From hypothesis to reality: enabling the future of AI with vertically integrated memory

Massive investments in cloud computing have exposed a critical shortcoming in modern AI infrastructure: the immense energy cost of moving data. To truly enable the future of compute, we must fundamentally change how memory is built and integrated into systems. Building a new semiconductor technology requires more than just scientific discovery: it demands rapid execution, collaboration with the entire ecosystem and a bold vision for the future. This presentation details the fast-paced evolution of a breakthrough memory concept into a tangible technology demonstration, and how we are enabling the future of AI compute through vertically integrated memory chiplets.