onsemi’s beat is a utilization story, not a demand inflection

The right read on this print is not that onsemi has turned the corner in automotive and industrial demand; it is that the company is managing through a shallow revenue trough while the earnings power remains capped by factory absorption. What was priced in was a stabilization quarter: street revenue sat at $1,452.2 million, EPS at $0.53, and the company needed to show that Q1’s margin shock was not the new run rate. What actually surprised was narrow but important: revenue came in at $1,468.7 million for a +1.1% surprise, while EPS of $0.53 was essentially in line with a -0.4% surprise. That split matters because the top-line beat did not flow through to upside earnings; the print supports a floor thesis on revenue, but not yet a re-rating thesis on margins.

The distinction is clearest in management’s own language, which committed to the recovery being operational rather than demand-led. CEO Hassane S. El-Khoury said the company delivered “Q2 revenue of $1.47 billion, exceeding the midpoint of our guidance and non-GAAP gross margin and EPS of 37.6% and $0.53, respectively.” The quote earns attention because it binds the quarter’s upside to guidance discipline, not a new cycle narrative. Revenue increased only 1.6% over Q1, and automotive revenue still decreased 4% sequentially. The company is no longer falling the way it was in Q1, but the core end market that investors most want to underwrite has not yet resumed leadership.

The financial trajectory argues for patience on the equity if the bull case rests on a quick return to the old margin model. Revenue has been compressed from the 2023 peak to $1,468.7 million, while gross margin has recovered to 37.6% from Q1’s 20.3% dislocation but remains far below the mid-40s level that defined the prior cycle. The most actionable point is that the sequential revenue improvement is too small to absorb the fixed-cost burden quickly. Management’s Q3 guide implies a revenue range of $1.465 billion to $1.565 billion, which brackets Q2 rather than breaking decisively higher. That is not a bad guide, but it is also not the kind of guide that lets investors ignore factory loading.

The capacity story explains the margin guide, because management explicitly put a number on the gap between today’s profitability and its long-term aspiration. CFO Thad Trent said, “In terms of the march to the target of 53%, you've got about 900 basis points of underutilization charges in our Q3 guide.” That is the fulcrum of the call. A 36.5% to 38.5% Q3 non-GAAP gross margin guide can look acceptable against Q2’s 37.6%, but the 900 basis points of underutilization means the company is still earning through a production system sized for a materially higher revenue base. The market may be mispricing the beat as evidence of normalized earnings, when management is effectively saying the margin bridge depends on volume returning or capacity actions biting.

That bridge is further complicated by mix, because the quarter’s sequential growth came from the business investors tend to treat as lower narrative quality than automotive electrification. Power Solutions Group revenue was $698 million and increased 8% quarter-over-quarter, while automotive revenue was $733 million and decreased 4% sequentially. Industrial revenue was $406 million and up 2% sequentially, but that was not enough to make the quarter feel broad-based. The clean variant perception is that onsemi has stabilized before its highest-value demand stories have recovered. If investors pay for automotive and silicon carbide upside now, they are paying ahead of the evidence in the reported segment data.

The segment details also push back on the idea that product-cycle optionality is already visible in the numbers. Analog and Mixed-Signal Group revenue was $556 million, down 2% quarter-over-quarter, and Intelligent Sensing Group revenue was $215 million, down 8% quarter-over-quarter. El-Khoury’s Treo comment, that the design funnel “has more than doubled quarter-over-quarter” toward a “$1 billion revenue target,” is useful because it shows where management wants investors to look beyond the trough. But a doubled funnel is not revenue, and ISG’s 8% sequential decline says current sensing demand is still pressuring the P&L. The market should give some credit for pipeline formation, but not capitalize the $1 billion target as if it has already entered backlog.

The cash story is better than the demand story, and that is why the stock is not simply a short on weak autos. Trent said the company repurchased “an additional $300 million of shares in the second quarter” and has returned “107% of our free cash flow to shareholders on a year-to-date basis.” The commitment matters because it shifts the downside debate: management is using balance sheet capacity and free cash flow to support per-share value while waiting for utilization to recover. Cash and short-term investments were $2.8 billion, with total liquidity of $4 billion, so this is not a capital-constrained trough. But free cash flow was $106 million in Q2, and the year-to-date free cash flow margin was 19% of revenue. Buybacks can cushion the equity, but they do not solve the 900 basis points of underutilization.

Working capital does not yet signal a clean demand snapback either. Inventory increased by $9 million on a dollar basis, while inventory days decreased to 208 days. Distribution inventory was 10.8 weeks versus 10.1 weeks in Q1, still within the company’s 9 to 11 weeks target range. The conflict is important: fewer inventory days suggest some progress inside the company, but higher distribution weeks say the channel is not tightening aggressively. That is exactly the kind of mixed signal investors should expect near a bottom, and it argues against treating the Q3 revenue range as conservative without corroboration from channel depletion.

The call delivery reinforces that management has become more constructive on the forward guide while remaining less expansive in prepared optimism, which is a useful tell for PMs reading the transcript rather than just the tables. In the tone history, Q2 FY2025 guidance_tone rose to 0.35 from 0.21 in Q1 FY2025, while ai_optimism fell to 0.44 from 0.70. That combination fits the numbers: management can guide with more confidence because inventories and utilization are being actively managed, but the broader demand language does not support a sweeping upcycle claim. The Q2 tone_confidence score of 0.52 also improved from 0.35, suggesting the call was more anchored in what management can control than in speculative recovery timing.

That tone read matters because the Q&A and guidance posture later became more positive in the tone series, but this event itself was still an inflection in control rather than a demand release. The Q2 FY2025 uncertainty score was 51.1, only modestly below Q1 FY2025 at 52.4, while qa_evasiveness moved to 5.4 from -2.1. Those figures are not alarming on their own, but they warn against over-interpreting management’s better guidance tone as a clean visibility upgrade. The print’s message is disciplined, not exuberant: costs are being cut, capex is being contained, and depreciation relief is coming, but customers have not yet pulled automotive back into growth.

The supplier read-through is therefore narrower than a simple “ON beat, suppliers win” conclusion. Tower Semiconductor, a specialty foundry for image sensors and power, gets some support from the fact that Power Solutions Group revenue increased 8% quarter-over-quarter, but Intelligent Sensing Group revenue declined 8% quarter-over-quarter, which offsets the read for image-sensor exposure. SICC and Wolfspeed, tied to SiC substrates and 150mm SiC wafers, should not be credited with a demand acceleration from this print because automotive revenue decreased 4% sequentially and management identified weakness in America and Europe. The better implication for suppliers is inventory discipline rather than volume upside: distribution inventory stayed inside the 9 to 11 weeks range, which lowers the risk of abrupt order cuts but does not imply a rebuild.

Relative to peers, onsemi is still priced against a margin recovery it has not yet earned in reported results. TXN is printing 58.0% gross margin, NXPI is at 56.2%, and IFNNY is at 38.7%. onsemi at 37.6% is no longer in Q1 emergency territory, but it is closer to IFNNY than to the analog leaders investors often use as through-cycle benchmarks. The peer point is not that the businesses are identical; it is that the current margin gap is too large to ignore when ON’s own Q3 guide embeds 900 basis points of underutilization. If the market is giving onsemi credit for a return to its 53% target, the next few quarters need to show leverage from revenue growth, not merely cost containment.

The expense and capex lines help the near-term EPS floor, but they also reveal how much of the thesis depends on self-help. GAAP operating expenses were $359 million versus $396 million in the second quarter of 2024, and non-GAAP operating expenses were $298 million versus $308 million a year ago. Q3 OpEx is guided to $280 million to $295 million, and Q3 capital expenditures are guided to $35 million to $50 million. Those numbers support free cash generation, especially with Q2 capital expenditures at $78 million or 5% of revenue. But they also mean management is optimizing the cost base into weak demand rather than expanding capacity into confirmed growth.

The closing test for the stock is straightforward: Q3 must prove that the Q2 stabilization has operating leverage behind it. Confirmation would be revenue at the upper end of the $1.465 billion to $1.565 billion range, gross margin above the Q2 level of 37.6%, and distribution inventory still within the 9 to 11 weeks target range. The thesis breaks if Q3 revenue merely holds near Q2 while non-GAAP gross margin stays inside 36.5% to 38.5%, because that would validate the underutilization overhang rather than narrow it. The next date is the Q3 FY2025 report for the period ending 2025-10-03, and the decisive numbers are not EPS optics but utilization evidence: the 900 basis points charge, automotive’s sequential direction, and whether free cash flow can stay on the path to the 25% full-year margin target.