From Accidents to Acceleration: How AI is Revolutionizing Materials Discovery
For most of human history, the discovery of new materials was driven by luck and tireless trial-and-error.
In 1839, Charles Goodyear’s eureka moment led to the accidental discovery of industrial grade rubber. When he dropped a mix of rubber and sulfur onto a hot stove, the mixture hardened rather than melting – unlocking a foundational material for modern transportation and industry.
Other discoveries required exhaustive persistence. In the 1870s, Thomas Edison tested more than 2,700 materials before finding the right filament for the lightbulb.
For generations, this was the innovation formula: sweat, serendipity, and iteration.
Not anymore.
We’re entering a new era in materials science: AI can hasten what used to be years-long processes into ones that now unfold in hours.
For example, in just 80 hours, Microsoft and Pacific Northwest National Laboratory screened 32 million materials to identify promising battery candidates. Meanwhile, Georgia Tech and Meta’s OpenDAC project is building an open-source database of millions of molecules for carbon capture, helping drive Direct Air Capture toward gigaton-scale viability. Scientists have also used generative AI to propose 23 new materials for solar-powered water splitting, a potential leap for green hydrogen.
But even once they’re discovered, energy technologies often take 30 or more years to reach commercialization, and another 20 years to achieve mass market adoption.
And that’s a problem, because we don’t have 50 years.
Early-stage clean energy R&D urgently needs a boost to tap AI’s full potential. Without targeted support, the sector risks missing out on the kinds of scientific breakthroughs already reshaping other fields.
Case in point: In 2022, DeepMind’s AlphaFold achieved a 45,000x acceleration in solving protein structures, transforming drug discovery. This same work would have taken until the end of this century using purely experimental methods. And it only happened because of a shared commitment to open-access data, cross-lab collaboration, and a willingness to bridge the worlds of machine learning and biology.
AI can similarly revolutionize clean energy materials, but the energy industry is slow on the uptake: only 2% of energy startups currently have an AI-related value proposition, compared with 7% in life sciences and 4% in agriculture.
At the Bezos Earth Fund, we’re proud to help begin leading this charge. We’re betting on bold, frontier technologies, especially where early philanthropic support can de-risk transformative ideas. But realizing this potential requires collaboration among experts and disciplinaries who, to date, rarely work side-by-side.
This is why the Earth Fund is bringing together leading voices in AI, materials science, and clean energy. At a recent event in DC, called AI x Materials Innovation: Breakthroughs for a Clean Energy Future, we explored what it will take to turn AI’s potential into progress.
So what will it take? Above all, scientists, software engineers, and energy experts will need to work together in new ways. By building bridges through tough-but-fruitful conversations, the Earth Fund is fostering this kind of collaboration so we can expand access to experimental materials testbeds, create standardized, open-access datasets, build pipelines between national labs, startups, and universities, and spur the innovations of tomorrow.
What early-stage R&D, if de-risked through strategic philanthropic support, can unlock game-changing advancements?
Imagine greener steel and cement that don’t drive 16% of global emissions. High-performance materials leading to 10x cheaper carbon capture. Gigawatts of renewable energy from superhot geothermal. Batteries that make affordable, long-duration energy storage a reality at scale.
The road to our clean energy future can be radically shortened by using AI to accelerate materials innovation, but only if we work together to drive progress. By combining the right questions, the right people, and the right tools, we can revolutionize discoveries that will support human progress for millennia to come.