THEA raises $8M to build blockchain settlement for AI model billing

THEA, a company that sells behavioral prediction models trained on 35 billion real-world decisions, has raised $8 million to build a blockchain-based settlement layer for AI services. The round, led by Maven11 Capital, Spartan Group, ManifoldTrading, HackVC, and Fisher8 Capital, will fund THEA Network—a coordination and payment infrastructure on Solana designed to handle continuous, cross-border billing for AI model queries.

The company's models analyze behavioral data from what it calls "risk markets"—trading platforms, esports analytics, and prediction markets where users make rapid decisions under financial pressure. THEA claims clients have seen retention increases up to 30% by using its predictive intelligence for customer segmentation, retention optimization, and compliance risk scoring, though these figures are unaudited and no client names are disclosed due to NDAs.

Why it matters

Traditional payment rails weren't built for machine customers. As AI agents gain autonomy to reason, transact, and execute tasks on behalf of humans, they'll need payment systems that don't require bank accounts, invoices, or manual approvals. THEA's bet is that blockchain settlement removes friction for the metered, continuous billing model that AI services demand—particularly when the customer is another AI.

Tokenizing demand, not hardware

Unlike projects that tokenize GPU supply or model ownership, THEA is tokenizing the demand side. Model providers continue to host and operate their own infrastructure off-chain. Only the settlement—the payment for each query—moves on-chain. "In practical terms, it's similar to a marketplace for purpose-built models and AI services," THEA explained in written responses to questions, first reported by Forbes.

The company chose Solana for what it described as a strong developer ecosystem and more than two years of 100% network uptime. THEA Labs operates the prediction models; THEA Network will handle the settlement layer. The company says the network will launch with usage from day one, though it has not disclosed tokenomics, whether it will use a native token or stablecoins, or details of any token sale.

Building for the agent economy

THEA expects AI agents to become autonomous economic actors within 18 months. In that scenario, agents would query specialized models, analyze customer data, and execute tasks according to enterprise KPIs—all without human intervention. The company describes its infrastructure as "architected not only around human demand for AI, but towards a future in which AI itself becomes a customer of purpose-built models."

That vision aligns with broader industry expectations. Animoca Brands co-founder Yat Siu has argued that agents with wallets will become autonomous economic actors, shifting advertising and transactional spend into what he calls an "invocation economy." Traditional financial institutions, as t54 Labs founder Chandler Fung noted, remain human-centric—creating barriers for machine-to-machine payments that blockchain rails are positioned to solve.

The data moat

THEA's competitive advantage lies in its dataset: more than a decade of accumulated behavioral signals from 35 billion real decisions made under genuine financial stakes. The company was formed as a corporate entity in 2024 after years of private operation. CEO Valentin Batura, who spent nearly two decades building quantitative trading systems at BF Corp and Triple Lemniscate Research, said in the funding announcement that "the organizations that can understand those signals fastest gain a significant advantage."

The company's external materials reference "ecosystem applications generating over $100,000,000 annually," though the term is not defined and it's unclear how much of that figure flows to THEA. The $8 million raise from crypto-focused funds sits in contrast to that revenue claim, and the company has not clarified the discrepancy.

Details of the THEA Network launch timeline and token structure remain undisclosed. The information in this article was first reported by Boaz Sobrado for Forbes.