Nvidia confirms a “rare” manufacturing issue affecting under 0.5% of RTX 5090 and 5070 Ti GPUs, which are missing render units, the latest hitch since launch
Where did the ROPs go? … In the grand scheme of things, that doesn't sound like a lot of affected GPUs …
Context & Ripple Effects
Nvidia’s recent product coverage has included deliberate configuration differences, such as the China-specific RTX 4090 D with fewer processing cores. This case is different: Nvidia describes an unintended production defect in a small share of current high-end cards.
The incident matters because the affected models sit at the top of Nvidia’s consumer lineup, where buyers expect the advertised hardware configuration and board partners must handle any corrective process.
First-order effects
- Owners of affected RTX 5090 and RTX 5070 Ti cards receive hardware with missing render output units, reducing the configuration from the product specification.
- Nvidia and its board partners must identify affected inventory and support replacements or other remediation for a defect Nvidia says is below 0.5% of units.
Second-order effects
- Retailers and board partners may face added verification, return, and customer-support work even though the stated affected share is small.
- The episode makes clear that a card’s reported hardware configuration—not just its model name—is relevant for buyers evaluating early production stock.
Third-order effects
- If launch-period component-configuration defects recur, premium GPU launches could increasingly require more transparent validation and replacement processes across Nvidia’s partner channel.
- The broader constraint is execution: product segmentation can be intentional, as with the reduced-core RTX 4090 D, but manufacturing escapes can erode confidence in that differentiation unless detection and remediation are reliable.
The trend: Premium GPU competition is making launch quality control and channel remediation more consequential alongside performance and product segmentation.