The institutional bridge to decentralized intelligence

Grayscale Investments has spent the last decade operating as the primary friction-reducer between legacy financial institutions and the volatile world of digital assets. Founded in 2013 as a subsidiary of Digital Currency Group, the firm is best known for its Bitcoin Trust (GBTC), which for years was the only way for Wall Street to gain exposure to Bitcoin without the complexities of managing private keys. Their victory against the SEC in 2023 essentially forced the agency’s hand on spot Bitcoin ETFs, a move that solidified the firm's status as a market-making force.

In July 2024, Grayscale expanded its focus toward the AI agent economy with the launch of the Grayscale Decentralized AI Fund. This move represents a bet on the decentralized AI (DeAI) stack as the necessary foundation for autonomous agents. While centralized players like OpenAI and Anthropic control the proprietary models that dominate the current market, Grayscale is backing the permissionless infrastructure these agents might eventually require to remain truly independent.

Investing in the DeAI stack

The Decentralized AI Fund focuses on three main categories: decentralized compute, data services, and AI protocols. Its initial basket includes Bittensor (TAO), Near (NEAR), Render (RNDR), Filecoin (FIL), and Livepeer (LPT). Each of these assets represents a different layer of the agent-enabling stack. Bittensor acts as a decentralized intelligence layer where subnets compete to provide the best model outputs, while Render and Filecoin provide the hardware—GPUs and storage—required to run them.

The inclusion of Bittensor is particularly telling for the agent ecosystem. Bittensor functions as a decentralized marketplace for machine intelligence, where miners are rewarded in TAO tokens for providing high-quality model outputs. For those building agents, this offers a glimpse of a world where model access is not governed by a single corporate API, but by a global, competitive network. Similarly, Render’s decentralized GPU network provides the raw compute power needed for inference, potentially lowering the costs for agent developers who cannot afford the premium on centralized cloud clusters.

Market position and leadership

The company operates out of Stamford, Connecticut, under the leadership of CEO Peter Mintzberg. Their model remains focused on management fees and the trust structure, though the conversion of several products to ETFs has changed their fee dynamics significantly. While the firm faces competition from traditional finance giants like BlackRock and Fidelity, Grayscale retains an advantage in its deep expertise in crypto-native protocols that larger firms often overlook. The challenge for Grayscale is justifying higher management fees compared to new low-cost ETFs, but its focus on thematic funds like the AI-centric one provides a specialized value proposition that generic products lack. This shift suggests that Grayscale sees AI and crypto as inherently symbiotic, where AI agents require internet-native programmable money to function autonomously.