Rohm raised its annual operating-profit forecast by 66.7% on November 5, citing stronger AI-related server demand and improving profitability. The Japanese semiconductor maker now expects ¥50 billion for the year ending March 2027. But its newly released accounts show why investors should distinguish a recovery in reported profit from growth in earnings before depreciation.

TECHi calculates that the ¥8.389 billion decline in first-half depreciation was equivalent to 85.3% of the ¥9.834 billion increase in operating profit. Rohm’s own simplified EBITDA measure rose just 4.2%, against 128.5% growth in operating profit. That does not erase the genuine revenue recovery; it changes the test for whether the higher forecast represents a durable improvement in the business.

Article Brief

Key Takeaways

4 Points24s Read

  • The newsRohm raised its March 2027 operating-profit forecast from ¥30 billion to ¥50 billion, a 66.7% increase.
  • The original findingFirst-half depreciation fell ¥8.389 billion, equivalent to 85.3% of the ¥9.834 billion operating-profit increase.
  • The counterpointRevenue rose 14.4% and server-related demand improved, but simplified EBITDA rose only 4.2%.
  • The next testNovember 6 presentation materials should provide more detail on the profit recovery; the semiconductor devices segment still posted a loss.

The upgrade is large, even after allowing for the accounting benefit

Rohm’s revised forecast covers April 1, 2026 through March 31, 2027. Expected revenue rises from ¥510 billion to ¥550 billion, an increase of 7.8%. Operating profit rises from ¥30 billion to ¥50 billion, while expected net profit attributable to owners increases from ¥29 billion to ¥45 billion. The revision was released through TDnet at 3:30 p.m. JST on November 5.

The difference between the two outlooks is ¥20 billion of operating profit against ¥40 billion of additional revenue. That 50% ratio is a comparison of forecast revisions, not a disclosed margin on individual products or proof that half of every future sales yen becomes profit. It captures a broader change in expected revenue, pricing, costs and business mix over the same fiscal year.

Rohm says server and storage demand remains strong as generative AI drives data-center investment. It also points to recovering factory-automation demand, resilient electrified-vehicle business, pricing initiatives and cost reductions. Management expects profitability to improve more than originally assumed, despite higher material, energy and logistics costs. The disclosure therefore describes a company-wide recovery with an AI contribution, rather than a business whose entire forecast can be assigned to AI chips.

Rohm’s revised March 2027 outlook
JPY billions unless statedPrevious forecastNovember 5 forecastChange
Revenue510550+7.8%
Operating profit3050+66.7%
Operating margin / calculated5.9%9.1%+3.2 percentage points
Net profit attributable to owners2945+55.2%
Basic EPS / JPY75.12116.56+55.2%
Swipe across to compare all columns. Rohm’s November 5, 2026 revision of its May 12 forecast; fiscal year April 1, 2026–March 31, 2027, Japanese GAAP. Margins and percentage changes calculated from the disclosed inputs before rounding. These are company forecasts, not achieved results or analyst consensus.

The annual operating margin implied by the new outlook is 9.1%, compared with 5.9% in the previous plan. That 3.2-percentage-point increase is the practical promise investors need to assess. A larger market opportunity helps, but the forecast also requires Rohm to retain more profit from its sales.

What the 85% depreciation comparison actually measures

The first-half earnings report shows operating profit increasing from ¥7.653 billion to ¥17.487 billion. Depreciation expense in the cash-flow statement fell from ¥27.148 billion to ¥18.759 billion. Subtracting each pair produces the ¥9.834 billion profit increase and the ¥8.389 billion depreciation reduction used in TECHi’s calculation.

Dividing ¥8.389 billion by ¥9.834 billion gives 85.3%. This is an arithmetic bridge between two reported measures. It is not an estimate that 85% of customers’ purchases were artificial, nor does it establish that sales growth, pricing or restructuring had no economic value. It shows that most of the increase in profit after depreciation has a counterpart in a lower depreciation charge.

Rohm explicitly says that reduced depreciation following impairment charges in the previous fiscal year contributed to the first-half profit improvement. An impairment reduces the recorded value of assets; subsequent depreciation can then fall. The earlier write-down and the later expense reduction belong to different reporting periods, so reading only the latest operating-profit growth rate misses part of the story.

The company’s simplified EBITDA, defined as operating profit plus depreciation, was ¥36.247 billion, up 4.2%. Adding the published statement inputs, truncated to millions of yen, gives ¥36.246 billion; the ¥1 million difference reflects reporting precision. That small rounding difference does not change the conclusion: earnings before this expense increased much more slowly than operating profit.

The first-half profit bridge
JPY billionsApril–September 2025April–September 2026Change
Operating profit7.65317.487+9.834
Depreciation27.14818.759−8.389
Operating profit plus depreciation / reconstructed34.80136.246+1.445
Swipe across to compare all columns. TECHi calculations from Rohm’s November 5 earnings and cash-flow statements. Inputs are truncated to millions of yen. The company separately reports simplified EBITDA of ¥36.247 billion, up 4.2%; the ¥0.001 billion difference does not affect the 85.3% comparison. Depreciation reduction / operating-profit increase = 8.389 / 9.834. This is an arithmetic bridge, not a measure of cash flow or a claim that other business drivers were absent.

The strongest counterargument is that a lower future cost base can still support shareholder returns. A past impairment does not make today’s reported profit unusable. But a stock forecast built on continued 128.5% operating-profit growth would need evidence beyond an expense reduction that has already occurred. EBITDA is also not cash flow: it excludes working-capital movements and does not fund factories by itself.

Server demand is real, but one major segment still loses money

First-half revenue rose 14.4% to ¥279.483 billion. Within the LSI business, Rohm reports strong server-related product growth alongside gains in automotive and industrial applications. Segment revenue increased 12.8% to ¥128.803 billion, and segment profit rose 35.5% to ¥17.556 billion.

The semiconductor devices business offers a more complicated picture. Revenue increased 20.7% to ¥120.720 billion. Rohm describes firm server demand for silicon-carbide power devices and a large increase in server-related silicon power-device sales. Nevertheless, the segment recorded a ¥4.393 billion loss, compared with a ¥9.809 billion loss a year earlier.

The loss narrowed by ¥5.416 billion, a meaningful improvement. It also means that higher sales have not yet produced profitability across this entire segment. Treating stronger AI infrastructure demand as proof that the silicon-carbide business has completed its turnaround would go beyond the results. Rohm continues to identify improving that business’s profitability as an objective.

Elsewhere, modules revenue fell 4.4% to ¥16.392 billion and segment profit declined 30.0% to ¥1.701 billion. The recovery is uneven. Neither the LSI result nor the semiconductor devices result should be presented as a separate AI revenue total, because the company’s reporting categories include several end markets.

For readers comparing Japanese suppliers, TECHi’s Shibaura forecast analysis examines a different issue: converting a much larger equipment order book into recognized revenue. Rohm’s fresh accounts instead make the relationship between depreciation, operating profit and cash generation central to the investment question.

The cash balance rose while operating cash flow fell

Rohm ended September with ¥477.535 billion of cash and cash equivalents, an increase of ¥48.821 billion from March-end. That provides substantial financial resources, but the cash-flow statement does not show that the increase came entirely from stronger trading.

Operating cash inflow declined to ¥25.546 billion from ¥47.190 billion a year earlier, a fall of 45.9%. The statement records a ¥23.331 billion working-capital outflow from receivables and a ¥17.122 billion inflow from trade payables. Those movements help explain why improved reported earnings did not translate into a similar increase in operating cash generation.

Investment activity provided ¥38.488 billion of net cash, including ¥62.754 billion from sales and redemptions of securities and investment securities. Meanwhile, purchases of property, plant and equipment fell to ¥26.682 billion from ¥81.624 billion.

A narrow cash-after-capex calculation is useful here: operating cash flow minus cash purchases of property, plant and equipment was negative ¥1.136 billion, compared with negative ¥34.434 billion a year earlier. This is TECHi’s explicitly defined calculation, not a company free-cash-flow forecast. It excludes securities transactions, subsidies and other investing items. It shows a much smaller funding gap, helped by sharply lower capital spending, while avoiding the assumption that asset sales are recurring operating income.

The balance sheet also lists ¥100 billion of short-term borrowing, ¥100 billion of long-term borrowing and ¥200 billion of bonds. Subtracting those ¥400 billion of financing balances from cash and cash equivalents leaves ¥77.535 billion. That limited comparison excludes other liabilities and is not a measure of cash available for an immediate distribution.

A stock forecast needs the right earnings denominator

At publication, the fixed price used for valuation is Rohm’s November 5 Tokyo close of ¥3,246 at 3:30 p.m. JST, from Yahoo Finance market data. Tokyo’s regular cash session has ended. The provider’s formal delay is unknown. The close was 6.0% below November 4’s ¥3,455, but its timestamp coincides with the disclosure time, so this daily move is not a clean measurement of the market’s response to the new forecast.

The company’s new annual basic EPS forecast is ¥116.56, compared with ¥75.12 previously. At the fixed closing price, that implies 27.8 times company-guidance earnings, versus 43.2 times using the previous forecast at the same price. These are company estimates for the year ending March 2027, not analyst consensus or a normalized earnings valuation.

If actual EPS were 20% below the new estimate, the multiple at an unchanged share price would rise to 34.8 times. That is a sensitivity, not a prediction of a miss or a share-price target. It illustrates how much the apparent valuation improvement depends on achieving the upgraded earnings denominator.

Share structure adds another limitation. Rohm reported first-half basic EPS of ¥40.36 and diluted EPS of ¥34.42. Those measures should not be interchanged, and the company’s annual basic forecast should not be described as a fully diluted estimate. Nor should the prior fiscal year’s large net loss be used to manufacture a meaningful positive trailing P/E.

What would strengthen or weaken the recovery case

The constructive case rests on more than depreciation. Server-related revenue is expanding, the semiconductor devices loss has narrowed, and the annual operating margin target has increased. Evidence that pricing and cost initiatives lift profitability while operating cash flow improves would make the recovery more convincing.

The weaker case would be revenue growth that leaves the power-device segment loss-making and fails to improve cash conversion. Currency also matters: the revised outlook assumes ¥153 per dollar for the second half. A different exchange-rate outcome, higher input costs or weaker automotive and industrial demand could alter the result. The release does not provide enough information to assign a precise stock-price effect to any one of these risks.

What to watch

Rohm says it plans to publish its earnings presentation and supplementary materials on November 6, followed by its semiannual securities report on November 9. The presentation is the next dated opportunity to assess the profit bridge, server demand and restructuring detail. Until those materials arrive, the useful forecast question is whether stronger demand produces growth in earnings and cash beyond the benefit of lower depreciation.