Nvidia has canceled the RTX 4090 Ti

In the past, rumors abounded that Nvidia intended to launch the TITAN or GeForce RTX 4090 Ti, founded on the Ada Lovelace architecture, and had prepared corresponding Printed Circuit Board (PCB) designs and thermal solutions. There were even exposures of its colossal cooling module that occupies four slots.

Given multiple considerations such as volume, weight, heat dissipation, and power consumption, it was circulated that Nvidia had ultimately chosen to abandon the project. This decision might also correlate with AMD not exerting sufficient competitive pressure. Recently, these suppositions have been corroborated by Twitter user @kopite7kimi, reconfirming the high probability of the GeForce RTX 4090 Ti’s non-existence. Considering that the current GeForce RTX 4090 doesn’t employ a complete AD102 chip, it remains uncertain whether there will be a TITAN based on the Ada Lovelace architecture. However, such a plan seems to be off the table for the moment.

In news more pertinent to the average gamer, Nvidia is likely to deploy the AD103/AD106 in the GeForce RTX 4070/4060. However, these will not be entirely new models, but rather strategic maneuvers to optimize chip shipments, primarily targeting the Chinese market. Given Nvidia’s past practice of repurposing high-end Ampere architecture chips for more mainstream models, the same operation could be replicated on the GeForce RTX 4070/4060.

Additionally, a monumental design shift could take place in Nvidia’s next-generation GeForce RTX 50 series flagship models, potentially incorporating a 512-bit wide memory design. The latest GDDR7 memory has already achieved a speed of 32Gbps, which means the memory bandwidth could reach an astonishing 2TB/s, doubling the bandwidth of the GeForce RTX 4090. According to Nvidia’s timeline, gaming graphics cards employing the next-generation consumer architecture are slated for arrival in 2025, rumored to be codenamed ‘Blackwell’, and will shift to a Multi-Chip Module (MCM) packaging.