At the Citi Global technology conference, the presenter tried to extract from NVIDIA CFO Colette Kress at least some hints of the proximity of the announcement of a new graphics architecture, which she had repeatedly sinned at past similar events. The experience made Ms. Kress less talkative, she only clarified that Ampere could stay on stage after the release of a successor.
This comment related specifically to the company’s plans in the server segment, since the representative of NVIDIA referred the peaceful coexistence of Volta and Ampere architectures. The first is not only present in client server systems, but also continues to be supplied by NVIDIA as part of serial products. On the other hand, carriers of the Ampere architecture now account for more than half of all NVIDIA revenues in the server segment, so the pace of expansion of the new architecture can be considered quite high.
At this event, the CFO of NVIDIA often referred to the fact that video cards of the GeForce RTX family, related to the Turing and Ampere architectures, now form no more than 20% of the fleet of operating video cards of this brand, and therefore the backlog of future upgrades is very large. The presenter used this thesis to ask about the relationship between the level of market penetration of the current architecture and the timing of the announcement of successors.
Colette Kress explained that NVIDIA can switch to a new architecture before the previous one gains a certain “sufficient” number of supporters. The company is working daily to create not only the next generation architecture, but also the one that will replace it in an even more distant future. The time of the announcement is determined by the degree of readiness of the new architecture to enter the market.