Amkor’s beat was margin, not demand, and the Arizona spend is the real revision

Amkor Technology printed a quarter that changes the debate from cyclical recovery to capital intensity, and that is the variant perception in this report: the surprise was not top-line acceleration, it was the combination of 16.7% gross margin, $0.69 EPS, and a newly explicit $2.5 billion to $3.0 billion CapEx envelope that shifts equity risk from near-term earnings delivery to return on U.S. advanced packaging capacity. What was priced in was a modest Q4 revenue beat, because consensus sat at $1,839.1 million and the company delivered $1,888.0 million, a +2.7% surprise. What was not priced in, or at least should not be treated as recurring without proof, was the +60.5% EPS surprise versus the $0.43 estimate, because the quarter’s own bridge includes gross profit of $315 million with “a benefit of approximately $30 million from asset sales,” per CFO Megan Faust. The market can like the beat and still misread its quality: the quarter says Amkor has pricing, mix, and utilization leverage when communications, automotive, and consumer line up, but the forward return profile now depends on filling Arizona and Vietnam capacity, not extrapolating one quarter’s EPS.

That distinction matters because Q4 demand was better than the Street expected but not a blowout by Amkor’s own seasonal history. Revenue of $1,888.0 million was down -5.0% sequentially from Q3 FY2025’s $1,987.0 million, even as it rose +15.9% year on year from Q4 FY2024’s $1,629.1 million. The EPS of $0.69 was the highest line in the provided quarterly history, but the company also had 16.7% gross margin against 14.3% in Q3 FY2025 and 15.1% in Q4 FY2024, and that margin level is where the earnings surprise lived. The Street had $0.43 in its model, which effectively anticipated a normal seasonal decline with some recovery in year-on-year revenue, not a quarter in which asset sales, mix, and utilization pushed diluted EPS above Q3 FY2025’s $0.51 despite lower sequential revenue. CEO Kevin Engel’s phrasing is useful because it commits to the quality of the guide comparison without overexplaining the source of the upside: “Q4 revenue was $1.89 billion and EPS was $0.69, outperforming the high end of our guidance.”

The financial trajectory shows why the market should separate the beat from the base. Amkor went from $1,321.6 million of revenue and 11.9% gross margin in Q1 FY2025 to $1,987.0 million and 14.3% in Q3 FY2025, then held $1,888.0 million and expanded gross margin to 16.7% in Q4 FY2025. That is not a simple revenue-beta story, because Q4 revenue declined -5.0% sequentially while gross margin rose to 16.7%. It is also not a pure structural-margin story, because the full year still landed at 14% gross margin and included a 90 basis point headwind from the ramp-up of the Vietnam facility. The print therefore supports a narrower but more actionable conclusion: Amkor’s earnings power is highly sensitive to mix and non-recurring items in the near term, while its medium-term margin path is being depressed by facility ramps. If investors pay a multiple on the $0.69 quarter without haircutting the $30 million asset-sale benefit, they are paying for transitory margin. If they ignore the company’s 2026 computing and automotive language because Q1 EPS steps down, they may miss the capacity-led revenue option.

The guide sharpens that tension rather than resolving it. For Q1, management sees revenue between $1.6 billion and $1.7 billion, with CFO Megan Faust saying the midpoint represents “a 25% year-on-year increase.” On the company’s own account, that is a material year-on-year recovery, and the quarterly history line for Q1 FY2026 shows $1,684.7 million of revenue, +27.5% year on year, 14.2% gross margin, and $0.33 EPS. But the EPS guide in the key points was $0.18 to $0.28, and the company’s quoted net-income framework was $45 million to $70 million, resulting in EPS between $0.18 and $0.28. The conflict is not demand, it is margin and investment cadence: operating expenses are expected to increase to approximately $135 million, while Q4 operating expenses were $130 million, and the Q1 gross-margin line in the quarterly history falls to 14.2% from 16.7%. That is why the Q4 print should not be read as a straight-line EPS reset. A 25% year-on-year revenue guide can coexist with an EPS step-down when the company is absorbing R&D, facility ramp costs, and a less favorable mix after a quarter that included an approximately $30 million asset-sale benefit.

The most important revision is the capital budget, because it tells investors where management wants to place the next-cycle bet. CapEx in 2025 was $905 million, lower than guidance due to timing of cash payments for the Arizona facility, while 2026 CapEx is expected to rise to $2.5 billion to $3.0 billion, with 65% to 70% projected for facility expansion, including phase one of the Arizona campus. That is the crux: Amkor generated full year free cash flow of $380 million in 2025, held $2 billion in cash and short-term investments at year-end, and had total liquidity of $3 billion, a 30% increase from the prior year, but the 2026 plan is a step-change in cash deployment. The company has the balance-sheet liquidity to begin the build, but the equity case now depends on converting computing and advanced automotive wins into loaded capacity rather than merely avoiding a downturn. Engel gave investors a scale marker for Arizona with language that is not in the model by default: “Phase one is can think about half of that $7 billion, and then the construction is about 60% of that.” That comment matters because it frames phase one as a large multi-year construction commitment rather than routine maintenance CapEx.

The segment clues justify the strategic spend, but not yet the return. Communications revenue grew 28% year on year in Q4 and 1% for the full year, reflecting a stronger footprint in the current generation of iOS phones and demand across both iOS and Android ecosystems. Automotive and Industrial revenue increased 25% year on year in Q4 and 8% for the full year, driven by advanced automotive content for ADAS applications. Consumer grew 9% for the full year, helped by a full year of the wearable product and improvement across traditional consumer applications. For 2026, management expects full year revenue growth to be driven by computing, expected to grow over 20%, and continued growth in advanced automotive. Those numbers argue that Amkor is not only riding handset seasonality; it is attaching to higher-package-complexity categories where customers value scale, yield, geography, and advanced process capability. But the full-year 2025 gross margin of 14% and the 90 basis point Vietnam ramp headwind show that richer content does not immediately mean richer margins when capacity is still ramping.

That is also the right lens for the supply-chain read-through. For Apple, Amkor’s 28% year-on-year Q4 communications growth and 1% full-year communications growth point to better content in the current generation of iOS phones, while consumer’s 9% full-year growth is consistent with wearables exposure in Apple Watch and AirPods. For Qualcomm, the 2026 statement that computing is expected to grow over 20% is the relevant read-through because Amkor supplies FC-BGA/FOWLP for Snapdragon, so the packaging mix is tied to higher-complexity compute rather than only smartphone units. Infineon’s read-through is the 25% year-on-year Q4 Automotive and Industrial growth and 8% full-year growth, tied to ADAS applications. Suppliers get a more direct volume signal from the CapEx plan: Advantest and Teradyne have ATE exposure, Cohu has test handlers and contactors, Kulicke & Soffa has wire bonding equipment, and TOWA has compression molding equipment for IC packaging. The magnitude is not vague: Amkor is moving from $905 million of 2025 CapEx to $2.5 billion to $3.0 billion in 2026, and 65% to 70% is facility expansion, so equipment and test suppliers should see order opportunity, but the facility-heavy mix means not all of that spend is tool pull-forward.

The peer comparison argues for respect on growth but restraint on margin multiple. Amkor’s Q4 FY2025 revenue growth of +15.9% sits below ASX at +17.4%, 4062.T at +18.6%, 3481.TW at +19.2%, and 6787.T at +24.5%, but above 6966.T at +13.2%, KYOCY at +6.9%, 7911.T at +5.0%, and 7912.T at +1.5%. On gross margin, Amkor’s 16.7% in Q4 is above 3481.TW at 14.2% and 6966.T at 16.3%, but below ASX at 20.1%, 6787.T at 21.3%, 7911.T at 23.4%, 7912.T at 23.8%, KYOCY at 29.0%, and 4062.T at 29.5%. That mix of above-median growth and below-top-tier margin is exactly why the stock should react to evidence of advanced packaging utilization, not simply revenue. The company is investing like a strategic capacity winner, but the current margin structure is not yet in the peer group’s upper band.

The call delivery adds a useful layer because management sounded more guided and less evasive, even though uncertainty rose. The tone history moved sharply in the right places for a company asking investors to fund a CapEx step-up: guidance_tone rose from 0.25 in Q4 FY2025 to 0.52 in Q1 FY2026, prepared_sentiment rose from 0.01 to 0.83, and tone_confidence rose from 0.36 to 0.43. That is not just a happier script, because qa_evasiveness fell from 41.6 to 12.0, a call-over-call delta of -29.6. The caveat is real: uncertainty rose from 41.3 to 55.4, a call-over-call delta of +14.1, while qa_sentiment fell from 0.35 to 0.28. The numbers conflict in a way that matches the business situation. Management is more explicit on guide and capacity plans, but investors are asking harder questions because the spending envelope has expanded before the return curve is visible.

That tone profile helps explain why the stock debate should not be reduced to whether Q1 seasonality looks fine. The company’s own language has become more specific on 2026 drivers, and the numerical guide is healthy on revenue, but the earnings bridge is under pressure from operating expense, facility ramp, and the absence of Q4’s asset-sale benefit. Q4 operating income was $185 million with operating income margin of 9.8%, EBITDA was $369 million with EBITDA margin of 19.5%, and net income was $172 million. For full year 2025, operating income was $467 million with operating income margin of 7%, EBITDA was $1.16 billion with EBITDA margin of 17.3%, and net income was $374 million. The gap between Q4 profitability and full-year profitability is the warning label: Amkor can print high quarterly earnings when mix and utilization align, but investors need proof that a 2026 revenue mix led by computing growing over 20% and advanced automotive can lift full-year margins while Vietnam and Arizona absorb capital and start-up costs.

The bear case is straightforward and should not be dismissed: a $2.5 billion to $3.0 billion CapEx year against $380 million of 2025 free cash flow can cap near-term equity enthusiasm if Q1 gross margin remains around 14.2% and EPS stays close to the $0.18 to $0.28 guide range. The bull case is more interesting because it is less visible in the consensus beat math: Amkor’s Q4 showed that incremental mix can drive 16.7% gross margin, and the company is now explicitly tying 2026 growth to computing expected to grow over 20% plus advanced automotive, while communications already delivered 28% year-on-year Q4 growth. The variant view is that the market may focus too much on the EPS step-down and not enough on the customer-content signal embedded in communications, computing, and ADAS. But that variant view only works if investors discipline the Q4 EPS quality and demand conversion into full-year margin, because the $30 million asset-sale benefit means the $0.69 quarter is not a clean run-rate.

What to watch next is specific. On 2026-03-31, confirm whether revenue holds near the Q1 FY2026 history line of $1,684.7 million and within management’s $1.6 billion to $1.7 billion guide, because a miss below that range would break the 25% year-on-year growth narrative before the Arizona spend begins to matter. Watch gross margin versus 14.2%, because staying materially below Q4’s 16.7% is acceptable only if management continues to show computing and advanced automotive mix building, while a failure to hold the 14.2% area would signal ramp cost or utilization pressure. Watch EPS against the $0.18 to $0.28 guided range and the $0.33 Q1 FY2026 history line, because the gap determines whether Q4’s $0.69 was mostly one-time or a bridge to better earnings power. Finally, watch whether 2026 CapEx remains at $2.5 billion to $3.0 billion, with 65% to 70% for facility expansion, and whether management repeats that computing is expected to grow over 20%; if either number is cut without an offsetting margin improvement, the Arizona thesis loses its funding logic.