Why The AI Race Is Also A Chemistry Race

While most discussions about AI focus on models and parameters, the reality is that every advanced processor depends on carefully engineered materials. Ajinomoto, the Japanese company best known for MSG, holds a near-monopoly on Ajinomoto Build-up Film (ABF), which insulates the dense wiring in AI chip substrates. Since its adoption by a leading CPU maker in 1999, ABF has captured over 95% of its market, reporting operating margins above 50% in 2026 as the AI boom accelerated.

This dependency exposes a critical blind spot in AI development. The industry's reliance on materials has grown dramatically—from about 12 chemical elements in the 1980s to more than 60 by the 2000s, and now over 85% of all non-radioactive elements according to a 2025 preprint. Some material problems, like ABF supply, may be solved through scale and capital investment, but others represent fundamental physics barriers.

Physics Barriers

Three major physical walls challenge further advancement: wiring, heat, and space. As chips shrink below 17 by 17 nanometers, copper wiring begins behaving like a resistor, requiring ruthenium which has lower resistivity. For cooling, Nvidia's thermal engineers have identified two-phase direct-to-chip cooling as essential for next-generation accelerators, but this relies on PFAS 'forever chemicals' that 3M exited manufacturing by the end of 2025.

Space computing presents unique challenges. On October 1, 2026, Google launched a prototype satellite with four AI chips to test performance in radiation and extreme thermal environments. Without air or water for heat dissipation, space acts as an insulator, requiring better heat-radiation coatings—another materials problem that must be solved before orbital data centers become viable.

Development Mismatch

Nvidia has committed to what CEO Jensen Huang calls a 'one-year rhythm' for new AI chips, while materials typically require 10 to 20 years to move from discovery to deployment. This creates a dangerous mismatch between an industry moving at software speed and its foundation that advances at chemistry pace. The solution may lie in AI itself helping accelerate materials discovery.

Investment Shift

Significant resources are now flowing into materials research. In November 2025, the U.S. launched the Genesis Mission, a Department of Energy initiative linking national lab supercomputers with AI-directed experiments. By July 2026, federal commitments exceeded $5 billion to expand this effort. Meanwhile, Google DeepMind announced plans for a UK lab where robots could create and test hundreds of materials daily.

Despite these advances, the simulation-to-reality gap remains substantial. In 2023, Google DeepMind's GNoME model predicted 2.2 million new crystal structures, but UC Santa Barbara chemists found 'scant evidence for compounds that fulfill the trifecta of novelty, credibility, and utility.' As Dr. Ismail notes from his research experience, materials often reshape themselves during reactions in ways that simulations cannot predict.

For technology leaders, the path forward requires mapping AI roadmaps to their material needs, treating experimental data—including failures—as strategic assets, and demanding physical proof for materials claims. The AI race and chemistry race have become one and the same, with the next breakthroughs likely coming from those who recognize this fundamental connection.