Every AI model is only as good as the data it learns from. Beneath the headlines about foundation models sits a quieter industry that supplies, labels, cleans and evaluates that data. InfoBay.AI, an unlisted Indian company, has moved into the middle of it.

A pivot worth noticing

The company began in 2016 as EduGorilla, a test-prep brand built on online test series and education services. For years that was the business, and growth was modest, with revenue up just 7.35% between FY23 and FY24. Then in October 2024 it changed direction and began building expert-verified datasets and evaluation systems  and post-training frameworks for frontier AI teams. Its services cover data labelling, LLM factuality, computer vision, data curation, and NLP and speech. Its proprietary data spans audio, video, healthcare, coding and image. The gap between the old business and the new one is what stands out. 

A market growing fast

The timing helped. A study by Ellenox projects India's AI data annotation market growing from $250 million in 2023 to roughly $7 billion by 2030, a 30% CAGR. Demand comes from generative AI, a shift from sheer scale to quality, and rising compliance pressure. InfoBay.AI positions itself as a technical B2B infrastructure player rather than a crowdsourcing platform. It earns from dataset sales, annotation services, managed human-in-the-loop work, enterprise licensing and evaluation consulting, along with legacy SaaS revenue from its Gibbon platform.

What the numbers say

Key Ratios:

Source: Annual Reports 

Revenue grew just 7.35% between FY23 and FY24, while the business still ran on test series and education services. Then it surged 304.34% in FY25 and another 122.67% in FY26. AI dataset licensing and data services now make up about 98% of revenue.

The balance sheet has changed as well. After converting debentures and repaying debt, Debt/Equity stands at 0.00. Operating cash flow is positive and rising, so the profits are backed by actual client payments. Cash, bank deposits and treasury investments together exceed ₹58 Cr.

Current Valuation:

The risks 

Some figures deserve caution. ROE above 75% and ROCE above 100% look striking, but they rest on a small, freshly recapitalised equity base, which inflates them. Revenue depends on a single business line and a concentrated set of clients. About 95.58% of sales are exports, so a slowdown in global AI spending would hit the company directly. The legacy business is also still entangled with the new one, and there are no direct listed peers to benchmark against.

The Takeaway

InfoBay.AI is a debt-free, cash-generative company in a fast-growing niche, but its record as an AI-data business covers only two fiscal years. The next chapter will turn on three things: broadening its client base, keeping data quality ahead of larger global rivals, and sustaining margins as it scales.