RESEARCHOpen ThesisMemo 
The investment question is when capacity becomes usable
NVIDIA’s October 21 supply-chain announcement gives investors a set of manufacturing milestones to monitor. Our reading is that it strengthens the case for a more deliberate supply expansion, while leaving the timing of incremental finished-system revenue unresolved. Announced capacity and usable output belong in different columns of a research model.
NVIDIA targets aggregate substrate capacity of 2.5 times the current level over three years. It also expects semiconductor testing capacity to double from calendar 2026 to 2027. These describe different production stages. They are not two independent increases that can be added together to estimate GPU shipments.
The practical research task is to identify which stage constrains a particular product and when that constraint might ease. A larger upstream footprint helps only if the remaining steps can absorb the output. Investors need evidence about the whole production path before applying a capacity multiplier to sales.
Samsung’s agreement gives a separate supplier perspective
Samsung Electro-Mechanics’ own October 21 announcement describes joint investment in dedicated ABF substrate facilities in Sejong and Busan, South Korea. Deliveries to NVIDIA are scheduled to begin in 2026. The agreement covers compute chipsets and high-speed networking switches, so its relevance extends beyond a single processor package.
The supplier expects double-digit annual volume growth over a multiyear period and says NVIDIA’s allocation could become a significant portion of its annual package-substrate capacity by 2028. It does not disclose an exact percentage. Samsung Electro-Mechanics also identifies fine circuit patterning, warpage control and substrate flatness as manufacturing challenges affecting yield and reliability.
For an investor, that creates a useful distinction between installed equipment and qualified output. A factory expansion can be strategically valuable before it produces a predictable earnings contribution. The announcement supports monitoring this supplier relationship; it does not supply the utilization, pricing or margin assumptions needed for a complete profit forecast.
Continue your research with ThesisMemo.
Explore NVIDIA research Free preview · No sign-up requiredKeep locations, dates and cash arrangements separate
NVIDIA says a $1.1 billion capacity reservation advance will support KYEC’s new US testing facilities and equipment, with high-volume production targeted for April 2027. Its Amkor plan concerns South Korean interposer wafer capacity, targeting an increase greater than 300% by the end of 2028. Both are future milestones.
A separate Amkor release, dated July 23, 2026, describes a $1.5 billion multiyear advanced-packaging and development agreement with NVIDIA. That earlier announcement concerns US expansion in Arizona, supported by a prepayment. It is background, not another October 21 transaction, and should not be confused with the South Korean expansion or KYEC’s advance.
Amkor’s July release explicitly warns that the Arizona campus may not meet expected construction timing, cost, specifications or sales outcomes. That qualification matters when evaluating the broader expansion strategy. Geographic diversification can improve resilience, but the presence of several locations does not establish that each will ramp on schedule.
What would change the assessment
The next useful evidence would be supplier confirmation that facilities have entered the promised production phase, followed by indications that customers are taking the resulting output. An announcement that equipment has arrived would answer a narrower question than evidence of sustained, qualified deliveries. Research notes should preserve that difference.
The positive scenario is coordinated execution: the relevant manufacturing stages become available when customers need the systems, allowing additional capacity to turn into deliveries without a new bottleneck. The weaker scenario is a mismatch between readiness and demand, or between one production stage and another. Those are analytical scenarios, not outcomes reported in these releases.
For NVDA, review future financial disclosures for the relationship between manufacturing commitments, cash deployed and realized sales. For AMKR, separate the specific facility and agreement being discussed before changing a model. The three sources here come from transaction participants; they establish what the companies announced, not independent proof of future execution. Until operating evidence arrives, the sensible update is a milestone checklist rather than a mechanical revenue upgrade.
Keep following NVIDIA after this article.
Explore company insights, recent news and SEC filings in one place. Start with the free research preview and follow the original evidence.
Explore NVIDIA research Free preview · No sign-up required- NVIDIA: October 21 supply-chain capacity announcement ↗
- Samsung Electro-Mechanics: October 21 ABF substrate partnership ↗
- Amkor: July 23 US advanced-packaging agreement (background) ↗
Research and education, not personalized investment advice. Facts and examples should be read in the context of their stated periods. Follow the original disclosures for the complete record.