Simply put
- Gradient has successfully raised $10 million from investors like Pantera, Multicoin, and HSG, while also launching Lattica and Parallax—a framework intended to run AI models on distributed devices instead of relying on centralized servers.
- This system taps into the unused computing power from smartphones, laptops, and IoT devices, utilizing Solana for data and payment coordination.
- While the team believes this method cuts costs and keeps user data local, detractors point out potential delays and complexity as drawbacks.
Gradient Network has completed its $10 million seed funding round, developing a decentralized AI infrastructure stack, with investments led by Pantera Capital, Multicoin Capital, and HSG (previously known as Sequoia Capital China).
The startup, based in Singapore, plans to use the funding to enhance two primary protocols: Lattica and Parallax, which aim to enable the running of AI models across a network of distributed devices, rather than relying solely on data centers. Both protocols are set to launch this week.
“We believe intelligence should serve the public good, not just be a business asset,” stated Eric Yang, co-founder of Gradient Network, in a press release. “This funding round provides us with the boost needed to build a decentralized infrastructure central to AI.”
As the timing aligns with growing criticism of AI companies regarding data privacy and the concentration of computational resources among a few tech giants, Gradient’s strategy involves using untapped processing power from everyday devices to essentially create a crowdsourced supercomputer.
Lattica functions as a peer-to-peer communication protocol for Bitcoin and torrenting, likened to the plumbing that facilitates data transfer between devices without the involvement of a central server. Notably, the network has reported over 1.6 billion connectivity instances across more than 190 regions.
AI decentralization
Parallax addresses the challenge of operating large AI models without traditional data centers. This protocol dissects large language models into smaller components, allowing simultaneous operations across multiple devices. Instead of sending data to companies like OpenAI or Amazon, Parallax processes information locally within the participating devices’ network.
Although some have voiced concerns about the complexity and latency that come with coordinating tasks over numerous devices, Gradient maintains that its decentralized approach could lessen costs compared to conventional cloud computing, while addressing privacy issues. Unlike centralized servers that collect and process user data remotely, Gradient’s system functions closer to data generation points.
Utilizing Solana’s blockchain, Gradient is choosing a platform known for its efficient transaction speeds and affordability compared to other networks. The blockchain oversees the coordination and financial interactions of the devices lending computing power.
This startup is stepping into an evolving landscape as companies seek to diversify their AI infrastructures. Competitors include those creating markets for AI services, and networks like the Superintelligence Alliance and SingularityNet, which explore various blockchain-based initiatives. While Bittensor and Gensyn take different technical paths, they share the goal of promoting distributed computing models.
Gradient has indicated plans to introduce additional protocols beyond Lattica and Parallax, though details remain unspecified. The company has also expressed intentions to publish research and invite developers to contribute.
