Scientific Computing Is Strategic Infrastructure
Why we believe regional scientific computing hubs can expand scientific ambition and strengthen national security
By Hangar | August 4, 2026
At Hangar, we work where technology meets problems of national importance. Few problems now sit closer to the intersection of science, government, and security than access to the computing behind modern artificial intelligence.
Artificial intelligence is helping researchers bring some of science’s hardest questions within computational reach. A physicist may need to model a system too complex to reproduce experimentally. A materials scientist may want to examine thousands of molecular structures before fabricating one. Medical researchers can use AI to find signals in images and datasets that are difficult for people to see. Similar tools are changing energy research, aerospace, chemistry, and the study of AI itself.
The constraint is access. Many researchers still cannot obtain enough computing time, move large datasets efficiently, or get the engineering help required to apply advanced systems to their work. That gap does more than slow individual projects. It limits the range of questions American science can practically ask.
The U.S. National Science Foundation’s new State and Regional Artificial Intelligence Infrastructure Hubs program addresses that constraint directly. It includes an approximately $100 million funding opportunity that asks states and regions to organize new AI and scientific computing hubs for researchers at universities and nonprofit research institutions.
We think that is important because we have helped do it before.
New York moved first. It did not wait for a federal program to prove that shared AI infrastructure could be built around the needs of researchers.
From computing equipment to a scientific institution
In New York, Hangar and the Secunda Innovation Fund partnered with Governor Kathy Hochul, research institutions, philanthropy, and industry to turn an ambitious idea into Empire AI. The consortium now has $550M+ committed behind it.
The build moved quickly. Alpha began serving researchers within months of launch. Beta is now fully live. It is the only NVL72 cluster anywhere devoted to scientific research computing. That capacity is supporting work such as modeling how neural circuits process uncertainty and drive decisions and improving protein structure prediction for drug discovery.
What has taken shape around the machines is just as important.
The computing clusters share a low-latency storage environment, so researchers can use different systems without repeatedly relocating their data. Research software engineers help translate scientific objectives into computational workflows. Security is built around both the machines and the information they contain.
That operating model is what transforms powerful equipment into a scientific instrument.
Researchers need help using the platform, and faculty need to know what the systems make possible. Education programs have to grow the next generation of scientists and engineers. Clear access rules and durable financing matter, too.
Hardware attracts attention. Institutions create lasting scientific capacity.
The science—and the national opportunity
We are especially interested in AI models that help scientists understand phenomena too fast, small, or complex to observe directly—watching catalytic reactions unfold, interpreting complex living systems, or finding order in physical processes that span many scales. Used this way, a model becomes an instrument for asking better questions, not merely a faster way to analyze familiar data.
Progress in these fields is changing how we develop medicines, materials, and manufacturing processes. It could strengthen agriculture and biological production while deepening our understanding of physical systems essential to energy, aerospace, public safety, and national defense.
These questions sit between basic and applied science. Answering them helps build a society that can meet new challenges and prosper.
National security is often discussed only in terms of finished systems. The deeper advantage begins much earlier: with the capacity to discover new materials, understand physical processes, develop energy technologies, protect sensitive research, and train people who can move between scientific questions and advanced machines.
It also includes protecting the public from threats. Some are hostile. Others emerge from disease, wildfire, infrastructure failure, or a natural disaster. In each case, science can help public institutions see danger sooner, understand it more clearly, and respond with better tools.
Wildland fire is one example. Computation, AI, and physics can model how a fire may move and orient firefighters during a fast-changing incident. They can also help leaders make better use of limited resources. Better AI research infrastructure accelerates the work upstream so improved models and methods are ready when they are needed.
Countries that build cutting-edge computational capacity will move faster from discovery to industrial capability and be better prepared when a new threat or urgent technical problem appears. A broader domestic base also reduces dependence on a handful of institutions or commercial platforms.
Regional AI hubs are one way to organize more of the country around scientific discovery. The point is not to distribute machines for their own sake.
A durable compact for discovery
At the dawn of the postwar era, the United States made a durable commitment to basic research. Public funding gave scientists room to pursue questions whose value could not be known in advance. Universities and independent research institutions opened new fields of knowledge and trained generations of researchers; industry carried many of their discoveries into the economy.
That compact produced generations of discovery and skilled researchers. It strengthened American prosperity and security, often through applications no one could predict at the outset.
Shared AI computing hubs are a modern test of that idea. Freedom to pursue an ambitious question means little if the instruments needed to answer it are out of reach. Today, those instruments include computing systems that few individual laboratories—and few universities—can build or sustain alone.
Government can provide continuity and orient investment toward national needs. Universities and nonprofit research institutions bring scientists, students, and difficult questions. Philanthropy can help a coalition take shape before every source of financing is settled. Industry supplies extraordinary hardware and software, along with the knowledge required to operate it well.
We are grateful to partner with some of the nation’s leading AI builders: NVIDIA, AMD, Intel, and Dell. The real opportunity is to turn that support into something scientists experience directly: dependable access to advanced AI systems, engineering help when a project becomes difficult, and a long-term role for industry in the regional institutions being built.
Building the next regional AI hubs
This is the kind of work Hangar exists to do: build the institutions hard public problems require.
Hangar and the Secunda Innovation Fund have already begun working with leaders outside New York. We are helping regions understand what prospective members and users need, bring the right partners to the table, and work through the practical questions of building an institution. That work has a federal counterpart: in June, the two organizations signed a Memorandum of Intent with NSF. The agreement identifies areas for possible cooperation, from regional convenings and researcher training to research security and shared standards for access.
These hubs should not be clones. Each should reflect the character of the institutions and researchers it serves.
We believe AI can expand the ambition and pace of science around the world. In the United States, that promise is beginning to take institutional form. NSF has launched a new initiative to unlock the value of existing scientific datasets for AI-enabled discovery. The Department of Energy’s Genesis Mission is applying AI to work in energy, basic science, and national security. Alongside regional hubs, these efforts could let researchers ask harder questions and move discoveries into use sooner.
AI has already begun to change scientific work. The United States now has to build institutions that help its researchers lead—and use that leadership to enlarge humanity’s capacity to understand the world and act on what we learn.
We intend to help. We will have more to say in the coming weeks about the work taking shape beyond New York.
And if you represent a university or nonprofit research institution—or are a public official or researcher interested in helping build one of these hubs—contact us at [email protected].