Cirrus beat the smartphone ramp, but the real misprice is the cash conversion into a seasonal downshift

The market went into this print prepared for a seasonal December peak, not for a simultaneous revenue beat, EPS beat, margin hold, and balance-sheet release. What was priced in was a smartphone-led Q3 with revenue near the street’s $534.5 million and EPS near $2.42, followed by the usual March-quarter reset. What actually surprised was the size and quality of the upside: reported street-comparison revenue was $580.6 million versus $534.5 million, a +8.6% surprise, and EPS was $2.97 versus $2.42, a +22.7% surprise. The important distinction is that the beat was not bought with a gross-margin concession. Gross margin was 53.1% in Q3 FY2026, versus 52.5% in Q2 FY2026 and 53.6% in Q3 FY2025, so the quarter landed inside the high end of the company’s recent band while revenue grew +3.5% sequentially and +4.5% year over year. That combination is the crux of the call: the stock can discount a Q4 revenue decline, but it should not discount December as a low-quality pull-forward unless margin or inventory tells the same story. They do not.

That quality point matters because the company’s own language tied the upside to units and mix, not merely backlog timing. CFO Jeff Woolard said Q3 revenue of $580.6 million was “above the top end of our guidance range driven by demand for components shipping into smartphones and a favorable mix of end devices.” The wording matters because “favorable mix” is the bridge between the +8.6% revenue surprise and the +22.7% EPS surprise: if the quarter had been only a unit shipment catch-up, the margin and cash data would be more vulnerable to reversal. Instead, non-GAAP gross profit was $308.2 million at 53.1%, non-GAAP operating income was $175.1 million or 30.2% of revenue, and non-GAAP net income was $156.7 million. The operating leverage was visible even with non-GAAP operating expense at $133 million, up $5.3 million sequentially and up $3.8 million year over year, both tied primarily to employee-related expenses. The company spent more, but the revenue and mix dropped through faster.

The financial trajectory also argues against treating this as a one-quarter spike. Revenue has now moved from $407.3 million in Q1 FY2026 to $561.0 million in Q2 FY2026 and $580.6 million in Q3 FY2026, while gross margin has stayed at 52.6%, 52.5%, and 53.1% over the same three quarters. That is not a straight-line growth story, because Q4 FY2026 history shows $448.5 million, down -22.8% sequentially, but it is a better economics story than the revenue seasonality alone implies. The relevant comparison is not just Q3 FY2026 against Q2 FY2026; it is Q3 FY2026 against the prior December quarter. Revenue was $580.6 million versus $555.7 million in Q3 FY2025, or +4.5% year over year, and diluted EPS was $2.66 versus $2.11. The street-comparison EPS of $2.97 is the number to use for the beat, while the quarterly-history diluted EPS of $2.66 is the number to use for reported period progression. On either basis, the point is the same: the December quarter demonstrated earnings sensitivity above revenue sensitivity.

The Q4 guide is where bears will focus, and it is a legitimate debate because the midpoint cannot be computed here without adding a number not in the pack. The company guided Q4 fiscal 2026 revenue to $410 million to $470 million, GAAP gross margin to 51% to 53%, and non-GAAP operating expense to $124 million to $130 million. That revenue range is below Q3’s $580.6 million, and the quarterly history already shows Q4 FY2026 revenue of $448.5 million with -22.8% sequential change. The market may see that as confirmation that Q3 was the handset high-water mark. The variant perception is narrower and more actionable: the correct question is not whether revenue falls in March, because the company has already framed that, but whether margins and cash stay disciplined inside that fall. The Q4 gross-margin range of 51% to 53% brackets the recent history of 52.6%, 52.5%, and 53.1% through the first three quarters of FY2026, and non-GAAP operating expense guidance of $124 million to $130 million is below the Q3 actual of $133 million. A seasonal revenue decline with OpEx falling and gross margin still guided around the recent band is not the same as a demand air pocket.

The balance sheet reinforces that interpretation because the quarter converted the handset ramp into cash rather than inventory. Cash and investments ended December at $1.08 billion, up $185.9 million from the prior quarter, while inventory ended Q3 at $189.5 million, down from $236.4 million in the prior quarter. Woolard gave the most important cash sentence on the call: “Cash flow from operations was $290.8 million in December, and CapEx was $5.2 million, resulting in a non-GAAP free cash flow margin of 49%.” That wording matters because it ties the revenue beat to cash realization in the same quarter, not to receivables or channel stuffing that would need to unwind in Q4. The trailing twelve-month cash profile is less euphoric, with cash flow from operations of $129.6 million, CapEx of $21.6 million, and a non-GAAP free cash flow margin of 31%, so the company is not suddenly a permanently 49% free-cash-flow-margin asset. But Q3’s inventory drawdown from $236.4 million to $189.5 million is exactly the kind of second-order data point PMs should demand before underwriting a December beat as durable.

That cash conversion also changes the capital-return math at the margin. Cirrus used $70 million in Q3 to repurchase approximately 591,000 shares at an average price of $118.33 and exited with $344.1 million remaining on its share repurchase authorization. The company did not need to choose between balance-sheet build and buybacks: cash and investments still rose by $185.9 million from the prior quarter. For a fabless name with handset concentration, the ability to repurchase stock after a seasonal peak is most valuable when inventory is falling, not rising. The print gives management more flexibility than the Q4 revenue guide alone suggests, because the company enters the March quarter with $1.08 billion in cash and investments and lower inventory than the prior quarter. If the stock trades as though the $410 million to $470 million guide is the whole story, it is ignoring the balance-sheet state in which that guide is being delivered.

The call delivery was consistent with a company that wanted investors to hear near-term discipline and longer-dated content optionality at the same time, but the tone data says management was more positive while also less clean. In the tone history, Q4 FY2026 sentiment rose to 0.40 from 0.27 in Q3 FY2026, guidance_tone rose to 0.48 from 0.28, qa_sentiment rose to 0.43 from 0.25, and ai_optimism rose to 0.90 from 0.26. That is the positive side of the tape. The caution is that tone_confidence fell to 0.36 from 0.44, uncertainty rose to 55.5 from 40.2, and qa_evasiveness rose to 55.8 from 40.4. The numbers conflict in a useful way: management sounded more constructive, especially in Q&A, but the model also detected more uncertainty and evasiveness. That makes the fundamental data more important than the adjectives. The quarter’s hard facts, +8.6% revenue surprise, +22.7% EPS surprise, 53.1% gross margin, $189.5 million inventory, and $1.08 billion cash and investments, carry more weight than the elevated uncertainty index.

The longer-dated content comments are where the call could matter beyond March, although they should not be overcapitalized yet. John Forsyth said, “Us sampling it right now means that that's really gonna be something that we see in calendar '27 and '28.” The value of that quote is not the product detail, which is absent from the data pack, but the timing commitment: sampling now is framed as revenue relevance in calendar '27 and '28, not as a Q4 FY2026 offset. He also said, “If you look out to '29, we think the SAM is, our products, is certainly north of $800 million.” That is a large future opportunity relative to the current quarterly revenue base of $580.6 million, but the investable constraint is timing. The thesis should therefore not be that new products rescue March. It should be that the current business is generating enough gross margin and cash to fund employee-related expense growth while optionality moves through sampling toward calendar '27 and '28 and a SAM north of $800 million by '29. The print bought time for that roadmap.

The supply-chain read-through is unusually constrained because the provided supply-chain table names no customers of CRUS and no suppliers to CRUS. That absence is itself relevant: this print cannot be used, from the data pack, to assign a quantified upside read-through to a named customer or supplier. The only named demand category in the call excerpts is smartphones, where Woolard attributed sequential revenue growth to higher smartphone unit volumes partially offset by a decline in general market sales. The magnitude available is the company-level result: revenue up +3.5% sequentially to $580.6 million, with gross margin at 53.1%, and inventory down to $189.5 million from $236.4 million. The second-order implication is therefore for competing fabless handset-content suppliers rather than a named CRUS customer: handset-exposed analog and mixed-signal content can still produce positive mix economics in a mature smartphone cycle, but only if the supplier can show the same evidence Cirrus showed, namely revenue above street by +8.6%, EPS above street by +22.7%, and inventory moving down rather than up. Without named customers or suppliers in the pack, any more specific chain read-through would be invented.

The peer context makes the same point from another angle: Cirrus is not being paid for a secular AI growth profile, so the debate must be about quality of handset earnings rather than top-line acceleration. In the provided fabless peer table, NVDA posted revenue YoY of +85.2% with gross margin of 74.9%, while Cirrus posted Q3 FY2026 revenue YoY of +4.5% with gross margin of 53.1%. That comparison does not make Cirrus a growth peer; it shows why the market is likely to discount a smartphone name when AI-exposed names print far higher growth and margin. But the relevant mispricing is relative to expectations, not relative to NVDA’s cycle. Cirrus beat revenue expectations by +8.6% and EPS expectations by +22.7%, while keeping gross margin at 53.1% and reducing inventory to $189.5 million. In a market trained to reward only high-growth fabless prints, a lower-growth company that turns a seasonal handset quarter into $290.8 million of operating cash flow can be misread as merely cyclical when it is also self-funding and accretive through repurchases.

The risk to the thesis is clear and numeric: Q4 has to prove that the December beat did not borrow too much from March. Confirmation would be Q4 fiscal 2026 revenue inside or above the $410 million to $470 million range, GAAP gross margin holding within the 51% to 53% guide, and non-GAAP operating expense staying within $124 million to $130 million. A break would be revenue below $410 million, gross margin below 51%, or OpEx above $130 million, because that would turn the Q3 upside into a lower-quality pull-forward with expense creep. The balance-sheet confirmation points are just as important: inventory should not reverse sharply from $189.5 million, and cash and investments should be judged against the $1.08 billion ending December base after the $70 million Q3 buyback and the $344.1 million remaining authorization. On the next call, the date markers are calendar '27 and '28 for sampled products and '29 for the SAM north of $800 million; any slippage in those time frames would reduce the value of the optionality. Until those numbers break, the cleaner interpretation is that the market is over-weighting the seasonal Q4 revenue reset and under-weighting the combination of 53.1% margin, $290.8 million operating cash flow, and inventory discipline delivered in the print.