STMicroelectronics’ revenue beat is not the recovery, it is the inventory-cost trough hiding in plain sight

The actionable read from this event is that STMicroelectronics is no longer an auto-led growth story with clean operating leverage, but it is also not the same earnings knife-catch it was at the Q1 FY2025 trough. What was priced in was a weak EPS quarter, continued Automotive pressure, and skepticism that the prior quarter’s revenue rebound was durable after revenue fell from $4,282.0 million in Q4 FY2023 to $3,322.0 million in Q4 FY2024 and then to $2,519.9 million in Q1 FY2025. What actually surprised was cleaner top-line delivery against the Street, with THE PRINT showing revenue of $3,329.0 million versus $3,282.7 million, a +1.4% surprise, while EPS was the negative surprise at $0.11 versus $0.27, a -59.3% surprise. The market should separate those two facts rather than average them into a muddled “mixed quarter” call: demand and mix were better than feared, but below-the-line and restructuring/tax/capacity effects still dominate the income statement. That distinction matters because the stock’s next move should be decided less by the reported EPS miss and more by whether the company can keep revenue near the $3.04 billion Q1 guide while unused capacity charges remain a quantifiable 220 basis points rather than widening.

That distinction is visible in the company’s own numbers, even though the street-comparison print and the company’s reported-basis commentary do not match exactly and should not be blended. On the company’s call basis, Jean-Marc Chery said, “We delivered revenues at $3.33 billion, above the midpoint of our business outlook range, driven by higher revenues in Personal Electronics and to a lesser extent in Communication Equipment and Computer Peripheral and Industrial, while Automotive was below expectations.” The wording matters because it names the rotation: the quarter did not beat because Automotive healed, it beat despite Automotive. In the quarterly history, revenue moved from $2,519.9 million in Q1 FY2025 to $2,802.4 million in Q2 FY2025, then $3,173.3 million in Q3 FY2025, and $3,326.5 million in Q4 FY2025, while gross margin moved only from 33.4% to 33.0%, then 33.2%, and finally 35.2%. That sequence is not normal cyclical operating leverage yet; it is revenue recovery plus mix relief against still-heavy factory absorption. EPS confirms the drag: diluted EPS went from $0.06 in Q1 FY2025 to -$0.11 in Q2 FY2025, $0.26 in Q3 FY2025, and -$0.03 in Q4 FY2025, even as revenue rose each quarter from Q1 through Q4.

The revenue chart should pull the eye to the fact that Q4 FY2025 gross margin of 35.2% is the first clear margin step-up after three quarters stuck near 33.0%, but it should not be mistaken for a return to the 2023 model. Gross margin was 49.7% in Q1 FY2023, 49.0% in Q2 FY2023, 47.6% in Q3 FY2023, and 45.5% in Q4 FY2023, then fell to 41.7% in Q1 FY2024 and 37.7% by Q4 FY2024 before bottoming around 33.0% in Q2 FY2025 and 33.2% in Q3 FY2025. The market may be underpricing the recovery in mix, but it is not underpricing a return to peak profitability because the data do not support that. Chery’s line that “Gross margin of 35.2% was also above the midpoint of our business outlook range, mainly due to better product mix” is important precisely because it attributes the upside to mix, not utilization. A mix-led 35.2% gross margin is investable if it repeats through the Q1 seasonal revenue decline; it is not enough to underwrite a 45.5% to 49.7% margin regime without evidence of absorption improvement.

The capacity story explains the margin guide, because Q1 FY2026 revenue is guided lower while fixed-cost absorption is still visible in basis points rather than hidden in adjectives. Chery said the company is “expecting Q1 '26 revenues at $3.04 billion, a decrease of 8.7% sequentially, plus or minus 350 basis points,” and separately guided gross margin “to be about 33.7%, plus or minus 200 basis points, including about 220 basis points of unused capacity charges.” This is the center of the thesis. A sequential revenue decline after Q4 is not bullish in isolation, but a 33.7% margin guide that explicitly absorbs about 220 basis points of unused capacity charges tells investors where the earnings sensitivity is. If revenue is merely seasonally weaker and underutilization is contained, Q4’s EPS disappointment becomes less informative than the margin bridge into Q1. If unused capacity charges persist at about 220 basis points or worse while revenue does not reaccelerate after Q1, then the company is still paying for an asset base sized for a demand curve that has not returned.

The EPS miss looks severe on THE PRINT, but the call gives enough one-off and operating-cost detail to argue that the miss is not the whole economic story. Lorenzo Grandi said fourth-quarter gross profit was $1.17 billion, down 6.5% year over year, and total net operating expenses excluding restructuring were $906 million, slightly higher year over year due to currency. He also said fourth-quarter operating income was $125 million and included $141 million for impairment, restructuring charges and other related phaseout costs. The company’s own reported-basis figures then complicate the EPS comparison further: Grandi said fourth-quarter 2025 net loss was $30 million including certain onetime noncash income tax expenses of $163 million, and diluted EPS was negative $0.03 compared to $0.37 of last year; elsewhere in the same call excerpts, he also said GAAP diluted earnings per share stood at $0.11 including certain negative onetime tax expenses impacting of $0.8 per share. These numbers conflict because the data pack includes different reporting bases and excerpted statements, so the right analytical move is not to normalize the quarter by inventing an adjusted EPS, but to treat operating leverage as impaired until the $906 million quarterly OpEx base and 220 basis points of unused capacity charges stop overwhelming the revenue recovery.

That operating-leverage caution also fits the full-year reset, where the company is spending through a downturn rather than harvesting cash from a lean model. Chery said full-year net revenues decreased 11.1% to $11.8 billion, with a strong decrease in Automotive and a lesser decrease in Industrial, while Personal Electronics and Communication Equipment and Computer Peripheral grew. Full-year gross margin was 33.9%, down from 39.3% in full year 2024, and the company invested $1.79 billion in net CapEx while generating free cash flow of $265 million. For Q4 specifically, management said it generated positive $257 million free cash flow and worked down inventories on both the balance sheet and in distribution. Inventory nevertheless ended the year at $3.14 billion, down from $3.17 billion at the end of Q3 but above $2.79 billion one year ago. That is the clearest balance-sheet tension in the print: sequential inventory clean-up is real, but the year-on-year inventory position says the company still has not fully digested the prior demand shock.

The forward CapEx plan makes the inventory and utilization debate more consequential because STMicroelectronics is not cutting itself into near-term EPS optics. Chery guided 2026 net CapEx of about $2.2 billion to support capacity additions for selected growth drivers including cloud optical interconnect and the manufacturing reshaping plan. That is higher than the $1.79 billion invested in net CapEx in the year just reported, while Q1 FY2026 gross margin is guided to about 33.7% with about 220 basis points of unused capacity charges. The variant perception here is not that CapEx discipline will rescue earnings. It is that management is deliberately carrying underutilization in mature and reshaped manufacturing while funding targeted areas where it sees a non-Automotive revenue leg. The proof point embedded in the call is explicit: in communication equipment and computer peripheral, data centers including cloud optical interconnect, and power and analog for AI servers and data centers, management believes it can deliver $1 billion revenue before 2030 with already USD 500 million in 2026. That gives the market a tangible offset to the Automotive reset, but it also creates a hard scorecard: the $2.2 billion CapEx plan must show up in that USD 500 million in 2026, not merely in strategic language.

The customer and supplier read-through is therefore narrow but investable: equipment and materials suppliers tied to STMicroelectronics’s reshaping and selected capacity additions have better signal than broad auto semi demand from this print. Aehr Test, as a wafer-level burn-in and test systems supplier, gets a directional support point from the planned about $2.2 billion 2026 net CapEx and from continued investment in selected growth drivers, but the negative 30.2% operating margin in Power and Discrete means SiC-linked demand should not be read as universally healthy. Wolfspeed, supplying 150mm SiC bare and epitaxial wafers under a >$800M agreement, faces that same mixed signal: long-term substrate commitments remain relevant, but Power and Discrete negative 30.2% says current profitability is under pressure. Tower Semiconductor, through the TPSCo JV with Tower, ST, and NXP in Japan, is more insulated by the structural manufacturing relationship, while Soitec’s FD-SOI and RF-SOI exposure lines up better with automotive MCUs, IoT, and RF-SOI than with the quarter’s Personal Electronics upside. The named suppliers with process-equipment exposure, Centrotherm, E&R Engineering Corp., Oxford Instruments, and Veolia Water Technologies, should care less about Q4 EPS and more about whether the about $2.2 billion 2026 CapEx budget survives the Q1 revenue decline to $3.04 billion.

The peer lens reinforces why this is a self-help and mix-recovery thesis rather than a sector-wide margin thesis. In the latest peer table, STM shows $3,095.0 million revenue, 33.8% gross margin, and +22.8% revenue YoY, which makes its growth rate optically higher than Texas Instruments at $4,825.0 million revenue, 58.0% gross margin, and +18.6% revenue YoY, and higher than NXP at $3,181.0 million revenue, 56.2% gross margin, and +12.2% revenue YoY. But the margin gap is the point: STM’s 33.8% gross margin sits far below Texas Instruments’ 58.0% and NXP’s 56.2%, and even below IFNNY’s 38.7% and Intel’s 39.4% in the peer table. A PM should not pay for STM as if a +22.8% revenue YoY line implies peer-level gross margin conversion. The mispricing opportunity, if there is one, is that the market may over-penalize the -59.3% EPS surprise and miss the stabilization in revenue and mix; it is not that STM has suddenly matched the analog margin structure of Texas Instruments or NXP.

The tone work supports that reading, but it also tells us management delivery became more forward-leaning while uncertainty rose, which is exactly the setup for a thesis that must be confirmed quickly. The tone history shows sentiment improving from 0.07 in Q4 FY2025 to 0.24 in Q1 FY2026, and guidance_tone rising from 0.28 to 0.54. Q&A sentiment also improved from 0.09 to 0.21. Those are meaningful changes, but not cleanly risk-free, because tone_confidence fell from 0.80 to 0.68 and uncertainty increased from 29.6 to 39.0. The call therefore sounded more constructive, but with less confidence and more uncertainty. That combination fits the numbers: management can point to Q4 revenue of $3.33 billion, Q4 gross margin of 35.2%, positive $257 million Q4 free cash flow, and a cloud/data-center path to USD 500 million in 2026, while still guiding Q1 revenue down to $3.04 billion and gross margin down to about 33.7%.

The most important implication of the tone series is that the call’s optimism was concentrated in guidance and Q&A rather than prepared-script sentiment. Prepared_sentiment was 0.02 in Q4 FY2025 and 0.01 in Q1 FY2026, while qa_sentiment moved from 0.09 to 0.21 and guidance_tone from 0.28 to 0.54. That is not cosmetic. Prepared remarks tend to anchor on reported facts, and those facts remain messy: Q4 FY2025 diluted EPS was -$0.03 in the quarterly history, full-year operating income stood at $175 million compared to $1.68 billion in 2024, and Q4 net operating expenses excluding restructuring were $906 million. The Q&A and guidance tone improved because the forward debate has moved to inventory normalization, cloud optical interconnect, and underutilization absorption. The risk is that uncertainty rising by +9.4 call over call means management’s confidence in the path is not as high as the guidance_tone jump alone would suggest.

The stock debate into next quarter should be framed around three numerical gates rather than a generic recovery narrative. First, Q1 revenue needs to land against the $3.04 billion guide and the plus or minus 350 basis points range without reintroducing the kind of sequential demand break seen in Q1 FY2025, when revenue was $2,519.9 million and revenue QoQ was -24.1%. Second, gross margin needs to hold near about 33.7%, plus or minus 200 basis points, and the unused capacity charge must not move materially worse than about 220 basis points, because that is the bridge between a mix recovery and a still-broken factory model. Third, inventory must continue down from $3.14 billion after already moving from $3.17 billion at the end of Q3, because a reversal would undermine management’s claim that it worked down inventories both on balance sheet and in distribution. The thesis breaks if Q1 revenue misses the $3.04 billion guide, if gross margin falls below the guided band around about 33.7%, or if the $2.2 billion 2026 CapEx plan starts to look like capacity being added ahead of demand rather than funding the USD 500 million 2026 cloud/data-center opportunity. It is confirmed if Q1 preserves the 33.7% margin framework despite the sequential revenue decline, inventory keeps falling from $3.14 billion, and management can attach real orders or revenue conversion to the $1 billion-before-2030 cloud/data-center target when it next reports.