The Startups and Technologies That Could Truly Enable an AI Revolution (An IP Perspective)

Over the weekend, multiple reports highlighted that both Nvidia and Oracle are making large investments in OpenAI to scale and enhance their AI technologies. Some news stories even mentioned that a 10-gigawatt buildout could be necessary to support these operations, a level of energy demand equivalent to powering a city like New York in summer. 

Meeting this demand may require significant infrastructure investment, including new nuclear plants or emerging energy technologies.

If this is the path forward, it raises a critical question: what types of IP and technology will ensure these data centers don’t become the office-space equivalents of tomorrow as a cautionary example of massive capital investments that quickly become outdated?

The U.S. Needs to Rethink “Abstract Subject Matter” Rejections

Since the Supreme Court’s ruling in Alice Corp. v. CLS Bank International (2014), many software and computing inventions have been invalidated under 35 U.S.C. § 101 for being “abstract ideas.” 

More recently, rejections have extended to inventions grounded in physical principles. For example, in American Axle & Manufacturing, Inc. v. Neapco Holdings LLC, a prop-shaft manufacturing method that accounted for vibrations was deemed unpatentable as a natural application of Hooke’s law.

As new technologies emerge that leverage previously unknown scientific principles, there is a real risk that courts or the Patent Office may strike down innovations on similar grounds. Such legal uncertainty could make even large-scale technological investments difficult to monetize.

Where to Focus for the Next Wave of Innovation

Energy production, material science, and advanced computing will define the feasibility of large-scale AI infrastructure. Incremental improvements to existing systems alone will not suffice. 

Areas that warrant attention and investment include:

Energy production: Nuclear fusion and other approaches to safe, cost-effective energy could meet the demands of both AI data centers and society at large. Expanding traditional nuclear infrastructure faces NIMBY challenges and long lead times.

Quantum computing: While still nascent, breakthroughs here could reduce reliance on massive GPU farms, lowering costs and energy consumption.

Material science: Developing advanced materials, particularly alternatives to rare-earth metals, will enhance performance, reduce dependency on foreign supply chains, and mitigate environmental risks from mining.

Universities and research institutions: The U.S. must attract the world’s top talent to universities, startups, and companies. Building the underlying systems that make large-scale AI feasible requires cutting-edge research and innovation across multiple disciplines. 

The next AI revolution won’t just be about software, it will be powered by energy, materials, and legal frameworks.