For years, the technology industry's been working to extract enough performance from silicon to make modern-day artificial intelligence (AI) possible. Now that it's done so, it's using that very same AI to discover new materials for a future generation of computer chips.
That's the big takeaway from the recent fundraising done by start-up CuspAI, anyway. It raised $450 million -- valuing the company at $2.6 billion -- from a handful of private investors, including Amazon (AMZN +15.32%) founder Jeff Bezos' private venture fund Bezos Expeditions (which also owns Blue Origin, The Washington Post, and more). Chipmakers Nvidia (NVDA +2.93%) and Advanced Micro Devices (AMD -1.90%) are also developmental partners, along with Facebook parent Meta Platforms (META +3.28%).
At first blush, it doesn't seem earth-shattering; most tech organizations are perpetually working on building a better (proverbial) mousetrap. This is one of those developments, however, worth putting on your radar even if it's going to take a while to bear any fruit.
Image source: Getty Images.
Opportunity knocks
It's not a complicated company. In its infancy, CuspAI was working on cost-efficient carbon capture and more effective water purification processes, both of which start with superior materials designed and tested by AI. The rapid proliferation of AI, however, has also highlighted shortages in its supply chain.
For instance, the International Energy Agency predicts that data centers' demand for gallium in 2030 could be 10% greater than today's supply. In the meantime, China controls 99% of the planet's refined gallium supply, leaving the rest of the world vulnerable to geopolitical tensions.
That being said, several other key elements needed by AI computing platforms, including iridium and tantalum, are also increasingly difficult -- and costly -- to obtain, limiting the industry's continued growth.
CuspAI's artificial intelligence platform can't create these critical metals, to be clear, or for that matter, procure them. It can, however, virtually identify alternatives that would work just as well, or even develop ways to synthesize a material that, up until this point, has been impossible to synthesize. Work that would normally take years and millions of dollars to complete can now be done in months at a fraction of the cost.
Yes, it's a relatively new and unproven business. But it's one that's expected to thrive, if only because it's rooted in real need. Outlooks from both Mordor Intelligence and Precedence Research suggest that the generative AI material science industry will grow at an average annual pace of more than 24% at least through 2030, if not longer.
This isn't Bezos' first foray into extracting more performance from generative AI platforms either. In November, Bezos Expedition made an investment in AI start-up Prometheus, which will help engineers design and manufacture complex physical products, perhaps even including microchips using materials discovered by CuspAI. It's another indication that there are still significant technological leaps to be made on all AI fronts, separately and collectively.
Not an investment option... yet
Unfortunately, CuspAI isn't publicly traded. There aren't even any whispers of an IPO in the foreseeable future.
Still, the idea of AI-designed materials for use in AI computing hardware is worth putting on your radar simply because it could be so game-changing -- in five to 10 years -- for companies like Nvidia or AMD, or perhaps even another chipmaker like Navitas Semiconductor (NVTS -1.36%). It's already done extensive developmental work with gallium nitride and could make good use of a cost-effective alternative to the expensive element if it became available in bulk at a good price.
