Profits Doubled, but Operating Cash Flow Fell: What Three 2026 Q3 Software Reports Actually Say

Why is net income up but operating cash flow down? Has the core business become more profitable? Not necessarily. The first questions are where the profit came from, whether cash followed it, and whether the headline refers to one quarter or nine months of cumulative results.

Kingsoft Office (688111) provides a useful real-world example. Its January–September 2026 net profit attributable to shareholders grew 123.77% year on year, while operating cash flow fell 13.07%. Look only at July–September, however, and attributable net profit fell 72.30% year on year. Those facts can coexist. Each describes a different accounting dimension or reporting period.

For readers asking “why is net income up but operating cash flow down?” or “why does a profitable company have negative cash flow?”, the starting distinction is accounting: earnings follow accrual recognition, while operating cash flow reflects net operating receipts and payments over the period. A mismatch is a reason to investigate receivables, contract liabilities, investment income and payment timing—not automatic proof of accounting fraud.

Rather than turn that contrast into a simplistic investment verdict, this article checks three companies’ original 2026 third-quarter reports: Kingsoft Office, iFlytek (002230) and Hundsun Technologies (600570). They have different business and revenue mixes, so this is a comparison of financial statement reading methods, not a claim that they are directly interchangeable valuation peers. All three reports state that the quarterly financial statements were unaudited. The evidence is historical, as available on November 10, 2026.

What the three filings show — the same period, different answers

The table uses January 1 through September 30, 2026 for every row. Monetary figures are in CNY millions, rounded to two decimal places, except year-on-year percentages.

January–September 2026	Kingsoft Office	iFlytek	Hundsun
Revenue	5,079.65	18,187.24	3,292.36
Revenue growth	+21.57%	+7.05%	-5.66%
Net income attributable to shareholders	2,637.04	21.89	575.72
Attributable net income growth	+123.77%	+132.83%	+26.49%
Attributable net income excluding non-recurring gains/losses	2,601.79	-763.00	495.40
Net operating cash flow	1,109.70	264.09	-750.87

Sources: the key-financial-data tables in each issuer’s 2026 Q3 report. Operating cash flow is shown consistently on the nine-month basis; a missing quarterly cash-flow comparator has not been replaced with a made-up number.

Even before considering market prices, the numbers challenge three popular shortcuts. Kingsoft’s profit growth far outpaced its revenue growth. iFlytek showed positive attributable net income but a substantial negative figure after excluding non-recurring items. Hundsun showed positive profit and negative net cash from operations. None of those facts alone establishes investment quality. Together they show why an earnings headline cannot replace a reconciliation.

1. Kingsoft Office: nine-month profit surged, but the third quarter fell

A quarterly filing presents two periods side by side: the reporting quarter (July–September) and year to date (January–September). They should never be treated as the same measurement.

Kingsoft Office year-on-year change	Q3 alone, July–September	Nine months, January–September
Revenue	+16.11%	+21.57%
Attributable net income	-72.30%	+123.77%
Income excluding non-recurring items	-72.85%	+126.48%
Net operating cash flow	-13.23%	-13.07%

A nine-month profit-growth headline can conceal a weak latest quarter. Conversely, focusing only on the steep single-quarter drop misses what happened earlier in the year. Both comparisons are valid only when their time bases are stated.

The company’s own explanation matters more than guessing from the percentages. In its report’s section on significant changes, Kingsoft says the increase in nine-month attributable profit was mainly related to higher investment income from associates and joint ventures. It cites lower investment income in that category as an important reason for the weak single-quarter result. This is management’s published explanation, not an independently measured attribution model. See the Kingsoft Office original Q3 report, especially pages 1 and 3.

Investment income can be a legitimate, recurring accounting component in the right circumstances; it is not inherently fabricated or a one-off gain. But recognizing a share of an investee’s earnings through the equity method is not the same economic event as invoicing software customers and collecting cash during the same period. The mechanism should be investigated separately.

One more distinction is essential: net income excluding non-recurring gains and losses does not necessarily equal profit earned solely from core software subscriptions. Whether a particular investment return is excluded depends on the accounting classification and applicable disclosure rules. Kingsoft’s nine-month adjusted-by-non-recurring-items net income rose 126.48%, while the filing explicitly points to associate and joint-venture investment income as a material driver. A strong growth rate in that reported measure is therefore not sufficient proof that recurring subscription profitability grew by 126.48%.

2. Cash conversion: a ratio worth checking, not a verdict

Kingsoft reported CNY 1,109.70 million of nine-month operating cash flow and CNY 2,637.04 million of attributable net income. Dividing the former by the latter gives roughly 42.1%.

It would be tempting to label that number an earnings-quality score. It is not. Consolidated operating cash flow and net income attributable to shareholders differ in scope and accounting mechanics. Reported earnings may include non-cash accruals or investment income, and operations may have timing differences between revenue recognition and cash collection. The ratio is a prompt to investigate, not a universal quality threshold.

A defensible investigation asks which transactions explain the gap:

1. Cash received and paid in operations. How did customer receipts move relative to revenue? How did payments to suppliers, employees and tax authorities change?
2. Receivables and contract liabilities. Was revenue recognized before cash arrived? Were subscription prepayments received before they became revenue?
3. Investment income and distributions. Did recognized profits from associates turn into actual dividends during the period? If so, where were those cash flows classified?
4. Quarterly seasonality. Can tax payments, payroll, bonuses, research expenditure or customer billing cycles explain part of the mismatch?
5. Comparable history and policies. Do the same ratios and drivers persist across comparable prior-year periods? Has the business mix or accounting treatment changed?

Kingsoft’s nine-month revenue increased 21.57% while operating cash flow declined 13.07%. That combination deserves analysis. It does not, by itself, establish that customers failed to pay or that booked revenue lacks substance. In financial research, an unanswered question should stay an unanswered question until the underlying entries can be traced.

3. iFlytek: positive statutory profit, deeply negative adjusted-by-item profit

The original iFlytek 2026 Q3 report provides a different kind of contrast. Its nine-month attributable profit was CNY 21.89 million, up 132.83% year on year, but net income attributable to shareholders after excluding non-recurring gains and losses was negative CNY 763.00 million.

Those two figures do not contradict each other. They describe different treatments of gains and losses. The report lists government grants and gains relating to financial assets among its non-recurring items and discusses sustained research-and-development investment. It would be inaccurate to claim the entire difference came from a single line item without examining the full reconciliation.

Yet nine-month operating cash flow was positive at CNY 264.09 million, up 115.28%. So a negative income figure after excluding non-recurring items can coexist with positive cash generated by operations. Income statements measure profitability under accrual accounting; cash-flow statements track cash receipts and payments over a period. They answer related, but not identical, questions.

For iFlytek, the follow-up research would be how much of the earnings and cash-flow structure is driven by R&D spending, customer collections, government grants, financial asset gains and timing — not merely whether one headline percentage looks impressive.

4. Hundsun: profit grew while revenue fell and cash flow was negative

Hundsun Technologies reported CNY 575.72 million of nine-month attributable net income, up 26.49%, and CNY 495.40 million of net income excluding non-recurring items, up 70.18%. Over the same period, revenue declined 5.66%, and operating cash flow was approximately negative CNY 750.87 million.

The company linked year-to-date profit growth to higher investment income and growth in its non-recurring-exclusion profit measure to investment income from associates. For the operating cash-flow movement, it cited higher tax payments than in the prior-year period. These descriptions appear in the Hundsun 2026 Q3 report, pages 1 and 3. The linked PDF is a public mirror; the issuer’s formal disclosure governs.

The responsible conclusion is not that negative operating cash flow makes the business unsafe. Nor is it that a 70.18% gain in the adjusted-by-item profit measure proves recurring revenue was up. It is that investment income, actual product revenue and cash tax payments can move in different directions. The size and durability of each contribution require separate evidence.

Kingsoft, iFlytek and Hundsun each make the same methodological point in a different way: statements should be read as connected tables, not as isolated percentages.

A five-step financial statement checklist

A practical workflow begins with the question being asked and ends with an auditable source trail.

Step 1 — Fix the period and units. Label each number as either the July–September quarter or January–September year to date. Confirm CNY yuan, millions or billions, and whether you are looking at consolidated figures, parent-company figures or shareholder-attributable profit. Never silently substitute one period for another.

Step 2 — Record the four earnings components. Capture revenue, attributable net income, net income excluding non-recurring items, and investment income. Record management’s explanation for material changes separately from your interpretation. Read the non-recurring-items schedule and consider how equity-method income is classified.

Step 3 — Reconcile cash. Capture customer cash collections, operating payments, tax outflows and net operating cash flow. Cross-check receivables, prepayments and contract liabilities. Write down plausible timing differences, but do not turn an explanation into a fact until the filing supports it.

Step 4 — Identify falsifying evidence. If the hypothesis is “subscription earnings are sustainably improving,” what would challenge it? Potential counterevidence includes receivables growing without corresponding collections, a decline in recurring subscription metrics, or reported growth heavily dependent on investment returns. Conversely, seasonal cash outflows that reverse later might weaken an initial negative reading. The companion investment review of linear-thinking traps explains why falsifying evidence should be recorded before projecting any single growth rate into the future.

Step 5 — Keep unresolved items unresolved. Quarterly reports often contain less detailed notes than full-year audited accounts. These three Q3 statements are unaudited. If an item cannot be isolated or verified from the available report, mark it unverified rather than assigning a model-generated value.

For AI-assisted financial reading, the same rule applies: preserve number → source table → page → accounting period → interpretation. In an AI financial statement analysis or AI Agent financial-report verification workflow, a model can help locate operating cash flow, earnings adjustments or equity-method investment income, but its citations must lead back to the actual filing; an unverified generated explanation is not an audit. A clean-looking AI summary with no traceable evidence is not an audit. XBSTACK’s seven-step financial report workflow and guide to building a financial-audit agent describe complementary ways to structure that verification. They do not imply that the public AI Finance tool has automatically processed or independently validated these three issuer reports.

What I would check in the next filing

Kingsoft’s 2026 nine-month profit increase and cash-flow decline are both real reported results. Its weak third quarter is also real. Management’s investment-income explanation helps reconcile the apparent conflict, but it does not remove the need to verify subsequent operating performance and cash receipts. The iFlytek and Hundsun examples reinforce why the answer cannot be reduced to a single “quality” ratio.

Three questions should survive any headline: What generated the profit? What generated the cash? What evidence in the next reporting period would disprove today’s interpretation? Valuation, investment horizon, competition, execution risk and market price require additional work. These filings alone cannot justify a buy-or-sell conclusion.

Original source reports and scope

• Kingsoft Office, 2026 Third-Quarter Report, published October 28, 13 pages — key financial data, explanation of significant changes, consolidated statements.
• iFlytek, 2026 Third-Quarter Report, published October 20 — financial-data table, non-recurring items, management notes.
• Hundsun Technologies, 2026 Third-Quarter Report, published October 31, 13 pages — key data and changes explanation. Public mirror; verify against the official issuer filing if using figures in a regulated setting.

This is a public-source study of how to read financial reports, not a description of personal holdings, an endorsement of any security, a return forecast or investment advice. All financial figures are historical, for the nine months ending September 30, 2026, unless explicitly marked as third-quarter figures.

Tool / Compound calculator

Test the assumptions behind the investing review

Use explicit starting capital, recurring contributions, return, time and inflation assumptions. The calculator supports planning and does not provide investment advice.

Open calculator → Continue investing reviews About Xiaobai