Nvidia posted a blowout quarter on August 26, reporting $96.2 billion in revenue and guiding Q3 to roughly $108 billion, a report that pushed its shares up by more than 5% in aftermarket trading. Despite that broadly strong headline, Asia-listed companies that feed Nvidias supply chain responded unevenly.
Korean memory stocks advanced notably, with SK Hynix and Samsung both recording gains. Taiwanese suppliers posted smaller increases, including TSMC and ASE. In contrast, Japanese equipment manufacturers such as Advantest and Disco fell, reflecting a market that parsed Nvidias commentary on margins and input costs rather than focusing only on the top-line beat.
The margin passage that shaped market reaction
Among the details in Nvidias release was guidance for Q4 gross margins to slip to a 71-72% range before firming toward 72-73% in fiscal 2028, "as higher memory prices take effect." That phrasing became the key interpretive lens for investors and traders assessing where demand and pricing power sit across the supply chain.
For memory manufacturers, the implication was straightforward and positive: Nvidia appears to be absorbing higher HBM costs rather than shifting them fully onto customers, and Nvidia explicitly described memory pricing as "extreme." That dynamic validates stronger pricing for memory suppliers and supports their recent rallies.
For makers of semiconductor manufacturing and testing equipment, the margin signal reads differently. If Nvidias margins are compressing because of rising input costs, market participants infer that hyperscale customers could press for concessions elsewhere in the chain. Equipment vendors, which rely on continued aggressive capital spending to expand wafer starts and test capacity, are being valued against a scenario in which the most active phase of capital investment may moderate.
Why test-equipment names underperformed
Advantest stood out as the largest decliner among the group. The stock had moved higher the previous day as markets anticipated a strong Nvidia print, but the drop after the results fits a classic "buy the rumor, sell the news" pattern. Beyond that trading dynamic, there is a structural connection: Advantest produces GPU test systems used on every Nvidia chip. Nvidias margin commentary suggests it may moderate the pace of incremental capacity additions as it works down current orders. A slowdown in new chip starts would eventually translate into fewer orders for test systems.
Foundry reaction was restrained
TSMCs relatively small move upward was notable given that it manufactures all of Nvidias GPUs. The muted response reflects two factors emphasized by market participants: Nvidias AI-related demand story for TSMC is already well known and priced in, leaving less room for an upside re-rating; and the same margin compression that benefits memory vendors could in time be a source of pricing pressure on foundries if Nvidia seeks wafer-cost concessions to protect its margins.
That said, there is also an explicit opportunity for TSMC to expand output to meet rising orders for advanced AI processors, a dynamic that underpins the longer-term link between Nvidias demand and foundry activity.
Packaging and assembly results diverge
Packaging firms showed a split performance. ASE recorded a modest gain while Powertechs response was weaker. The difference aligns with exposure: advanced packaging solutions such as CoWoS benefit ASE more directly, whereas Powertech has a larger footprint in legacy integrated circuit testing, a segment that is not as directly tied to the recent AI-driven surge.
Bottom line
Nvidias earnings were broadly bullish for the company itself and for parts of its supply network, but the results were not uniformly positive across the ecosystem. The key margin disclosure set off a selective market reaction: companies with immediate pricing power in memory were rewarded, while those reliant on continued heavy capex for future growth faced greater investor caution. The market is interpreting margin compression as a potential signal that the most intense phase of capacity expansion could be leveling off, even as demand tied to AI remains strong.