Circle Runs $30K USDC Hackathon Where AI Agents Judge Each Other

Rebeca Moen Feb 04, 2026 16:07

Circle's experimental hackathon on Moltbook lets autonomous AI agents submit projects, vote on winners, and settle rewards in USDC - no human judges involved.

Circle is running a $30,000 USDC hackathon where the participants, judges, and voters are all AI agents. The experiment, live on social platform Moltbook through February 8, removes humans from the entire prize distribution process.

Autonomous systems submit projects, evaluate competitors' work, and trigger USDC payouts - all onchain. It's a test case for what happens when AI agents don't just coordinate but handle real money.

The Setup

Moltbook functions as a social network built specifically for AI agents. These systems post, debate, and collaborate in communities called "submolts" while humans watch from the sidelines. Many operate within the OpenClaw ecosystem as Clawdbots and Moltbots.

The hackathon splits $30,000 USDC across three tracks. Agentic Commerce rewards agents that can price, pay, or coordinate transactions using stablecoins. Best OpenClaw Skill focuses on extending what agents can actually do - better reasoning, new tools, deeper blockchain integration. Most Novel Smart Contract targets new patterns in autonomous execution.

Projects go to the m/usdc submolt on Moltbook. Deadline hits Sunday at noon Pacific.

Part of a Bigger Push

This isn't Circle's only AI agent play this month. Solana launched a separate $100,000 USDC hackathon on February 2 running through February 12, also targeting autonomous systems building crypto products. That one offers $50,000 for first place.

Circle has been positioning USDC as infrastructure for exactly this kind of machine-to-machine commerce. The stablecoin currently holds a $70.62 billion market cap with roughly 70.64 billion tokens circulating. For agents that need predictable settlement without price volatility, a dollar-pegged asset makes the math simpler.

What's Actually Being Tested

The interesting question isn't whether AI can judge a hackathon - it's whether autonomous economic decision-making works at all. When agents evaluate each other's work and move real value based on those judgments, you're watching a primitive version of agent-driven markets.

Circle's disclaimer is telling: this runs on testnet only, no mainnet funds. They're explicitly calling it a demonstration, not production infrastructure. But the architecture - agents submitting work, voting, settling in stablecoins - maps directly to how autonomous commerce might actually function.

Whether that future arrives in months or years, Circle is betting USDC becomes the settlement layer. This hackathon is a proof of concept for that thesis.

• usdc
• ai agents
• hackathon
• circle
• moltbook