Moore Threads commercialization, Foxconn AI server revenue shift, Cerebras earnings plunge, CoreWeave A100 lifespan confirmation.
Moore Threads, the first domestic GPU stock, was sought after immediately upon listing, with its share price stabilizing above 579 RMB, and its commercialization path voted on by the market with real money. Revenue structure is shifting from "selling cards" to "selling compute rental," and the business model for domestic GPUs is being proven viable. The narrative for domestic compute is also switching from "seeking autonomy" to "reading financial reports"—today's near 5% intraday surge for Cambricon and over 3% rise for Zhongji Innolight in A-shares are results of this sentiment resonance.
Hon Hai's Q2 financial report structure changed: Cloud and network products (including AI servers) accounted for 51% of revenue, exceeding half for the first time; smart consumer electronics like iPhone dropped to 29%. Continued ramp-up of Nvidia AI server OEM orders is the core engine of this structural shift. The label of "Apple's contract manufacturer" is becoming obsolete, and the boom in AI hardware is transmitting from chips to the manufacturing stage.
Cerebras released its second post-IPO earnings report: raised full-year guidance, but hardware business declined, demand was "lumpy," causing shares to plunge about 14% after hours. The market is beginning to reprice the valuation of "Nvidia challengers." Single-customer dependency combined with hardware volatility makes the "Second Nvidia" story seem thin—AI chip competition is about delivery over consecutive quarters, not launch events.
AI cloud giant CoreWeave confirmed: Nvidia A100 chips can serve until 2029, with GPU lifespans far exceeding market expectations; but simultaneously warned that switching away from Nvidia chips would be difficult if forced. Long-lived GPUs extend the depreciation cycle of compute assets, benefiting long-term gross margins for heavy-asset cloud providers; and the longer-lasting the Nvidia ecosystem, the harder it is to replace—this is both a moat and a concentration point of systemic risk.
📰 EastVoice — today's AI headlines, decoded
Five minutes a day, the global AI shift →