On February 23, 2026, Based, the innovative decentralized AI compute protocol, announced the successful close of an 11.5 million dollar Series A funding round. The investment was led by the crypto-native venture capital firm Variant, with significant participation from Cyber Fund, 1kx, and a group of high-profile angel investors including former Coinbase CTO Balaji Srinivasan and Ethereum researcher Justin Drake. This latest capital injection brings Based’s total funding to over 15 million dollars and establishes a post-money valuation of approximately 85 million dollars. Founded in late 2024 by a team of distributed systems engineers, Based is building a permissionless marketplace that allows developers to rent high-performance GPU clusters for training large language models (LLMs) without relying on centralized cloud providers like Amazon Web Services or Google Cloud. By leveraging a custom Layer 2 blockchain architecture, the protocol ensures that compute contributors are paid instantly and transparently, while providing cryptographic proofs that the requested training tasks were performed accurately.
The primary mission of Based is to address the persistent “GPU bottleneck” that has plagued the AI industry throughout the mid-2020s. As demand for specialized hardware like NVIDIA’s H100 and B200 chips continues to outstrip supply, smaller AI startups often find themselves priced out of the market or trapped in long-term, restrictive contracts with major cloud providers. Based addresses this by creating a secondary market where underutilized data centers and private mining operations can list their excess capacity for lease. The protocol’s native token, BASE, serves as the medium of exchange and the governance mechanism for setting compute standards and security protocols. CEO and co-founder Sarah Chen noted that the 11.5 million dollars will be utilized to double the engineering team and launch a new “Incentivized Testnet” in March 2026, which aims to onboard over 50,000 global GPUs into the Based ecosystem. This decentralized approach not only lowers the cost of entry for AI innovation but also provides a more resilient and censorship-resistant infrastructure for the development of “sovereign” AI agents.
The Series A funding arrives at a critical moment for Based as it prepares to integrate with the burgeoning “agentic economy” of 2026. A key feature of the protocol’s upcoming mainnet is the “Auto-Lease” function, which allows autonomous AI agents to manage their own compute budgets by interacting directly with the Based marketplace. This allows an AI model to effectively “pay for its own survival” and growth by sourcing the cheapest available training and inference power across a global network of providers. This synergy between decentralized finance and distributed compute is being hailed by investors like Variant’s Li Jin as the “missing piece” of the Web3 stack. However, the project still faces significant technical hurdles, including the challenge of verifying complex training outputs in a trustless environment and maintaining low-latency communication across geographically dispersed nodes. As Based moves toward its full production release, the 11.5 million dollars in fresh capital provides the necessary runway to prove that a decentralized, community-owned AI infrastructure is not only possible but superior to the legacy models of the 20th-century tech giants.
