Power Integrations’ EPS beat masks the wrong surprise: the GaN story is not yet big enough to break the revenue band

The clean read on this print is that Power Integrations is a better margin story than revenue acceleration story, and that distinction matters for the stock. What was priced in was $127.1 million of revenue and $0.34 of EPS, a setup that assumed the recovery in demand would show up in the top line without sacrificing profitability. What actually surprised was the split: revenue came in at $115.9 million, an -8.9% miss, while EPS was $0.35, a +2.9% beat. That is not a normal “better quality” beat, because the company did not beat revenue and then drop incremental dollars through the model. It missed the sales bogey by a meaningful amount, while non-GAAP EPS still reached management’s reported $0.35 because gross margin stayed high, tax was low, and capital returns reduced the denominator. The market may be mispricing the beat as confirmation that the product cycle is already taking over; the evidence says the recovery remains narrow, with GaN and industrial doing the work while appliances remain the offset.

That distinction becomes clearer when the quarter is placed against the recent financial trajectory rather than the Street model alone. Revenue has not escaped the same broad band that has defined the company since the post-2023 reset: the recent history runs from a $89.5 million trough to a $125.5 million high, and Q2 FY2025 at $115.9 million sits closer to stabilization than breakout. Gross margin tells a different story, because it has held around the mid-50s after the company worked through the 2024 trough in demand. The important point is not that sales grew, since management’s own framing was revenue up 9% year-over-year to $116 million, but that the revenue base remains too small to absorb rising legal, transition, and R&D costs without leaning on mix, tax, and buybacks. In other words, Power Integrations has fixed enough of the model to defend EPS, but it has not yet proven the demand slope needed to make the $1 billion revenue ambition investable on this quarter’s facts.

The margin line is why the quarter did not look worse in EPS, but the guide says that cushion is starting to narrow. Sandeep Nayyar put the reported non-GAAP gross margin at 55.8%, down 10 basis points from the prior quarter, and guided Q3 non-GAAP gross margin to between 55% and 55.5%. That is still high for a power-discrete peer set, but the sequential direction matters because the company tied the pressure to higher input costs flowing through inventory and a smaller dollar-yen benefit. At the same time, non-GAAP operating expenses were $46.7 million in Q2, and Q3 is expected around $47.5 million. If revenue only moves to the guided $118 million midpoint while gross margin steps down and OpEx steps up, the earnings bridge relies less on operating leverage than a bullish GaN-cycle thesis would require.

The demand guide reinforces the same conclusion because management did not describe a broadening recovery. Jennifer A. Lloyd’s wording matters because it separates the businesses that are working from the one still blocking the inflection: “Our third quarter revenue outlook of $118 million, plus or minus $5 million reflects continued strength in the industrial category and in GaN products, tempered by softness in appliances, which make up most of our consumer category.” That sentence is the quarter’s operating thesis. Industrial and GaN are real, but appliances are large enough to keep the whole company near $118 million rather than back toward the prior $125.5 million high. The mix disclosure sharpens the point: Q2 revenue was 40% industrial, 37% consumer, 12% computer and 11% communications. With consumer nearly as large as industrial, appliance softness is not a footnote. It is the reason the company can talk about product-cycle upside while still guiding a small sequential revenue step.

The most constructive interpretation is that Power Integrations is deliberately trading breadth for quality, leaning into higher-value conversion ICs and GaN where the economics can support the model before the volume recovery arrives. Lloyd said almost 90% of sales is power conversion ICs for appliances, consumer electronics and industrial applications, and she linked GaN to SAM expansion and the path toward $1 billion in revenue. That framing is credible as a product strategy, but the print does not yet validate the timing. Metering is called out as on track to grow 20%-plus this year, and higher-voltage GaN is said to create a path to ASP expansion as customers upgrade existing silicon designs. The catch is that these positives have not lifted company revenue above $115.9 million in the reported quarter or above a $118 million Q3 midpoint. A PM paying for the $1 billion pathway today is underwriting a conversion from design-win narrative to revenue slope that this print still leaves unproven.

The capital return bridge also cuts both ways. On one hand, the company generated $29 million in cash from operations and returned $44 million to stockholders, including $32.6 million of buybacks. That is a useful signal of balance-sheet confidence, especially with repurchases at an average price of about $46 and a diluted share count of 56.4 million, down about 700,000 from the prior quarter. On the other hand, using buybacks to help protect per-share results after an -8.9% revenue miss is not the same as demonstrating end-market acceleration. The remaining $42 million repurchase authorization gives management another lever if demand remains uneven, but it does not solve the core question of whether GaN and industrial can outgrow appliance softness enough to drive operating leverage through the model.

The cost disclosures make that question more urgent because several expenses are not purely cyclical. Nayyar said Q2 non-GAAP OpEx rose sequentially due mainly to annual salary increases, executive transition costs, and litigation expenses; he also identified a $9 million charge related to an employment litigation case in California. Some of that may fade, but Q3 non-GAAP operating expenses are still expected around $47.5 million, driven mainly by legal costs and R&D activity. The R&D component is strategically understandable if the company is investing behind GaN and higher-voltage platforms, yet the near-term income statement now has less room for a sluggish revenue quarter. This is why the EPS beat should not be overcapitalized: the company beat $0.34 by $0.01, but the next leg depends on revenue acceleration showing up before the higher cost base absorbs the margin benefit.

The call delivery was consistent with a management team trying to reset the narrative around growth without overpromising the next quarter. The tone history shows Q2 FY2025 sentiment at 0.03 and Q&A sentiment at -0.01, which fits a call where the prepared story leaned into GaN and long-term targets while the analyst exchange pressed on turns, appliances, and guidance. The later call-over-call profile in the table shows how delivery can change when confidence improves: Q1 FY2026 versus Q4 FY2025 sentiment rose by +0.15 and guidance_tone rose by +0.33, while qa_evasiveness fell by -44.5. For this Q2 FY2025 event, however, the relevant signal is not a triumphant tone but a constrained one. Management offered a $118 million guide, plus or minus $5 million, and described industrial and consumer as “kind of flattish” with growth coming from the other two segments. That is disciplined language, not acceleration language.

The second-order read-through is narrow because the data pack identifies no named Power Integrations customers and no named suppliers, so this print should not be stretched into a company-specific signal for any external account. The end-market read-through is still useful: appliance-exposed customers are not yet creating enough demand to offset the drag from the consumer category, while industrial demand and GaN adoption are carrying the positive side of the mix. The magnitude is visible inside Power Integrations rather than through a named supply-chain counterparty: consumer was 37% of Q2 revenue, industrial was 40%, and major appliances were described as about 50% of total consumer. That combination means appliance softness has enough weight to cap company-level growth even when industrial and GaN are moving in the right direction. For suppliers, the absence of named disclosures prevents a direct read-through, but higher input costs are already showing up in the guided margin range of 55% to 55.5%, so the internal evidence points to cost pressure rather than a clean pass-through environment.

Against the peer set, Power Integrations still screens as structurally higher margin but not clearly higher growth in this print. Its reported gross margin of 55.2% in Q2 FY2025 is far above Diodes at 31.8% and Vishay at 21.0%, which supports the view that integrated power conversion and mix are differentiated. But the revenue comparison is less favorable to a re-rating argument: Power Integrations’ actual revenue was $115.9 million, while Diodes reported $405.5 million and Vishay reported $839.2 million. The smaller revenue base should make growth easier if the product cycle is truly inflecting, yet the Q3 guide is only $118 million at the midpoint. That is the crux for portfolio positioning: the company deserves credit for margin architecture, but a premium multiple needs evidence that GaN and industrial can move revenue faster than the current guide implies.

The leadership transition gives bulls a credible long-term story, but it also raises the evidence bar. Lloyd arrived from Analog Devices after running multiple businesses with $1 billion plus in revenues, and she said she will have more to say “in the months ahead” about plans to deliver growth and get Power Integrations on a path toward $1 billion in revenue. That phrase matters because it is both ambitious and deferred. The company is not yet giving investors a quantified bridge from $115.9 million of quarterly revenue to that longer-term target in the data provided here. Until that bridge exists, the investable debate should be anchored in the quarterly guide, segment mix, and margin pressure rather than in the destination. A new CEO can change portfolio choices, customer targeting, and capital allocation, but the Q2 print shows a company still operating inside the old revenue range.

What would break the cautious variant view is straightforward. For Q3, the first confirmation point is whether revenue lands above the $118 million midpoint and closer to the high end of the plus or minus $5 million range, because anything near the midpoint keeps the company in the same band despite the GaN narrative. The second is whether non-GAAP gross margin holds inside the guided 55% to 55.5% range despite higher input costs and a smaller dollar-yen benefit; a miss there would remove the main support behind per-share resilience. The third is whether non-GAAP operating expenses stay around $47.5 million, because further legal or R&D creep would make the model more dependent on revenue acceleration that has not yet appeared. On the demand side, the thesis turns more constructive only if management can show that industrial and GaN are no longer merely offsetting appliances but lifting total revenue beyond the recent ceiling. Until the next quarter proves that, the actionable read is to respect Power Integrations’ margin quality but fade the idea that this Q2 EPS beat marks the start of a broad revenue inflection.