In January 2026, New Jersey, a group of its higher education institutions, NVIDIA, and the New Jersey Artificial Intelligence Hub signed a memorandum of understanding to build a supercomputer for the state. The New Jersey Economic Development Authority (NJEDA) will manage and deploy roughly $25 million toward what the announcement called "a statewide supercomputer initiative for New Jersey's institutions of higher learning," meant to give students, researchers, and entrepreneurs access to modern computing for AI work. Then-Governor Phil Murphy signed it at what turned out to be nearly his last press conference in office; Governor-elect Mikie Sherrill's transition team confirmed her administration would carry the project forward, and she has since taken office and continued to back it.
If you work at a small New Jersey college or university, that phrase, "statewide" and "institutions of higher learning," should get your attention. It's the same phrase that describes exactly the gap I write about most on this site: research computing capacity that exists, is publicly funded, and is nominally available to every campus, but that small institutions miss because nobody on staff is tracking it. This article lays out what's actually confirmed about the NJ AI Hub's supercomputer plan, what's still open, and what a small campus can do right now, before the details are finalized, to make sure it isn't left out again.
What the NJ AI Hub is, in plain terms
The NJ AI Hub is a state-university-industry partnership that opened in West Windsor in March 2025. Its founding partners are the NJEDA, Princeton University, Microsoft, and CoreWeave (a cloud computing company headquartered in New Jersey), and together they committed more than $72 million to it: $25 million from the NJEDA and the rest in donated services and support from Princeton and the industry partners. The Hub runs on three tracks: helping companies apply AI to their own research and development, building an accelerator for AI startups (operated by Plug and Play, which selected its first cohort of 14 startups), and workforce and talent development. Liat Krawczyk serves as its executive director.
The January 2026 supercomputer MOU is a separate, later commitment layered on top of that original partnership, and it's the piece most directly relevant to campus researchers rather than startups. NVIDIA co-founder and New Jersey native Christopher Malachowsky signed on behalf of the company, and New Jersey Institute of Technology president Teik Lim spoke at the announcement about the opportunity for "higher education, government, and industry to partner on a flagship, statewide approach to advancing shared AI infrastructure, education, and research capacity."
What's confirmed, and what isn't yet
Read the announcement carefully and there's a real distinction between what's been decided and what's still being worked out.
Confirmed: the dollar figure (about $25 million, managed by the NJEDA), the stated purpose (a statewide resource for New Jersey higher education, not one campus), the partners at the table (the state, NVIDIA, the NJ AI Hub, and a group of higher education institutions), and the political continuity (both the outgoing and incoming governors have backed it).
Not yet public, as of this writing: which institutions get direct access and on what terms, whether county colleges and smaller four-year institutions are included on the same footing as the research universities at the table, what the allocation process will look like (a formal proposal, a lighter request, or something else), when the system will actually be usable by a researcher, and who at each campus is supposed to request access. Those are exactly the questions that, in the national research computing ecosystem I write about elsewhere on this site, tend to get answered informally, through word of mouth among people who already have a research computing office. Small campuses without one are structurally the last to find out.
Why this matters more for small institutions than it might seem
The instinct at a small college, on hearing "New Jersey is building a supercomputer with Princeton and NVIDIA," is to assume it's for the big research universities and move on. That instinct is usually wrong, and it's worth being specific about why. The language in the MOU is deliberately statewide, not university-specific, and New Jersey Institute of Technology's public comments framed the project around "New Jersey students, innovators and workers" broadly. The state has also, through the NJ AI Hub, already run a parallel initiative aimed squarely at its 18 community colleges (AI Ready NJ, which I cover in a companion article), which signals that "statewide" is meant literally rather than as boilerplate.
That said, stated intent and lived access are two different things, and the gap between them is where small institutions lose out. A flagship research university with an existing research computing office will have someone in the room when the access policy is written, someone who knows to ask about allocations at the next NJ AI Hub event, and a grants office that already tracks state funding opportunities. A small college with no dedicated research computing staff has none of that infrastructure, even though its researchers, and its students in AI-adjacent programs, are named in the same sentence as everyone else's.
The projects most likely to serve small institutions are the ones where a small institution shows up early enough to ask how.
What a small campus can do now, before the details are final
- Name one point person, even at a fraction of their time. Someone needs to be the person who reads NJ AI Hub announcements and NJEDA press releases and asks, on your campus's behalf, "does this apply to us?" Without that person, the information arrives too late to act on, the same pattern that keeps small colleges out of the national OSPool and ACCESS systems.
- Get on the NJ AI Hub's mailing list and watch njaihub.org's news page. The Hub has been consistent about publishing updates as programs launch (the accelerator cohort, the AI Ready NJ partnership, the supercomputer MOU); the access details for the supercomputer, when they're announced, will likely appear there first.
- Ask directly, and ask now, rather than waiting for a public call. A short email to the NJ AI Hub or your institution's state relations contact asking how a small institution requests access costs little and puts your campus on record as interested before the process is fully designed.
- Build the case with a real research need, not a hypothetical. The clearest way to get taken seriously in a new allocation process is to show up with a specific, funded, or fundable research question that needs the capacity, the same advice that applies to first allocations on ACCESS or the OSPool. "We might use it someday" competes poorly against "our biology department has a genomics pipeline that currently takes a week on a laptop."
- Don't wait on this to start using what's already free. The OSPool and NSF ACCESS exist today, require no state MOU, and already accept researchers from any US academic institution at no cost. A campus that gets one researcher running on those systems this semester will be in a far better position to use a new state resource later, because by then someone on staff will already know what a research computing allocation looks like.
A parallel to watch: the national ecosystem's slow start
It's worth remembering that the national research computing ecosystem, the OSPool, NSF ACCESS, and the campus networks around them, took years to become something a small college could actually use without a translator. Access was technically open from early on; usable access, with clear documentation and a first researcher who could walk a colleague through it, came later and unevenly, favoring campuses that had someone paying attention. A state-level effort with a single funder, a single set of partners, and a compressed timeline could move faster than that, or it could repeat the same pattern at a smaller scale. Which one happens will depend heavily on whether small institutions insert themselves into the process early, rather than waiting for an invitation that assumes they're already in the room.
The takeaway
New Jersey has committed real money and real partners to a statewide computing resource for higher education, and the language used to describe it, deliberately, includes institutions beyond the flagship research universities. What isn't decided yet is exactly how a small college requests a seat at that table. That's an opening, not a wall: campuses that ask the question now, while the access process is still being designed, have a real chance to shape an answer that includes them. Campuses that wait for a formal announcement risk finding, again, that the process was built around institutions that already had someone in the room.
If your institution wants help figuring out where it stands with the NJ AI Hub, or wants to get a researcher running on the national resources that are already free today while the state system comes online, get in touch and we can map out the first step.