The transition to programmable shaders was risky and nearly broke the company, but it ultimately led to significant advancements in gaming technology.

For the first five years, we ignored customer feedback to focus on making our graphics processors better, believing that 3D graphics were insatiable.

The invention of CG (C for Graphics) programming paradigm opened new avenues for using GPUs beyond gaming, leading to breakthroughs in general-purpose computing.

Our first programmable chip, GeForce effects, was a gamble that almost led to our downfall, but it was necessary for our survival.

In 1993, we believed that the PC would evolve to run 3D graphics programs, allowing users to explore new worlds and play games.

One breakthrough in graphics was designing a pipeline that avoids the triangle rasterization hardware in GPUs, allowing for better results through pixel shader calculations and ray tracing.

We made these chips programmable via TensorFlow, allowing models to run on CPUs, GPUs, or TPUs with minimal modifications.

In the middle of 2022, we realized that the transformer models were going to fundamentally disrupt our GPU virtualization business, so we decided to pivot the company to focus on AI applications instead.

We're witnessing a fundamental transition in computing, moving from general-purpose computing to generative neural network processing, which opens up unthinkable applications.