摩根大通看多安靠科技二季度业绩与AI先进封装盈利拐点

AMKR delivered a record 2Q and, more importantly, guided 3Q to another record
with gross margin stepping to levels the Investor Day model did not contemplate
until CY28, supporting our thesis that the AI-driven advanced packaging cycle
would inflect AMKR’s earnings power sooner than the Street appreciated.
Composition of the 2Q beat was notable - upside came from record compute and
auto/industrial, not smartphone seasonality, and margin expansion was driven by
utilization and mix rather than one-timers (with mgmt raising the bar again with
guidance that steps margins up another leg). Five takeaways drive our higher
estimates: (1) compute growth is shifting into high gear with record 2Q revenue
(+20% Q/Q vs. +MSD guided), accelerating to nearly +30% Q/Q in 3Q on the
HDFO datacenter CPU ramp and broad AI datacenter demand, lifting compute to
~27% of revenue (from 21% two quarters ago) at well-above-corporate
incremental margins (~60% implied fall-through in the 3Q guide); (2) the
unseasonal 3Q Comms decline (down HSD% Q/Q in what is normally a ~+30%
iPhone-build quarter - a ~$300M gap vs. seasonal) is half strategy and half market:
a deliberate SiP migration to Vietnam (~$150M revenue air pocket) that frees
Korea cleanroom for compute (revs return with the 2H27 launch cycle) and a
memory availability/pricing drag concentrated in Android (now just 15-20% of
comms, per JPMe), which we view as a mix trade that is margin-accretive either
way; (3) mgmt was explicit that the current business is “not margin-bound”: two-
thirds of 2Q’s +260bps of GM expansion came from utilization (now high-70s%),
3Q’s step-up is predominantly mix, and mgmt agreed with our framing that 4Q GM
biases higher still, net of a D&A step-up; (4) AZ Phase 1 is now fully committed
behind three named anchors (an Apple agreement, NVIDIA’s $1.5B prepayment
(2027 receipt), and the 10-year TSMC agreement) with Phase 2 decision work
currently underway, de-risking the $7B build and its funding; and (5) CY26 now tracks to ~$7.7B revenue, ~17.8% GM, and ~$2.73+ EPS (JPMe), hitting AMKR’s Investor Day 2028 targets two years early, with 2027 growing through a
manageable 1-2pt AZ OpM drag, and the 2030 model ($11B+ / 22%+ / $5.00+) in
our view increasingly shaping up to be a floor. We raise our estimates and establish
a Dec-27 price target of $85 (prior Dec-26 PT of $85). We maintain our Overweight rating on AMKR.
Summary Investment Thesis and Valuation
Investment Thesis
Our Overweight thesis is based on Amkor’s strong market position in the outsourced assembly and test manufacturing market and its differentiated portfolio of advanced packaging solutions. We believe the company is well-positioned to
intercept the strong underlying demand trends in semiconductor
and the technology transition towards more advanced
packaging. We see Amkor capturing more value in the
manufacturing supply chain over time on new packaging
innovation/complexities.
We see Amkor’s revenue growth outpacing industry growth on
1) Amkor’s strong technology/product differentiation in advanced packaging solutions, which are leveraged to the fast growing AI market; 2) increased emphasis by customers to
diversify their manufacturing away from Taiwan/China; 3)
continued strong share gains at its largest customers; and 4)
continued trend to outsource manufacturing capacity (due to
increasing manufacturing complexities).
Valuation
We establish a Dec-27 price target of $85 (prior Dec-26 PT of
$85) based on a ~25x P/E multiple (vs ~22-23x peer-group
median) applied to our CY27 EPS estimate of $3.47. We believe
Amkor should garner a premium multiple to its peer group given
its broad manufacturing footprint/geographical positioning,
strong leadership profile in advanced packaging solutions (75-
80% of sales), and increasing exposure to the fast growing AI
demand trends.