Cerence Inc. (CRNC) Cash Flow Statement

The $278.0M and $313.5M losses in 2024Q2 and 2024Q3 sit alongside $1.0M and $12.9M of positive operating cash flow, which appears consistent with large non-cash impairment-type charges that never consumed cash, though the composition warrants closer inspection in the filings. Even excluding those distorted quarters, the pattern of negative net income paired with positive operating cash flow persisted through 2025Q4, implying that reported profitability is a poor proxy for cash generation here and that investors should lean on the cash statement rather than the income statement. The 2026 quarters show the reverse tension: net income finally turned near-breakeven-positive while the OCF/NI ratio remained wildly elevated at 46.9x in 2026Q3, pointing to accrual reversals rather than a genuine step-change in earnings quality.

Stripping the 2026Q3 outlier, quarterly FCF has climbed from $4-12M in 2024 to a $10-36M band across 2025Q3 through 2026Q2, which does look like a genuine improvement trajectory against an 8.5% margin in 2024Q4. However, that improvement closely tracks the working capital contribution rather than operating income, which has hovered near breakeven in line with the earlier finding of an R&D cost base that limits operating leverage. The durability of the trajectory therefore appears contingent on collections continuing to normalize, and a reversion in the cash conversion cycle would likely compress FCF meaningfully even if revenue holds the recent $65-70M run rate.

Nine of the ten quarters posted a positive working capital contribution, which for a software business of this size is unusual and may indicate either elongated contract prepayment terms or a one-time unwind of receivables built up during the revenue troughs of 2024-2025. The 2024Q4 print of $21.8M and the 2025Q3 print of $20.1M are large relative to single-quarter revenue of roughly $50-70M, so the cash flow statement appears to be obscuring how much of the balance represents genuinely collected cash versus deferred revenue timing. The lone negative print in 2025Q2 coincided with the strongest net income quarter of the window ($21.7M), an inverse pattern that further suggests the working capital line is decoupled from underlying operating performance.

The capital intensity is consistent with an asset-light software model, and the elevated 12.3% CapEx/Rev print in 2025Q3 alongside the $7.7M quarterly spend appears to be an isolated project rather than the start of an investment cycle, given that subsequent quarters reverted to 0.8-5.0%. With D&A running modestly ahead of CapEx, the asset base is shrinking in book terms, which may be rational if legacy licensed-intellectual-property balances are running off as the business shifts model, but it would be a warning sign if it coincides with declining R&D effectiveness. Given the earlier finding that R&D runs at 42-48% of revenue, the very low tangible CapEx suggests essentially all of that spend is being expensed rather than capitalized, which flatters near-term FCF at the expense of long-term asset depth.

The near-total absence of deployment alongside consistent FCF generation is a notable posture, particularly given that the buybacks that did occur (a $7.5M repurchase in 2026Q1) came in the same quarter the company posted a $5.2M net loss, suggesting opportunistic rather than programmatic capital allocation. There are no acquisition outflows in any of the ten quarters despite a market context where voice-AI peers such as SoundHound are consolidating, which may indicate Cerence lacks conviction in inorganic growth or is preserving optionality. Investors should monitor whether the accumulated cash is eventually deployed into restructuring, M&A, or a buyback program, since idle cash drag on ROIC would compound the already-thin operating margins identified in the income statement analysis.