Tokens linked to decentralised AI infrastructure have surged since NVIDIA's GTC conference, with TAO up 60 percent, FET up 66 percent and NEAR gaining more than 10 percent in a single session.
The combined market capitalisation of artificial intelligence-linked cryptocurrency tokens has reached $28 billion, driven by a sustained rally that began during NVIDIA's GTC developer conference in mid-March and has continued into April as investors rotate out of broad-market positions and into niche infrastructure plays.
NVIDIA CEO Jensen Huang used the conference to project roughly $1 trillion in chip demand backlog through 2027 and to unveil OpenClaw, an open-source framework for training robotic AI agents. The announcements triggered immediate buying across AI-adjacent tokens. Bittensor's TAO surged 60 percent in a single week. The Artificial Superintelligence Alliance's FET token rose 66 percent over the same period. NEAR Protocol climbed more than 10 percent in a single trading session, reaching its strongest level since late January.
Render Network's RNDR token, which powers a decentralised GPU marketplace connecting idle graphics cards with users who need compute capacity for 3D rendering and generative AI workloads, has tracked closely with the broader AI token rally. The network processes rendering jobs at a fraction of centralised cloud pricing, a value proposition that has gained traction as AI model training costs have risen sharply over the past 18 months.
The sector's performance contrasts with the stagnation in major tokens. Bitcoin has traded in a range between $62,000 and $75,000 since early February, and Ethereum has hovered around $2,100 for weeks. The divergence suggests that capital is flowing into tokens with specific AI-related utility rather than into the market as a broad risk-on trade.
Decentralised AI infrastructure projects differ from traditional cloud computing in that they aggregate underutilised hardware from a distributed network of providers rather than relying on centralised data centres operated by a single company. Bittensor operates a network of subnets, each dedicated to a specific AI task such as text generation, image recognition or financial prediction. Validators stake TAO tokens to evaluate the quality of machine-learning models submitted by miners, and rewards flow to those who produce the most useful outputs.
The FET token underpins the Artificial Superintelligence Alliance, a merger of Fetch.ai, SingularityNET and Ocean Protocol that completed in mid-2024. The combined entity aims to build a decentralised AI platform spanning data marketplaces, autonomous agents and machine-learning model hosting. Its token reached a 52-week high during the post-GTC rally before pulling back modestly in early April.
Institutional interest in AI tokens remains limited compared with Bitcoin and Ethereum, but venture capital activity in the subsector has accelerated. PitchBook data show that AI-crypto crossover deals attracted $1.4 billion in the first quarter of 2026, up from $620 million in the same period a year earlier. Notable rounds included a $200 million Series B for Ritual, a project building confidential AI inference on blockchain, and a $150 million raise for Gensyn, which operates a decentralised compute network for model training.
The rally has not been uniform across all AI tokens. Smaller projects without meaningful network usage or revenue have lagged, and some have declined. Analysts at Messari noted in an April research report that only 12 of the roughly 180 tokens tagged as AI-related on CoinGecko had generated more than $1 million in annualised protocol revenue, suggesting that much of the sector's market cap is driven by narrative rather than fundamentals.
NVIDIA's own share price rose approximately 1.5 percent during the GTC conference and has held those gains. The company's projection of $1 trillion in chip demand through 2027 was framed around enterprise and sovereign AI spending, not crypto, but the knock-on effect on decentralised GPU and compute tokens was immediate.
The AI token sector now accounts for roughly 1.15 percent of the total cryptocurrency market capitalisation of $2.43 trillion. That share has more than doubled since the start of the year, when AI tokens collectively held a market cap of approximately $11 billion.
Trading volume in AI tokens on decentralised exchanges has also risen. Uniswap and Raydium processed a combined $1.8 billion in AI token swaps during March, according to data from DeFiLlama, up from $740 million in February.
Fetch.ai's FET token closed Monday's session at $2.47, down from its post-GTC peak of $2.89 but still more than double its January price of $1.12.
