AMKR: Flat Sales, Higher Profit — When Does the Cash Catch Up?

The 60-second read

Five things to know

1. More profit without more sales. Revenue rose only0.2%, while attributable earnings rose50.6%; consolidated margin explains the change.
2. The mix broadens. Computing reached26% of revenue and auto/industrial22%, offsetting weaker communications.
3. The prior guide was mixed. Revenue and EPS landed inside July ranges, but18.3% gross margin missed the18.5% floor.
4. Cash investment still leads. Nine-month cash PP&E exceeded operating cash by$368M under our simple FCF definition.
5. The spending ceiling narrows. FY2026 capex guidance is now≈$2.5B; returns on that investment remain the valuation test.

Three scenarios, one probability-weighted number

Explicit analyst scenarios, not company forecasts.

Signposts: what would change our mind

Next source review: Q4/FY2026 release; February2027 is a research check window, not a confirmed issuer date.

The tape — the close came before the release

Our decision starts with a timing distinction. The $36.81 reference is Amkor’s October 26 regular-session close. The company released results after that session. The decline from the preceding $39.30 close therefore cannot be called the market’s reaction to these earnings. A provider timestamp or a conference-call appointment does not turn a pre-release trade into a post-release signal. We preserve the closing prices and use their computed change only as background. The first completed post-release trading session remains pending in this packet.

The company has become materially less expensive than the stale targets visible in our provider snapshot, but that difference does not automatically establish mispricing. A target is an analyst’s dated estimate under a set of assumptions. It is not a claim on cash and it is not guaranteed to update when investment plans, customer demand or the stock price changes. The retrieved 52-week range is $29.50–$96.68 and beta is 2.224. These are descriptions of historical trading variation, not probabilities for our valuation scenarios.

We rate the shares HOLD with three-out-of-five conviction and High uncertainty. Our rounded base fair value is $35, with a $27–$40 band across three valuation methods. The probability-weighted scenario value is $37.475, slightly above the price, because a successful expansion carries meaningful upside. That upside does not remove the downside if capacity spending outruns demand. We require a 25% margin of safety against the $35 anchor, or roughly $26.25, before price alone would support a BUY. Better operating evidence could change the anchor instead.

This quarter’s research question is the conversion of margin gains into durable owner cash. That differs from the prior Q2 episode’s record-quarter and AI-investment framing. The company is earning more on roughly unchanged annual-comparison revenue. Our task is to determine how much of that improvement survives the capital requirements of the business, without explaining a pre-release price change using information investors had not yet received.

The print — profit improved much faster than sales

Revenue of $1,990.557 million barely exceeded the $1,986.968 million earned a year earlier. Yet net income attributable to Amkor rose from $126.589 million to $190.613 million, and diluted EPS increased from $0.51 to $0.76. The improvement is primarily a margin story rather than a broad acceleration in total demand. Gross profit increased by $79.039 million while operating expenses rose by $13.255 million. Operating profit consequently increased by $65.784 million, a bridge that explains most of the stronger earnings before considering financing and taxes.

The quarter was not a clean beat against every comparison. The July release guided Q3 revenue to $1.95–$2.05 billion, gross margin to 18.5%–19.5%, and diluted EPS to $0.72–$0.82. Actual revenue and EPS landed inside those bands, while the reported 18.3% gross margin was below the bottom of its range. The dated provider consensus was $2,036.668 million revenue and $0.783 EPS; actual results were approximately 2.26% and 2.94% below those measures. The provider snapshot is not an immutable record of what every analyst expected immediately before release, so the comparison is described as retrieved consensus.

Q4 guidance calls for $1.9–$2.0 billion revenue, 18%–19% gross margin, and $0.65–$0.75 diluted EPS. The $1.95 billion revenue midpoint is approximately 2.0% below Q3. That outlook suggests that current profitability may hold at a substantially better level than last year without requiring another immediate step up in revenue. It also prevents us from annualizing one strong quarter as if seasonality had disappeared.

Adding the Q4 revenue midpoint to reported nine-month sales produces an illustrative full-year revenue figure of $7,523.223 million. This is our arithmetic, not a newly issued company annual revenue forecast. The distinction matters because the annual provider consensus remains higher. Investors should ask whether estimates incorporate the new range before treating the current consensus as the company’s promised trajectory.

Products and customers — diversification has economic value

The mix is changing in a useful direction. Communications represented 37% of Q3 revenue, down from 51% a year earlier. Computing rose to 26% from 19%, and automotive, industrial and other applications rose to 22% from 16%. Consumer applications accounted for the remaining 15%. Applying those rounded shares to reported revenue gives approximate end-market dollar amounts, not separate audited revenue lines. The approximation shows why total sales can remain flat while the company describes record computing and automotive activity: growth in those markets is offsetting a weaker communications contribution.

That diversification should reduce dependence on any single device cycle, but it does not make demand predictable. Packaging customers can change schedules rapidly, and the annual report warns about short commitments, utilization sensitivity and concentrated customers. The top ten customers still represented 66% of Q3 sales. That is below the prior-year 73%, yet it remains a substantial concentration. The latest annual report separately identifies one customer at 29.8% of FY2025 revenue and a second at 11.1%; those annual figures are not substitutes for an undisclosed current-quarter customer split.

Advanced products generated approximately $1,632 million and mainstream products $359 million in Q3. The advanced label encompasses several packaging technologies; it is not a disclosed pure-AI revenue measure. Advanced-product revenue was below the prior-year $1,684 million, while mainstream revenue exceeded the prior-year $303 million. Calling all advanced sales AI, or attributing all margin expansion to one AI program, would go beyond the reporting. We therefore judge the economics using the product mix, margin and investment evidence actually supplied.

Amkor reports the business as one segment. Product families and end markets are useful operating disclosures, but their individual operating profits are not separately supplied. We show consolidated operating profit rather than manufacture segment margins. The evidence supports broader demand participation and better consolidated profitability; it does not yet identify the return earned by each new factory or each customer program. That missing allocation is one reason the valuation retains High uncertainty.

Profit — distinguish earned margin from cash available

The gross-margin expansion is substantial. The exact statements imply a gain of approximately 394 basis points versus last year; the rounded release presents 18.3% against 14.3%. Materials fell from 57.5% of revenue to 50.7%, while labor, depreciation and other manufacturing costs absorbed higher shares. The release does not provide a complete causal decomposition among mix, purchasing, efficiency and customer pricing. We can identify the accounting bridge without asserting that every dollar is permanent or that one operational cause explains the entire change.

Operating expenses increased with the business’s demands. Selling, general and administrative expenses were $99.026 million versus $83.211 million, while research and development was $39.792 million versus $42.352 million. Together they absorbed part of the gross-profit improvement. Below operating profit, interest expense was $17.916 million and other net income was $23.367 million. Those lines produced pretax income of $230.162 million, followed by $39.243 million of tax expense. The resulting effective tax rate is approximately 17.1%; a different geographical earnings mix or incentive treatment could alter future tax expense.

The company’s $435 million EBITDA measure adds interest, taxes and approximately $187 million depreciation and amortization to consolidated net income. Those add-backs are transparent, but depreciation is especially relevant in a capital-intensive packaging business. Excluding it improves comparability of current operating cash generation; it does not mean the equipment is free or will never need replacing. We therefore use EBITDA as one valuation route and cross-check it against a cash-flow method.

There is no separate adjusted-EPS series in this release that justifies relabeling $0.76 as non-GAAP EPS. GAAP net income attributable to shareholders is slightly below consolidated net income because of noncontrolling interests. The current minority allocation is small, but it remains part of the ownership bridge. These distinctions keep valuation denominators consistent with the security an outside common shareholder actually owns.

Earnings quality — stronger operations, unfinished cash test

Our quality assessment separates an accounting improvement from the investment return it may eventually support. Operating cash flow for the first nine months was $868.345 million, well above consolidated net income of $449.754 million. Depreciation contributes materially to that conversion, as expected for a manufacturer with substantial installed equipment. The conversion is real, but it cannot be read independently of capital expenditures. After $1,236.806 million of cash property-and-equipment payments, our simple free-cash-flow measure is negative $368.461 million.

Receivables and inventories also deserve attention. The current balance sheet reports $1,511.028 million receivables and $603.792 million inventory. The charts use ending balances divided by the quarter’s revenue or cost of sales, multiplied by 91 days. This is a consistent monitoring proxy, not an average-balance days-sales-outstanding calculation. Seasonality and product mix can move it without indicating a collection or inventory problem. A sustained rise alongside softer demand would nevertheless weaken the quality case.

The current earnings release does not isolate stock-based compensation for Q3. We mark that quarter-specific check unavailable rather than accept a provider zero as proof of no expense. The annual report supplies share-based compensation of approximately $20.1 million, $18.4 million and $8.3 million for 2025, 2024 and 2023. Those amounts are retained in the annual dataset. We do not subtract an invented current adjustment or describe an undisclosed quarterly reconciliation as complete.

The guidance record is mixed in a useful, testable way: revenue and EPS were inside the prior bands, but gross margin fell below the lower boundary. The latest 10-Q before this release and the annual controls assessment did not identify a reporting failure that disqualifies current figures. That is source eligibility, not a guarantee of future accuracy. Investors should watch whether Q4’s 18%–19% margin range holds while capacity investment continues, and whether the balance sheet carries signs of demand being pulled forward.

Cash and financing — spending is still ahead of self-funding

Capital spending is the central qualification to the profit improvement. The company updated full-year capital expenditure guidance to approximately $2.5 billion, the bottom of its previous $2.5–$3.0 billion range. Lowering the upper end reduces one uncertainty, but it still describes an unusually large investment requirement relative to recent operating cash generation. Spending supports future capacity and customer commitments; its economic value depends on utilization, pricing, yields and cash receipts after the assets enter service.

Our simple cash-flow measure subtracts cash PP&E payments from operating cash flow. It differs from company free-cash-flow presentations that may add disposal proceeds, grants or insurance recoveries. We keep the definition consistent across the history. Q3 alone generated approximately $486.772 million operating cash flow and required $548.381 million cash PP&E payments, leaving negative $61.609 million under that definition. Both quarterly amounts are computed by subtracting the first-half cumulative figures from the nine-month statements. The provider’s incomplete current capex fields were rejected in favor of those source statements.

Liquidity is substantial, but gross obligations have increased. Cash and short-term investments total $2,376.196 million against $2,467.235 million debt, producing approximately $91.039 million net debt before lease obligations and restricted-cash adjustments. Restricted cash is not silently included in freely available liquidity. Capital expenditure payables were $496.129 million; these obligations help explain why a cash-spending measure and an investment plan may differ in timing. Subtracting cash spending to date from the annual capex plan is only a rough cash-planning comparison, not a promised fourth-quarter payment schedule.

The company repurchased $32.328 million of common stock during the quarter and paid its regular dividend. These returns coexist with borrowing, investment commitments and the convertible-note structure described in the filings. The relevant question is the eventual per-share return from all uses of capital, not whether buybacks alone signal confidence. We would prefer evidence that new capacity earns attractive cash returns before assigning the entire investment program a premium valuation.

Valuation — cash discipline earns a lower anchor than stale targets

Three methods frame the decision, and their disagreement is informative. The earnings route uses normalized 2027 EPS of $2.80 and a 14-times multiple, producing $39.20. That EPS is close to the retrieved 2027 consensus, but the multiple is our judgment for a cyclical, capital-intensive company. It is not a reported fact or an assertion that the Street endorses our discount. The enterprise-value route applies six times $1.65 billion normalized EBITDA, subtracts $91.039 million net debt, and divides by 249.526 million diluted shares, producing about $39.31 per share.

The third route is deliberately cash constrained. We assume annual owner cash flows of $150 million, $350 million, $550 million, $650 million and $725 million across the next five modeled years. We discount those at 11% and grow the terminal cash flow at 2.5%. After subtracting current net debt and dividing by diluted shares, the value is approximately $27.16. These are analyst scenarios, not management forecasts. They show the penalty when investment recovery takes time even if earnings look attractive today.

The equal-weight average of these three routes is approximately $35.22, rounded to a $35 base anchor. The method band is $27–$40. We do not hide the lower cash-flow result by selecting only the two higher multiples. At the $36.81 reference price, a 14-times earnings framework requires about $2.63 of sustainable EPS. A six-times EBITDA framework requires approximately $1.546 billion of EBITDA after adding narrow net debt to diluted equity value. Neither hurdle is absurd, but each requires the profitability improvement to endure.

Our bear case is $20.40, using $1.70 EPS at twelve times; the base is the $35 three-route anchor; the bull is $59.50, using $3.50 EPS at seventeen times. Probabilities of 25%, 50% and 25% produce a $37.475 weighted value. The mean is not a target date or a promise, and it is higher than the base because the upside scenario has a longer tail. High uncertainty and the narrow expected cushion support HOLD rather than a valuation-driven BUY today.

Wall Street — date the marks before comparing them

The dated provider snapshot shows a Buy consensus label with seven Buy, six Hold and one Sell ratings. Its aggregate target is $78.75, bounded by $65 and $90. Those targets are materially above our valuation. The gap is not evidence that one side has made an arithmetic error; it indicates different assumptions, update dates or valuation horizons. We retain the individual target dates to make that difference reviewable.

The latest returned target news includes Morgan Stanley at $70 on July 28 and UBS at $90 on July 24. Those observations precede this quarter’s release. They should not be introduced as analysts’ fresh reactions to the October results. Older UBS and Morgan Stanley entries in the snapshot are retained as history, not additional independent current votes. The ratings distribution also aggregates records over time, and its provider label is not equivalent to a newly published research note.

Annual consensus revenue for 2026 was approximately $7.656 billion in the retrieved data. Reported nine-month sales plus the new Q4 midpoint produce approximately $7.523 billion, below that consensus. For 2027, the snapshot shows revenue around $8.653 billion and EPS around $2.80, based on eight revenue contributors and seven EPS contributors. Different fiscal horizons have different analyst counts; zero values in sparse distant forecasts are not treated as forecasts of zero profit.

We use this material to identify the expectation gap. The company’s margin improvement can be credible while forward estimates still need to fall or spending takes longer to produce cash. Conversely, if computing growth and utilization deliver better cash returns than our cautious sequence, the Street’s optimistic direction could prove right even if the exact target remains too high. Our disagreement is therefore explicit and conditional: current cash economics and execution uncertainty warrant a much smaller premium than the stale target set implies.

Management — grade promises against the next statement

The most useful management scorecard begins with the prior numerical commitments. Revenue and EPS fell inside the July Q3 ranges, while gross margin missed the lower boundary. We give credit for meaningful year-over-year improvement without rewriting that forecast record as an across-the-board beat. The current Q4 range preserves a strong margin level but does not promise another sequential revenue increase. That is a more restrained message than extrapolating record demand in selected end markets into the whole company.

Capital guidance moved to the low end of the prior spending range. That change is directionally helpful for cash requirements, but the difference between timing changes and permanent reductions matters. Equipment delivery schedules, capacity milestones, customer advances and capital payables can shift cash outlays between periods. The next annual filing should help distinguish those effects. Until then, a lower annual spending number does not automatically raise the long-term return on the program.

Ownership and governance also belong in the analysis. The annual report states that the Kim family and affiliates owned approximately 49.4% of outstanding common stock as of February 12, 2026. A voting agreement applies to certain conversion shares. This is a material influence on shareholder decisions, but it is not a reason to invent a separate class premium or assume outside shareholders have no rights. We retain the actual common-security structure and discuss the practical influence instead.

PricewaterhouseCoopers has audited the company since 2000, and the annual report provides the controls assessment. Employee counts were 28,700 in 2023, 28,300 in 2024 and 30,800 in 2025. Workforce growth supports a larger operating footprint, but headcount does not establish factory productivity by itself. The next management assessment will focus on whether the spending, staffing and customer commitments translate into utilization and cash returns rather than simply larger installed capacity.

Risks — the best bull case still needs utilization

The leading risk is an investment-return shortfall. New packaging capacity requires capital before it earns stable revenue, and a high fixed-cost base makes utilization important. We assign this risk High likelihood of affecting the path and High potential impact, rather than claim that a particular project will fail. Customer commitments improve visibility but do not remove the industry’s short ordering cycles or guarantee an attractive return on every asset.

Second is demand concentration and mix reversal. Communications remains the largest end market, and the top ten customers still account for two-thirds of revenue. A computing recovery could coexist with weakness elsewhere. The bull case is stronger if multiple applications support capacity at the same time; the bear case is stronger if new demand merely replaces declining demand while capital requirements keep rising. We cannot assign undisclosed customer-level profits to settle that debate.

Third is financing and dilution. The balance sheet’s near-neutral narrow net debt can obscure the size of gross debt, restricted cash, capital payables and future investment commitments. Convertible notes and capped-call arrangements add terms that matter to per-share outcomes. We include the current diluted share base but do not assume every future dilution possibility is already captured. A refinancing, conversion or capital-allocation decision could change the denominator or the cash available to owners.

The strongest bull argument is that the current margin expansion demonstrates valuable technology and customer positioning just as computing and automotive programs diversify revenue. If capacity ramps successfully, earnings and cash can scale faster than sales. The strongest bear argument is that the apparent cheapness rests on earnings multiples while equipment spending consumes the cash, leaving shareholders exposed to a cyclical slowdown. Our HOLD balances those cases and requires evidence, rather than a slogan about AI, to move the conclusion.

Checkpoints — what would change the call

The next reported quarter should test whether the gross-margin improvement is durable at a slightly lower revenue midpoint. A margin at or above 18.5% would support the case that the business has improved structurally; a result below the 18% guide floor would make that conclusion harder. Revenue at or above the $1.95 billion midpoint would reduce concern that the latest outlook signals a broader demand shortfall. These checkpoints are analytical thresholds, not promises that the share price must respond in a particular direction.

Cash deserves a separate checkpoint because a good income statement can coexist with substantial investment. We will compare year-end cash capex, operating cash flow, capital payables and any identified grants or disposals. A positive simple free-cash-flow quarter would be evidence of better funding balance, although seasonality and payment timing must still be explained. A larger deficit accompanied by rising payables and weaker margins would be more concerning than a planned deficit with stronger utilization.

The portfolio test is whether computing participation holds while communications stabilizes. We use current 26% computing mix and 66% top-ten-customer concentration as reference points. A lower concentration percentage can be helpful, but only if it reflects healthy new customer revenue rather than a sudden loss at a major existing customer. The disclosure must be interpreted together with total sales and profitability.

We will revisit these measures with the Q4/FY2026 release, expected in the next annual reporting cycle; a firm release date has not been verified here. For scheduling the research check, we use February 2027 as an analytical review window, not an issuer-confirmed event date. The prior Q2 research packet was not available in the canonical local packet directory, so no historical signpost grade or invented prior fair value is presented. The prior published title is preserved to prevent repeating its angle.

Source notes — preserve the definitions and the limits

This packet uses the SEC-filed October 26 earnings release for the current quarter, the latest available Q2 10-Q and FY2025 10-K for structural context, historical SEC-filed earnings releases for operating mix and prior guidance, and dated Financial Modeling Prep responses for market and historical statement data. The raw files and retrieval manifests are saved with the episode. Nine quarterly financial records and three annual cash-flow records remain available in the supporting dataset, including their source links.

Several provider fields required correction. The current statement response returned zeros for EPS and diluted shares despite explicit issuer values; those were replaced with the release figures. Current depreciation and EBITDA were also taken from the issuer’s rounded reconciliation. Quarterly cash capital expenditures were incomplete or incorrectly signed in the newest provider rows. We computed Q2 and Q3 payments from cumulative primary cash-flow statements and documented the subtraction. A zero in a sparse provider field is not evidence that the company had no expense or investment.

All derived margins, changes, working-capital proxies, free cash flow, enterprise value, sensitivity values and valuation routes are computed in the saved authoring workflow. Analytical assumptions are labeled separately from reported figures. The model uses diluted weighted-average shares as a transparent valuation denominator; it is not a forecast of the exact future share count. The narrow enterprise-value bridge excludes leases and restricted cash and therefore should not be compared casually with a provider’s differently defined enterprise value.

The present work verifies an actual Q3 reporting period and continuing listed common security. Ordinary capacity expansion, customer programs, debt issuance and capital allocation are analyzed as financial risks rather than blanket eligibility exclusions. Exact release minutes remain unconfirmed, and no stock move is attributed to the print before a completed post-release window exists. Human audio audition is outside this research packet; production will separately preserve the fixed voices, automated provenance checks and actual visual review required for the episode.

Current SEC earnings release · FY2025 annual report · Charged Alpha research