Before modern GPUs, retro consoles relied on dedicated graphics chips designed to draw specific elements like sprites and backgrounds. These chips were efficient at their narrow tasks, but they could not be reprogrammed to handle new kinds of visual effects. Their work was hardwired into the hardware.
The Engadget article explains that the key change came with the original Xbox. Its graphics processor could not only draw but also perform computation, meaning developers could write shader programs that ran on the GPU itself. This opened the door to lighting, reflections, and other effects that fixed-function hardware could not achieve.
That architectural shift had lasting consequences. The Xbox's approach made programmable shaders standard in consoles, and later generations pushed GPUs further toward general-purpose parallel computing. The history shows that the modern GPU is less a pure drawing machine and more a general-purpose processor optimized for graphics.