Nvidia chief financial officer Colette Kress told the UBS Global Technology and AI Conference that the industry is not inflated fluff but going through a massive shift from CPUs to GPUs.

Kress said: “Yes, it's a very interesting discussion as a lot of words are really focusing on some very interesting thoughts regarding the supposed AI bubble. No, that's not what we see [...]. But what our focus is on is transitioning that to GPUs. It's necessary because there's just not going to be any improvement we can see in other CPU usage. So that's one of our first pieces”.

She was pressed on whether ASIC competitors are chipping away at Nvidia’s lead. Kress said the outfit is not pursuing a single narrow AI task but is running the entire development process from training to inference.

She boasted that Nvidia has “seven different chips” working together to build an accelerated computing environment while ASIC makers stick to a single-purpose model.

Kress argued that this mix of silicon, sprinkled with Nvidia’s beloved co-design buzzwords, leaves ASIC rivals looking rather lonely. CUDA remains the real moat, she stressed, claiming it has delivered an “X factor improvement” across libraries and keeps customers glued to Team Green.

The big reveal for the faithful was confirmation that the next-generation Vera Rubin architecture has taped out. “Yes. So, Vera Rubin, we're pleased to say it's been taped out. We have the chips and are working feverishly right now to get ready for the second half of next year to bring that to market. We're very pleased with what occurred with Ultra,” she said.

Rubin is among the most hyped AI releases on the horizon. With chips and networking gear already flowing through the early pipeline, Nvidia looks set for mass production by the second half of 2026. The outfit is betting that this will keep its GPU dominance humming even as rivals push their own hardware stacks.