Case 01 · Equity Research
A company lifted by the AI narrative — but how profitable is the core business, really? We traced the story back to the three statements, and the conclusion differed from the market consensus at the time.
In 2025 iFlytek had a domestic large-model narrative on one side and a profit statement under pressure on the other. The market told stories about revenue growth and AI orders, but rarely answered a more basic question: does the core business actually make money?
Our task was to produce, in limited time, a report that could support an investment judgment — not a compilation of materials.
PEST and Porter's five forces set the industry boundary; segment-level breakdown and the three DuPont factors handled the company; revenue drivers built the three statements bottom-up, and sensitivity analysis tested the key assumptions.
Data came from annual reports, Wind and public order information. Every key assumption is listed separately in the report so readers can change the inputs and re-run it.
Once profit is taken apart, iFlytek's earnings quality looks more fragile than the headline: government subsidies account for about 78% of profit, and the core business is close to break-even once they are excluded.
The conclusion is not “sell”. It is a reminder: when pricing this company on an AI narrative, strip out the subsidies and look at the core business on its own, or its profitability will be overstated.
The deliverable was a 50+ page report and a reusable three-statement model. The project won a national first prize, but what mattered more was confirming one thing: the value of financial analysis is translating a narrative into numbers.
If I did it again, I would bring in peer comparison earlier and frame “iFlytek's problem” as “a common problem of AI companies” — the conclusion would be more persuasive.