Now Reading: Gradient Network Raises $10M From Pantera and Multicoin to Build a Fully Decentralized AI Runtime
-
01
Gradient Network Raises $10M From Pantera and Multicoin to Build a Fully Decentralized AI Runtime
Gradient Network, a crypto-native startup aiming to radically decentralize artificial intelligence (AI) infrastructure, has raised $10 million in seed funding to build what it calls the world’s first “fully decentralized AI runtime.” The round was led by Pantera Capital and Multicoin Capital, with participation from HSG (formerly Sequoia Capital China) and other strategic investors across the AI and crypto sectors.
The funding arrives as Gradient unveils its two foundational protocols—Lattica, a peer-to-peer data delivery layer, and Parallax, a decentralized inference engine capable of scaling large AI models across heterogeneous compute networks.
According to the team, Gradient’s goal is to shift AI away from centralized monopolies and toward a sovereign, peer-powered future—a response to growing concerns around surveillance, platform dependency, and AI bias.
“Centralized intelligence is giving way to open, agentic systems,” Gradient said in its announcement. “What comes next is hosted, served, and owned by the people.”
Rebuilding AI From the Ground Up
Gradient is building an open-source AI infrastructure stack from the ground up. The team identifies three core primitives: compute, communication, and orchestration.
- Lattica is designed as a universal data motion engine, forming the connective tissue for decentralized computing. Built on learnings from a massive peer-to-peer CDN experiment involving millions of nodes, it aims to enable fast, efficient, censorship-resistant data exchange across a global network.
- Parallax, launching this week, is the protocol’s answer to the increasingly urgent need for scalable, privacy-respecting AI inference. Unlike local endpoint models, Parallax breaks large models into orchestrated segments, distributing them across a global mesh of nodes to enable verifiable, fault-tolerant execution at scale.
Together, the two protocols aim to lay the groundwork for a new machine internet—one not controlled by a few data monopolies, but distributed across community-owned infrastructure.
A Push Against Centralized AI Power
As AI becomes more embedded into daily life, questions about privacy, data sovereignty, and platform control are reaching a boiling point. Gradient’s thesis is clear: AI must be decentralized to remain democratic.
Rather than relying on closed APIs or centralized model hubs, Gradient envisions an ecosystem where developers and users can collectively govern the data, infrastructure, and model lifecycles. This approach, the team says, doesn’t just protect against abuse—it fosters innovation by removing gatekeepers.
The announcement also marks a brand refresh that reflects Gradient’s evolving identity, from a research-heavy team into a protocol-building network shaping the future of verifiable AI infrastructure.
While many competitors in the space focus on decentralized storage or inference alone, Gradient’s full-stack approach—bridging data, compute, and orchestration—may set it apart in a quickly growing space that includes projects like Gensyn, Bittensor, and Morpheus.
What’s Next?
Lattica and Parallax will both launch this week, with more protocols planned later in 2025. Gradient is actively inviting developers, researchers, and node operators to join what it calls an “expedition to decentralize the world’s intelligence.”
Editorial Note: This news article has been written with assistance from AI. Edited & fact-checked by the Editorial Team.
Interested in advertising with CIM? Talk to us!
