Morgan Stanley Raises Astera Labs Price Target to $389, Bullish on AI Interconnect Platform Expansion
nashnova research
Morgan Stanley lifted its Astera Labs price target to $389, implying 74% upside, arguing the company is evolving from selling individual chips to selling rack-scale AI interconnect solutions — but whether the stock re-rates hinges on the market's confidence that 2027 growth can move beyond a single customer.
How strong were the results?
Q3 revenue hit $584.9 million, up 153.7% year-over-year. Q4 guidance midpoint came in roughly 12.8% above consensus.
This means → top-line momentum is not the issue. The drag is external: semiconductor valuations are compressing and AI financing costs are rising. Strong earnings do not automatically translate into a higher stock price.
In plain terms = the student aced the exam, but the whole grading curve shifted down, so the reward for a high score shrank.
From "connecting chips" to "connecting full racks" — what matters most to Morgan Stanley?
Astera Labs started with Aries retimers — chips that restore high-speed signal integrity — solving point-to-point data quality between processors.
The product line now spans three layers: Scorpio switch chips handle cluster-level communication, Leo controllers expand usable memory, and COSMOS software manages the entire switch fabric and server rack.
Orion, expected in 2027, goes further — bundling chips, boards, compute trays, switch trays and memory trays into a rack-scale reference architecture delivered through OEM partners. Management estimates a fully loaded 80-GPU rack can carry roughly $100,000 in Astera Labs product, software and licensing value.
This means → the business model is shifting from "selling components" to "selling full-stack solutions plus software licenses," sharply raising per-customer value.
What underpins the 2027 growth story?
Morgan Stanley's confidence rests on multiple product curves arriving in sequence, not a single catalyst: a second Scorpio X hyperscaler customer is still in pre-production; additional customers are expected to qualify in H1 2027 and contribute revenue in H2; Leo memory products should ramp mid-year; near-package optical interconnects — technology that uses light instead of electrical signals to move data between chips — are expected to begin contributing in H2.
The Scorpio family already generates annualized revenue above $1 billion, with the X series outpacing the P series. Management says the next-generation switch chip alone could reach $1,000+ per accelerator in value.
Morgan Stanley raised its 2026–2028 revenue forecasts to $2.03 billion, $3.40 billion and $4.41 billion, implying year-over-year growth of 138.1%, 67.5% and 29.7%. Adjusted EPS estimates are $4.37, $7.00 and $8.64.
Why are Wall Street firms maintaining buy ratings yet cutting targets?
Post-earnings, several banks kept their buy ratings but lowered price targets: RBC cut from $500 to $400, Stifel from $460 to $370, Jefferies from $450 to $350, Needham from $425 to $300.
This reflects a shared conclusion: no one doubts the growth, but the valuation premium the market is willing to pay is shrinking. Rising AI financing costs are compressing the multiples investors will accept.
In plain terms = everyone agrees the company will earn more money, but they are simultaneously marking down how much each dollar of profit is worth in stock price.
What worries the cautious camp?
UBS held its "Neutral" rating and trimmed its target from $300 to $290, focusing on AWS's future architecture choices — if Amazon shifts to in-house alternatives, Astera Labs's switch-chip share could erode.
TD Cowen kept "Hold" and cut its target from $300 to $225, the lowest among major brokerages.
This means → the real debate is not "can Astera Labs grow?" but "can 2027 growth diversify beyond heavy reliance on Amazon's Trainium platform?" New customer production timelines, Leo and optical-interconnect revenue scale-up are the key proof points ahead.
市场有风险，内容仅供研究参考，不构成投资建议。