Spiral is adding Bitcoin payments to Mesh LLM so machines can buy AI compute

Spiral is building Bitcoin payments into Mesh LLM, its peer-to-peer AI inference network. The goal is to let machines buy and sell AI compute from each other without touching traditional payment rails.

As of early October 2026, the payment feature is still under active development. Nothing has shipped yet, but the direction is clear: Spiral wants AI agents to have a wallet, not a corporate expense account.

How the Mesh LLM payment layer is supposed to work

Mesh LLM is a decentralized network that pools idle GPU power from its users. That combined hardware runs open-source AI models collaboratively.

Mesh LLM can run distributed inference on large models exceeding 256GB, the kind of workload that would normally require serious dedicated hardware.

The payment integration runs through the Lightning Network, a layer built on top of Bitcoin for small, fast transactions. GPU contributors would get paid for the compute they lend to the network, and AI agents would be able to pay for inference directly, without an intermediary sitting between buyer and seller.

On October 3, 2026, the team publicly discussed its ongoing progress on Mesh LLM payments and flagged challenges around verifiable compute. If you pay a stranger’s GPU to run your model, how do you know it actually did the work?

From Bitcoin plumbing to AI infrastructure

Spiral was initiated by Block in 2019 to fund and build open-source Bitcoin development. The Lightning Development Kit gives developers building blocks for adding Lightning payments to their own apps. The Mesh LLM payment work builds directly on that existing foundation.

Spiral has also moved into AI infrastructure through the Goose Development Kit, known as GDK, and through Agent Communication Protocol standards.

In September 2026, Lightning was integrated into the x402 payment standard, which is designed specifically for AI agent micropayments. Recent discussions around Mesh LLM have highlighted x402 as a potential route for handling those micropayments. That remains a possibility under discussion rather than a confirmed implementation detail.

What this means for AI agents, GPU owners and Bitcoin

Coverage of the work has so far stayed mostly inside Bitcoin-focused and developer communities. There are no adoption numbers, transaction volumes or usage metrics to point to yet.

The verifiable compute problem is the single biggest thing to monitor. A payment system for anonymous GPU providers only works if buyers can trust they received what they paid for, and the team has openly said this is a challenge.

Other signals worth following include whether x402 ends up powering the micropayment flow and how Spiral ties Mesh LLM into its GDK and Agent Communication Protocol work.