The leaker suggests the delay is due to a focus on enterprise silicon, specifically the Vera Rubin accelerator, which is expected to generate $20 billion in Q3 2027. Both the current Ada Lovelace and upcoming Blackwell architectures in the RTX 40 and RTX 50 series GPUs share the same TSMC 4N process node, so it would take Nvidia six years to move its consumer GPUs to the more advanced TSMC 3N or 3NP node used by Rubin accelerators. The delay could also be due to the potential sidelining of an RTX 50 Super refresh, which might have been cancelled due to the use of denser 3GB GDDR7 VRAM chips. This is particularly bad news for consumers, as the RTX 5090 is already experiencing a pricing squeeze, with costs soaring above $9,000.

The delay comes as Nvidia's highly profitable data centre arm is experiencing soaring demand from the ongoing AI buildout, meaning there is less onus to deliver a potentially less profitable consumer GPU. It is not unusual for enterprise products to use more advanced nodes before arriving in consumer products, but this time around, the Blackwell product cycle is poised to be unusually long. For now, consumers are advised to hold on to their current GPUs unless they are willing to pay the dizzyingly high costs that the market demands right now.