Why materials science is key to unlocking the next frontier of AI development

But this pace of innovation is not guaranteed, and the next frontier of technological advancement—from an artificial intelligence future to new computing paradigms—will only happen if we think differently.

Atomic Challenge

Modern microchips push the limits of physics and reliability. This is atomic precision. A few atoms can determine the functionality of the entire wafer. This engineering marvel is the result of more than 50 years of exponential expansion to create faster, smaller transistors.

But we are reaching physical limits of size, costs are growing exponentially with complexity, and efficient power consumption is becoming increasingly difficult. At the same time, artificial intelligence has increasingly higher requirements for computing power. Data from Epoch AI shows that the amount of computing required to develop artificial intelligence is rapidly surpassing Moore’s Law, doubling every six months since the “deep learning era” in 2010.

These interconnected trends create challenges not only for industry, but for society as a whole. Without new semiconductor innovations, today’s AI models and research will lack computing resources and struggle to scale and evolve. Key areas such as artificial intelligence, self-driving cars and advanced robots will encounter bottlenecks, and the energy consumption of high-performance computing and artificial intelligence will continue to soar.

Material intelligence

At this inflection point, a complex global ecosystem—from foundries and designers to highly specialized equipment manufacturers and materials solution providers like Merck—is working together more closely than ever before to Find answers. All of these can play a role, and the role of materials goes well beyond the silicon that makes up the wafer.

Instead, materials intelligence is present at nearly every stage of the wafer production process, whether it’s chemical reactions that carve circuits at the molecular scale (etching) or adding incredibly thin layers to a wafer with atomic precision (deposition): The layers in the leading edge node are 25,000 times thicker than a human hair.

Yes, materials provide the physical basis for wafers and the substance for more powerful, compact components. But they are also integral to the advanced manufacturing methods and novel chip designs that have underpinned the industry’s rapid growth in recent decades.

Therefore, materials science becomes increasingly important as we work to address the limitations of miniaturization. More than ever, the industry needs advanced materials to unlock new designs and technologies that increase chip efficiency, speed and power. We see novel wafer architectures employing three dimensions and stacked layers to optimize surface area usage while reducing energy consumption. The industry is leveraging advanced packaging technology to merge independent “chips” with different functions into more efficient and powerful single chips. This is called heterogeneous integration.