OpenGradient announces $9.5 million in funding to build a verifiable AI compute layer

Backed by a16z crypto, Coinbase Ventures, SV Angel and others, OpenGradient is building a network for open intelligence that can host, run, and validate AI models at scale.

OpenGradient, a verifiable AI compute layer, announced that it has raised a total of $9.5 million in funding to scale its network for open and auditable model execution.

Leading investors include a16z crypto, with participation from Coinbase Ventures, SV Angel, Foresight Ventures, Pragma, SALT, Symbolic Capital, Canonical Crypto, Black Dragon, NEAR, Celestia, and Thanefield Capital. Angel investors include Balaji Srinivasan (former Coinbase CTO), Illia Polosukhin (NEAR co-founder), Sandeep Nailwal (Polygon co-founder), Bruno Faviero (Magna), Daniel Cheung and Ryan Watkins (Syncracy Capital), and Ekram Ahmed (Celestia).

problem

AI is becoming the backbone of software, finance, and autonomous agents, but the infrastructure on which it runs remains opaque. Developers building AI-native applications today face a choice between trusting black-box cloud endpoints or building costly validation layers from scratch. As AI moves from an assistive tool to autonomous execution (executing trades, managing assets, issuing decisions), its opacity becomes a systemic risk.

Also read: AiThority Interview with Glenn Jocher, Ultralytics Founder and CEO

What OpenGradient is building

OpenGradient is a network for open intelligence. A distributed infrastructure network designed to host, run, and validate AI models at scale. Rather than acting as a standalone blockchain, OpenGradient operates as a specialized AI coprocessor. This allows other applications, blockchains, or agents to outsource compute-intensive tasks to specialized GPUs for computation and a dedicated network of trusted execution environment (TEE) nodes. For example, enterprises can run AI workloads such as Sybil detection and content generation on OpenGradient, and clients can query cryptographic proofs from the network and independently verify the results.

The platform consists of three core components.

• Verifiable Inference Network — A dedicated compute layer that runs AI workloads and attaches cryptographic proofs to all inferences, allowing downstream applications to verify exactly which models were run with which inputs and what was returned.
• Decentralized Model Hub — The world’s largest on-chain model repository with over 2,000 models. Authors can publish, monetize, and create open models without intermediaries.
• Developer Tools — SDKs and APIs that provide access to testable inference through a familiar interface, so builders can benefit even if they don’t understand proof systems.

“AI stacks are consolidated around a small number of closed providers, and the applications built on top of them have no way to audit what is running underneath,” said Matthew Wang, co-founder and CEO of OpenGradient. “We are building an open alternative, an infrastructure where models are testable, execution is provable, and developers own the intelligence their products rely on. This funding allows us to expand that vision.”

traction

OpenGradient showed significant early traction ahead of the broader ecosystem launch.

• Over 2 million users across network and adjacent products
• Over 2 million verifiable inferences processed
• Generates over 500,000 cryptographic proofs
• 2,000+ models from 100+ developers on Model Hub
• 6 active revenue streams across platforms

This shift signals a growing demand for AI systems that are not just black boxes but can actually be programmed, measured, and create value.