.
All current AI agents were created with older LLMs, while the focus was on their social media success and speculative trading. The emergence of the newly trained DeepSeek is also disrupting the AI agent space. Projects are scrambling to add the new model, or completely replace their previous tools for launching agents.
Crypto projects are trying to tokenize the trend, by adding a new batch of agents based on the DeepSeek trend. The earliest risky AI agent tokens based on DeepSeek’s training have already emerged on Solana. SwarmNode has also adapted, allowing all agents to switch to DeepSeek, aiming to re-spark the agent narrative. As usual, Solana builders almost immediately responded with an unofficial DeepSeekAi token on Solana, also a risky bet.
Agent creators on Internet Computer also posted an example of a new AI personality built on ICP, posted directly to the DeepSeek GitHub repository.
Net buying pressure for AIXBT and AI16Z has decreased, as smart money and whales are accumulating other assets.
After boosting AI agents, funds are now flowing back into infrastructure tokens. Virtuals Protocol (VIRTUAL) was the most widely bought asset, with $200K advantage of buyers over sellers.
Instead of tokens, traders also rolled over into cbBTC, Coinbase’s wrapped version of BTC. As the market consolidates and traders switch to safer assets, the AI space now has to show its projects can survive beyond the initial hype.
Traders are also turning back to meme tokens, as AI agents are seen as a similar type of asset. Trading accelerated for Base meme tokens TOSHI, MIGGLES, and MOCHI. Some of the funds flowed back to DeFi tokens like Aerodrome (AERO).
Cryptopolitan Academy: Are You Making These Web3 Resume Mistakes? - Find Out Here