Nvidia Blackwell AI GPUs Face Three-Month Delay Over Design Flaws
Nvidia reportedly pushes its highly anticipated Blackwell GPU launch into 2025 due to undisclosed design flaws, forcing major tech clients to wait.
Sources from both Nvidia and Microsoft reveal that Nvidia delays its upcoming Blackwell AI GPUs due to design flaws, pushing the expected launch from late 2024 into the first quarter of 2025 or later. Although Nvidia delivers engineering samples of the B100 and B200 chips this week, the company takes extra time to resolve these underlying hardware issues before beginning mass production.
This delay significantly impacts major tech giants like Microsoft, Meta, and xAI, which rely on these expensive chips to power their artificial intelligence operations. Meta already expresses doubt about receiving Blackwell shipments this year, and xAI faces a setback in its plan to acquire 300,000 B200 GPUs, with full order fulfillment now stretching into the second quarter of 2025.
Nvidia willingly takes this extra time to ensure product reliability, avoiding the catastrophic launch failures that recently plague competitors like Intel. With individual AI superchips costing up to $70,000 and complete server racks exceeding $3 million, Nvidia cannot risk a massive hardware recall. Furthermore, Nvidia holds a dominant 98 percent share of the data center GPU market, giving the company the flexibility to delay its release without losing ground to rivals.