Series context This is a follow-up to Mirae Asset’s KRW 1.3M Samsung Electro-Mechanics target. The May note asked whether the market could apply a 37x 2017-style MLCC shortage multiple to 2028 earnings. The June 1 note moves the debate forward: can investors underwrite 2029 earnings for an AI package power-integrity and FC-BGA bottleneck company? Related hubs: AI PCB and Substrate Hub and AI HBM Hub.
TL;DR
Mirae Asset’s June 1 target price for Samsung Electro-Mechanics is mechanically clear:
Target price = 2029F EPS of KRW 93,242 x 30x PER
= KRW 2,797,260
= roughly KRW 2.8 million
The important point is that the target went up while the multiple went down. The May 6 report used 2028F EPS of KRW 34,764 x 37x = KRW 1.3M. The June 1 report uses 2029F EPS of KRW 93,242 x 30x = KRW 2.8M. The applied PER falls from 37x to 30x. The target-price upgrade is therefore not multiple expansion; it is a large upward revision to the 2027-2029 earnings path.
That changes the debate. Samsung Electro-Mechanics is no longer being valued as a cheap electronics-component recovery stock. The market is being asked to value it as a long-duration AI infrastructure bottleneck. If it wants a premium versus memory names, it now has to prove that its earnings duration and pricing power are stronger than memory.
1. From KRW 1.3M to KRW 2.8M: What Changed?
The May 6 Mirae Asset report valued Samsung Electro-Mechanics at around KRW 1.3 million by applying a 37x PER to 2028F EPS of KRW 34,764. The debate then was whether a 2017-style early MLCC shortage multiple could be applied again. (Mirae Asset Securities)
The June 1 report, based on the user-provided original document, uses 2029F EPS of KRW 93,242 and a 30x PER. I could not verify a public URL for the June 1 report, so the detailed June 1 numbers below are attributed to the uploaded report text.
| Item | May 6 Report | June 1 Report | Interpretation |
|---|---|---|---|
| Target price | KRW 1,300,000 | KRW 2,800,000 | +115.4% |
| Reference price | KRW 918,000 | KRW 2,127,000 | Stock was already up +131.7% between reports |
| Implied upside | 41.6% | 31.6% | Target rose, but risk/reward compressed |
| Valuation year | 2028F EPS | 2029F EPS | Valuation year moved one year further out |
| EPS | KRW 34,764 | KRW 93,242 | +168.2% |
| Applied PER | 37.0x | 30.0x | -18.9% |
| Formula | 34,764 x 37 = 1,286,268 | 93,242 x 30 = 2,797,260 | Earnings revision drives the target |
| Core logic | Capex, FC-BGA, AI MLCC, pricing upside | Multi-year exclusive contracts, FC-BGA +20% ASP, MLCC +10% ASP, Si-Cap demand | Pricing moves into the base case |
This is the key distinction. In May, pricing was an upside scenario. In June, pricing is embedded in the base case. That is positive because the framework is more explicit, but it also raises the bar. If the pricing does not show up, it becomes downside.
2. 2026 Is Unchanged; 2027-2029 Are Rewritten
Figures are in KRW trillions. Changes compare the June 1 estimate with the May 6 estimate.
| Year | Revenue: Old to New | Operating Profit: Old to New | Net Profit: Old to New | OPM: Old to New |
|---|---|---|---|---|
| 2026F | 13.52 to 13.52 / 0.0% | 1.56 to 1.56 / 0.0% | 1.31 to 1.31 / 0.0% | 11.6% to 11.6% |
| 2027F | 15.92 to 16.47 / +3.5% | 2.29 to 3.41 / +48.6% | 1.91 to 2.83 / +48.4% | 14.4% to 20.7% |
| 2028F | 18.76 to 22.24 / +18.5% | 3.21 to 5.87 / +82.8% | 2.70 to 4.93 / +82.6% | 17.1% to 26.4% |
| 2029F | 21.59 to 28.90 / +33.9% | 4.22 to 8.59 / +103.6% | 3.57 to 7.24 / +102.7% | 19.6% to 29.7% |
| 2030F | 24.82 to 32.54 / +31.1% | 5.37 to 10.59 / +97.0% | 4.59 to 9.04 / +96.9% | 21.7% to 32.5% |
The signal is clear: the near-term year is not the driver. 2026 is unchanged. The report rewrites the 2027-2030 earnings curve.
Investors should therefore stop asking only whether 2Q26 is good. The real questions are whether FC-BGA ASP can rise 20%, whether MLCC ASP can rise 10%, whether Si-Cap demand extends beyond the first contract, and whether margin expansion survives capex and depreciation.
3. Segment Read-Through: Components and Package Substrates
Components: MLCC and Si-Cap Must Earn Bottleneck Rent
| Year | Component OP: Old to New | Component OPM: Old to New |
|---|---|---|
| 2027F | 1.35 to 2.45 / +81.5% | 16.9% to 28.7% |
| 2028F | 1.70 to 3.58 / +110.2% | 17.9% to 33.9% |
| 2029F | 2.09 to 4.33 / +107.1% | 18.9% to 35.1% |
| 2030F | 2.63 to 5.44 / +107.0% | 20.0% to 36.6% |
This is a powerful assumption. A component OPM in the mid-30s is hard to justify with a normal MLCC cycle alone. The market has to believe Samsung Electro-Mechanics is selling a die-near AI package power-integrity bottleneck, not merely generic ceramic capacitors.
The official KRW 1.5 trillion silicon-capacitor supply contract supports that direction. Samsung Electro-Mechanics says the product is used inside high-performance semiconductor packages such as AI server GPUs and HBM to stabilize power delivery. (Samsung Electro-Mechanics)
But the first contract is not enough to prove 2029 earnings. SEMCO needs repeat orders, customer expansion, and evidence that Si-Cap becomes a platform socket rather than a one-off project.
Package: The Real Swing Is 2028-2029
| Year | Package Revenue: Old to New | Package OP: Old to New |
|---|---|---|
| 2027F | 3.94 to 3.94 / 0.0% | 0.78 to 0.79 / +1.6% |
| 2028F | 5.20 to 7.63 / +46.6% | 1.32 to 2.10 / +59.4% |
| 2029F | 6.43 to 12.45 / +93.7% | 1.92 to 4.05 / +111.3% |
| 2030F | 7.33 to 13.36 / +82.2% | 2.50 to 4.90 / +95.8% |
The package-substrate revision is not really a 2027 story. The major swing begins in 2028. Mirae Asset is effectively assuming that new FC-BGA capacity, AI server/network demand, and customer lock-in become material from 2028 onward.
That may be right, but FC-BGA is capex-heavy. Without customer deposits, long-term volume commitments or price-protection clauses, high growth can still be diluted by depreciation and yield risk.
4. Why SEMCO Must Prove It Is Stronger Than Memory
Once SEMCO is valued this way, its peer set changes. It is no longer only compared with passive-component companies. It is compared within the broader AI infrastructure basket, including Samsung Electronics and SK Hynix.
Memory has a very direct earnings engine: HBM, DDR5, eSSD and SOCAMM attach to AI servers. But memory stocks always carry peak-earnings doubt. When pricing rises, investors remember that supply usually follows, and ASPs eventually roll over.
For SEMCO to deserve a premium versus memory, it must prove something different:
| What SEMCO Must Prove | Why It Would Be Stronger Than Memory |
|---|---|
| Long-term contracts and minimum-purchase terms | Longer earnings visibility than a spot-driven memory cycle |
| Durable FC-BGA +20% ASP | Customer lock-in, not just temporary shortage |
| Repeatable MLCC +10% ASP | AI server power-integrity bottleneck, not a generic MLCC cycle |
| Additional Si-Cap customers and post-2028 extension | Platform entry, not one large contract |
| Component OPM in the 30s | Quality margin above traditional component cycles |
| Free-cash-flow conversion | Capex-heavy substrate growth does not destroy economics |
Put simply: memory has bigger earnings but gets discounted as cyclical. SEMCO is trying to earn a premium because its parts may be smaller but more durable. That durability now has to be proven.
5. 2029 Is Far: The Evidence Checklist
Using 2029F EPS is not wrong. Growth stocks often pull forward future earnings. But 2029 is far enough that investors need milestones.
2027 Evidence
| Checkpoint | Pass Signal | Fail Signal |
|---|---|---|
| MLCC +10% ASP | Component gross margin and OPM improve | Shipments rise but margin does not |
| FC-BGA +20% ASP | Contract evidence, customer deposits, long-term allocation | Price uplift offset by yield, depreciation or mix |
| Si-Cap revenue | Contract revenue starts on schedule in 2027 | Ramp, yield or customer schedule slips |
| Customer diversification | Second AI ASIC or hyperscaler customer appears | Single-customer, single-project dependence remains |
2028-2029 Evidence
| Checkpoint | Why It Matters |
|---|---|
| Package revenue path toward KRW 12T | The largest swing factor behind 2029F revenue |
| Component OPM near 35% | Core justification for a 30x PER |
| Package OPM approaching high 20s | Proof that FC-BGA is a lock-in business, not just capex |
| Post-2028 Si-Cap repeat demand | Turns the thesis from a contract event into a duration trade |
| Separation from memory ASP slowdown | Shows SEMCO can hold earnings even if memory price momentum fades |
6. Investment View
For existing holders, Core Hold still makes sense. The June 1 report gives a clearer formula and a larger earnings path. The use of 30x instead of 37x also weakens the criticism that the target is purely multiple expansion.
For new buyers, the setup is different. At the June 1 reference price of KRW 2,127,000, the upside to KRW 2.8 million is 31.6%. That is meaningful, but if the thesis fails, the downside can be larger because the current framework now embeds several years of success.
The cleaner interpretation is:
| Objective | Better Instrument |
|---|---|
| Direct AI memory exposure | Samsung Electronics or SK Hynix |
| AI package power-integrity duration | Samsung Electro-Mechanics |
| Lower peak-earnings doubt than memory | Samsung Electro-Mechanics, but only with evidence |
| Earnings momentum already showing in reported numbers | Memory mega-caps |
| A long-duration 2027-2029 earnings path | Samsung Electro-Mechanics |
Final Take
The most important sentence in the KRW 2.8 million report is not the target price. It is the valuation statement:
SEMCO is being valued as an AI package power-integrity and FC-BGA bottleneck company that can earn 2029F EPS of KRW 93,242 and deserve 30x earnings.
That can be true. But 2029 is far. To command a premium over memory, SEMCO must now prove longer earnings duration, stronger pricing power, repeat Si-Cap demand and high-margin FC-BGA/MLCC economics.
Existing holders can stay with the position. Fresh chasing should wait for evidence. From here, the stock is no longer about discovering a good components company. It is about collecting the future that the market has already started to price.
Evidence Classification
[Fact]
- Mirae Asset’s May 6 target framework used 2028F EPS of KRW 34,764 and a 37x PER. (Mirae Asset Securities)
- Based on the user-provided June 1 Mirae Asset report, the KRW 2.8M target uses 2029F EPS of KRW 93,242 and a 30x PER.
- Based on the user-provided June 1 report, 2029F operating profit was revised from KRW 4.22T to KRW 8.59T, with 2029F OPM at 29.7%.
- Samsung Electro-Mechanics announced an approximately KRW 1.5T silicon-capacitor supply contract for 2027-2028. (Samsung Electro-Mechanics)
[Inference]
- The KRW 2.8M target is driven by the 2027-2029 earnings revision, not by a higher multiple.
- SEMCO must prove longer duration than memory through contracts, ASP durability, Si-Cap repeat orders and high OPM.
- The key validation period has shifted from near-term earnings to 2027-2029 contract, pricing and yield evidence.
[Blocked]
- I could not verify a public URL for the June 1 Mirae Asset report. June 1 figures are therefore attributed to the uploaded user-provided original.
- Customer-level FC-BGA and Si-Cap contract terms, minimum volumes, prepayments and price-protection clauses are not publicly confirmed.