Morgan Stanley Previews Earnings for Five Semiconductor Companies: Strong Fundamentals, but Cycle Sustainability in Question
nashnova research
Morgan Stanley's latest semiconductor report previews five companies reporting this week, noting strong industry fundamentals — but the market's core debate has shifted from near-term numbers to how long the upcycle can last, which will set the next repricing.
Amkor: Nvidia committed $3 billion — what comes next?
Morgan Stanley rates Amkor (AMKR) "Equal Weight" with a $70 price target. Q3 gross-margin guidance of 18.5%–19.5% came in above prior second-half expectations.
Nvidia and Amkor announced plans to expand Korean interposer wafer — a middle layer connecting chips — capacity by over 300% through 2028, signing a new $1.3 billion deal that brings total commitments to $3 billion.
This means → AI packaging demand is real and scaling, but costs are stacking up: Arizona investment has grown to $12 billion, and management expects a 1–2 percentage-point operating-margin drag starting in 2027. The Arizona fab won't turn profitable until after late 2029.
In plain terms = orders keep getting bigger, but the new factory only burns cash in the near term. Whether margins hold through 2027 is the key question.
NXP: is the auto-chip shortage turning into pricing leverage?
Morgan Stanley rates NXP (NXPI) "Overweight" with a $338 target. Q3 results and Q4 guidance are expected roughly in line.
A distributor survey shows respondents expecting below-seasonal auto demand fell from 15% to 11%. Inventory is now below the historical median. This reflects a stabilizing auto-chip supply–demand balance.
63% of respondents report longer lead times; customers are locking in forward supply. This means → NXP enters 2027 pricing negotiations with more leverage: 55% report firm analog pricing, 54% report firm MCU — microcontroller, one of the most common chips in a car — pricing.
In plain terms = inventory at the bottom + slower deliveries = seller's market. NXP has room to hold prices in the next round.
Allegro: data-center surge, auto recovery ahead of peers
Morgan Stanley rates Allegro Microsystems (ALGM) "Overweight" at $52, expecting Q3 results at the high end of guidance.
The auto recovery is driven by higher content per vehicle, not restocking. June-quarter orders rose 30% year-on-year. This means → it is not a one-time inventory refill — each car genuinely uses more chips.
Data-center revenue grew 32% quarter-on-quarter in the June quarter, reaching 17% of total sales. Current sensors — components that measure power delivery to chips — surged 66% sequentially. Management expects data-center revenue to more than double in fiscal 2027; Morgan Stanley models 114% growth.
This reflects hyperscaler capex spilling over into "supporting-cast" chips like sensors.
Intel: server shortage props up the surface — PC and foundry hide two risks
Morgan Stanley rates Intel (INTC) "Equal Weight" at $84. The report dissects three core issues.
Servers look strong, but Morgan Stanley is "skeptical." AI-agent CPU demand has kept server processors in shortage. Intel is using the window to push double-digit price increases on select customers, likely driving a beat. In plain terms = the strong print comes from scarcity, not product dominance — once supply catches up, the pricing story breaks.
The PC market is deteriorating sharply. Q3 ODM assembly volumes fell 17% quarter-on-quarter (normal seasonality is +5%) and 26% year-on-year. Morgan Stanley forecasts 2027 PC shipments down 11%. This means → if PC revenue tracks shipments, consensus faces a $3–4 billion downward revision — and PC remains Intel's largest revenue segment.
Foundry faces a structural contradiction. 18A and 14A process development is on track, but if Intel's own CPUs are sold out through 2028, where does the spare capacity for external customers come from? This reflects a core tension — "we can't even meet our own demand" is the worst possible pitch to foundry prospects.
Sandisk: storage is strong, but this quarter's print won't be the catalyst
Morgan Stanley rates Sandisk (SNDK) "Overweight" at $1,750. The storage market is very strong, but this quarter's report is unlikely to move the stock.
Three months ago the market was deeply pessimistic on near-term pricing, yet Q3 and Q4 pricing has come in with strong double-digit increases. Data-center eSSD — enterprise solid-state drive — demand is durable; DRAM spec-downs have actually increased the need for high-speed local storage as cache.
Industry Q4 ASPs are expected to rise 15%–20%, though Sandisk's increase may trail slightly. This means → the company tends to guide conservatively during strong periods. Q3 will likely beat meaningfully, but Q4 guidance won't surprise.
Across all five — what is the core tension?
Morgan Stanley's overarching view: strong fundamentals are not in dispute, but "how long can the cycle last" has replaced "are this quarter's numbers good" as the market's central question.
Among the five, Intel's structural risks stand out most. Whether 2027 estimates can withstand PC weakness + foundry capacity contradiction will be the key repricing test after this earnings season.
NXP and Allegro offer a different angle — pricing power. Depleted inventory and lengthening lead times are shifting leverage to sellers. This reflects the next phase of the semiconductor cycle: from volume recovery to pricing negotiation.
市场有风险，内容仅供研究参考，不构成投资建议。