9fin.

Every leveraged buyout, high-yield bond and syndicated loan arrives wrapped in the same thing: a document. Often hundreds of pages of covenants, definitions and financial disclosures, written by lawyers, buried in PDFs. For decades, the job of a credit analyst was, in large part, to read those documents and copy the numbers that mattered into a spreadsheet.

9fin was built on the unglamorous observation that this work is enormous, repetitive, and ripe for automation. Founded in London in 2016 by Steven Hunter, a former J.P. Morgan credit banker, and Huss El-Sheikh, a former Deutsche Bank engineer, the company set out to fix a market that still ran on manual data entry and fragmented sources. Its answer was machine learning that extracts structured data directly from financial documents - and, on top of that data, a platform of news, analytics and AI-powered workflows.

The pitch has aged well. In December 2024 the company raised a $50M Series B led by Highland Europe. Fifteen months later, in March 2026, it raised a $170M Series C led by HarbourVest with Canada Pension Plan Investment Board, at a $1.3 billion valuation - unicorn status, and more than $250M raised in total.

The incumbents in financial data are broad: Bloomberg, LSEG, S&P Global, Moody's. 9fin's wager was the opposite - go narrow and deep on the one corner everyone else found too messy, leveraged finance, and own it before expanding. Rather than compete on breadth, it built a proprietary data layer by extracting information straight from documents, then layered AI on top.

That matters because generic AI struggles here. The data isn't on the open web; it's specialised, opaque and high-stakes. A large model trained on the internet has never seen most credit agreements. 9fin's answer is to own the data first and make the AI reliable second.

The team - self-described "9finners," roughly 350 to 430 people - is unusual for a data company: financial journalists breaking news sit alongside machine-learning engineers building extraction pipelines and credit analysts validating the output. That blend of journalism, domain expertise and engineering is the company's real moat as much as any single model.