Surf Raises $15 Million to Build AI Platform Specialized for the Crypto World

Exclusive: Surf, an AI platform just for for crypto, raises $15 million | Fortune

Ryan Li has long been steeped in the worlds of both AI and crypto. He started working with AI about ten years ago as an undergrad at UC Berkeley and has since built two crypto startups. That experience has led him to conclude that popular AI platforms like ChatGPT and Perplexity give rise to far too many hallucinations about crypto -- and that those errors can cost traders millions, if not billions, of dollars. That's why he cofounded Surf, an AI platform specialized for the world of crypto. On Wednesday, his startup announced that it raised $15 million from Pantera Capital, with participation from Coinbase Ventures and Digital Currency Group. Li, the startup's CEO, did not disclose the company's valuation in an interview with Fortune. "Surf was something that we wanted to build for ourselves, and it turns out the market loves it," he said. "I think Surf is the best model for crypto, and we want to make sure people trust Surf instead of relying on hallucinations from other models." The product launched as an invite-only service in July before opening to the general public in September. It has more than 300,000 people using the platform and has generated more than a million search results, according to the company. Li says that the platform can answer a spectrum of questions about crypto, from the most basic to the most technical concepts. He says that 80% of the platform's users are regular people who don't work in crypto, and the remaining 20% are from venture capital firms, exchanges, and other crypto-affiliated companies. The startup says that it has a couple of million dollars in revenue and hopes to reach $10 million by the end of 2026. Surf generates revenue by charging a subscription fee, with users paying $15 to $399 per month on different tiers of the service. People can access a free version in which they can only ask a few questions per day. The company currently has just under 30 employees. Li views his startup's competition as mainstream AI models like ChatGPT, Perplexity, and Grok. The company says that its platform performs four times better than ChatGPT and Grok on crypto tasks, based on benchmark testing it co-authored with Princeton University. While ChatGPT may be slightly faster, Surf had answers that were more specific to crypto and had more citations, based on a side-by-side comparison of the two platforms that Li shared with Fortune. "ChatGPT is a really good generalist, but it doesn't know the crypto industry enough," Li said. "Surf is like that expert that worked in the industry for a long time by us training the model just for crypto." With the new funding, the startup plans to launch Surf 2.0, a more advanced model, in February 2026.

Pantera, Coinbase back Surf's $15M push to build crypto-native AI models

The funding will support Surf's next AI model aimed at delivering deeper onchain analysis and automating research for crypto companies and traders. Surf, an AI platform built for digital-asset analysis, raised $15 million in a round led by Pantera Capital with participation from Coinbase Ventures and DCG, to expand its AI models and enterprise tools. The company offers a domain-specific model used by exchanges and research companies to analyze onchain activity, market behavior and sentiment. The funding will go toward Surf 2.0, which will introduce more advanced models, broader proprietary data sets and additional agents designed to handle multi-step analytical tasks. Surf said its platform has seen rapid uptake since its launch in July, generating more than one million research reports and claiming millions in annual recurring revenue, with usage from a large share of major exchanges and research firms. Surf's model uses a multi-agent architecture that evaluates onchain data, social sentiment and token activity, delivering its analysis through a chat interface for research and reducing manual workloads for analysts and traders. Related: How to turn ChatGPT into your personal crypto trading assistant Artificial intelligence and blockchain are increasingly intersecting as more companies develop tools that leverage both technologies. In April, decentralized AI startup Nous Research closed a $50 million Series A round led by Paradigm. The company is developing open-source AI models powered by decentralized infrastructure and uses the Solana blockchain to coordinate and incentivize global participation in training. In May, Catena Labs, led by Circle co-founder Sean Neville, announced it had raised $18 million to develop a bank built around native AI infrastructure. The company said the system will be designed for both AI agents and human contributors, with AI handling day-to-day operations under human supervision. In October, Coinbase introduced "Based Agent," a tool that lets users create an AI agent with an integrated crypto wallet in just a few minutes to perform onchain actions such as trading, swapping, and staking. As crypto and AI continue to converge, the role of human traders may also be shifting. The decentralized exchange Aster is running a "human vs AI" trading showdown, funding up to 100 human traders with $10,000 each to compete against top-performing AI agents Dec. 9-23. Though the competition still has 13 days to go, Team Human was in the lead as of Wednesday, with a return on investment (ROI) of 13.36% compared to Team AI's ROI of 0.54%.