Table of Contents
HBCU AI Research Campus: R1 Is the Real Test
Morgan State is building an HBCU AI research campus with Google Public Sector. The Carnegie R1 thresholds make the ambition checkable, and they are specific.
An HBCU AI research campus is being built in Baltimore, and the most useful detail is a number almost nobody quotes. On September 23, 2026, Morgan State University announced a partnership with Google Public Sector to build an AI research campus, giving faculty and students high-performance GPU computing through Google’s Program for Accelerated Research, with the stated aim of reaching Carnegie R1 status. R1 is not a reputation. It is a published threshold: $50 million in research and development spending and at least 70 research doctorates awarded. That makes this an announcement you can check.
Key Takeaways
- R1 requires $50 million in R&D spending and at least 70 research doctorates. R2 requires $5 million and 20. These are published thresholds, not reputational judgements.
- The partnership supplies GPU computing, which is the upstream constraint. Without compute, a university cannot produce the AI research that generates the spending that qualifies for R1.
- Donated or discounted cloud compute is not automatically research expenditure. Whether it counts toward the $50 million is an accounting question worth asking.
- Research classification determines whether undergraduates can get paid research positions, which is the part that reaches a 15-year-old choosing a university.
- Running a research programme on one company’s cloud creates a dependency. That is a trade-off, not a scandal, and it is worth naming.
What was announced
A research campus, in this context, is institutional capacity rather than a set of buildings: computing, faculty positions, graduate funding and the administrative apparatus to win and manage grants.
The AI-in-education tracker entry dated September 23, 2026 reports Morgan State University partnering with Google Public Sector to build an AI research campus, with the collaboration giving faculty and students high-performance GPU computing through Google’s Program for Accelerated Research, and notes the university aims to reach Carnegie R1 status.
Morgan State’s own newsroom did not carry that specific item on the page served when checked, which is ordinary for a university news page weeks later. It did list a Google HBCU RISE Summit scheduled for October 5, 2026 among its events, which independently corroborates an active relationship with Google. If you want the primary announcement, it will be in the university’s press archive rather than on its front page.
Why R1 is the number that matters
Here is the part that converts an announcement into a measurable claim.
The Carnegie Classification of Institutions of Higher Education publishes research activity designations with explicit 2025 thresholds. R1, labelled “Research 1: Very High Spending and Doctorate Production,” requires spending at least “$50 million on research & development” and awarding “at least 70 research doctorates.” R2 requires at least $5 million and at least 20 research doctorates.
Two things follow immediately. First, “aims to reach R1” is a falsifiable statement with a scoreboard, which is more than most university announcements offer. Second, the gap between R2 and R1 is an order of magnitude in spending and more than triple in doctorate production. This is a decade-scale institutional project, not a two-year push.
| Designation | R&D spending threshold | Research doctorates | What it means for an undergraduate |
|---|---|---|---|
| R1 | At least $50 million | At least 70 | Many funded labs; undergraduates can get paid research positions |
| R2 | At least $5 million | At least 20 | Some funded labs; research experience exists but competes with teaching load |
| Research Colleges and Universities | Below the R2 spending bar | Fewer or none | Research is faculty-initiated and usually unfunded |
That right-hand column is why a parent should care. A paid undergraduate research position is often the thing that turns an interested 19-year-old into a graduate student, and whether such positions exist is largely a function of how much external funding the institution brings in.
Why GPU access is the binding constraint, mechanically
This is the technical core, and it is more interesting than the press-release language suggests.
Training or fine-tuning a modern model requires large amounts of parallel computation, which in practice means clusters of GPUs. The chips are expensive, they are supply-constrained, and they depreciate fast. A university that cannot provide compute cannot host AI researchers, because an AI researcher without compute cannot produce results, and a researcher without results cannot win the federal grants that count as research expenditure.
So the causal chain runs: compute enables research, research wins grants, grants count as R&D spending, spending plus doctorates produce classification. Compute sits at the head of that chain. A partnership that supplies GPU access is attacking the upstream bottleneck rather than the downstream metric, which is the right order of operations.
Now the honest complication. Discounted or donated cloud compute may not count as institutional research expenditure in the way that a federal grant does, because expenditure accounting generally tracks money the institution spends. If a partner provides compute in kind, the research gets done and the expenditure line may not move proportionally. That is an accounting question with a real answer, and it is exactly the sort of question an institution’s research office can address. Ask it before treating the partnership as progress toward the $50 million threshold.
The access argument, stated carefully
The equity case here is structural rather than statistical, and it is stronger for being specific.
Research classification shapes three things that determine who becomes a researcher: which faculty an institution can recruit and retain, how many graduate students it can fund, and how competitive it is for federal grants. Change the classification and you change who gets paid to do science. For institutions that serve student populations historically underrepresented in computing, that chain is the mechanism by which representation in AI research changes or does not.
There is a related pattern in how young people already reach these tools, and it complicates the simple access story. Pew Research Center’s survey of 1,391 U.S. teens, fielded September 18 to October 10, 2024, found 26% had used ChatGPT for schoolwork overall, with 31% among Black teens and 31% among Hispanic teens compared with 22% among White teens. Usage is not the scarce resource. What is scarce is the path from using a model to building one, and that path runs through funded research positions at institutions with compute. Our look at the AI access gap along income lines covers the school-level version of the same gap.
The dependency worth naming
A research programme built on one company’s cloud has a structural relationship with that company’s pricing, roadmap and priorities. This is not unique to Morgan State or to Google; nearly every university AI effort now rests on commercial infrastructure, and the alternative, buying and operating your own cluster, is far more expensive and depreciates in your own building.
The useful question is not whether to accept the dependency but how it is governed. Is there a term? Who owns the resulting work? What happens to in-progress research if terms change? Those are standard research-administration questions, and a well-run partnership will have answers. The time to ask is at the announcement, not at renewal.
What to do with this at home
Check the Carnegie designation of any university your teen is considering
It takes two minutes and it predicts whether paid undergraduate research positions exist. For a 16-year-old who wants to work on AI, this is a more informative data point than a ranking, because rankings blend prestige with outcomes while designations measure research capacity directly.
Ask admissions one question about compute
“Do undergraduates get access to GPU computing, and through what mechanism?” Any institution serious about AI education can answer specifically. A vague answer means your teen will be learning about models rather than training them.
Separate “uses AI” from “builds AI” in your own planning
The first is now universal and is not a career path by itself. The second requires mathematics, programming and access to compute, in roughly that order. Our overview of whether schools are preparing students for an AI economy covers the preparation question for younger students.
Watch which institutions are building capacity, not which are announcing partnerships
Partnerships are announced constantly. Capacity shows up in published research expenditure and doctorate counts, both of which are public. Following the second list is slower and far more informative.
What not to do
Do not read a classification goal as a classification. Morgan State aims at R1; it has not been designated R1, and the thresholds involved mean the aim is a long project. Treating an aspiration as an achievement does the institution no favours, and a parent who repeats the distinction accurately is being more useful than one who repeats the headline.
What to Watch For Over the Next 3 Months
- Week 4: Watch for details of the Program for Accelerated Research allocation. The useful specifics are how much compute, for how long, and whether undergraduates are included or only faculty and graduate students.
- Month 2 red flags: Coverage describing the university as “now an R1 institution.” Any announcement where compute access is described without a term length. Partnerships announced with no named research questions.
- Month 3 self-check: Can you name the Carnegie designation of the three universities your teenager is most interested in? Three facts, publicly available, more predictive of research opportunity than any ranking.
Frequently Asked Questions
What are the Carnegie R1 requirements?
The 2025 research activity designations require at least $50 million in research and development spending and at least 70 research doctorates awarded for R1. R2 requires at least $5 million and at least 20 research doctorates.
Has Morgan State reached R1?
No. The September 23, 2026 announcement describes the partnership with Google Public Sector and states the university aims to reach R1 status. Given the thresholds involved, that is a long-term institutional goal rather than an imminent designation.
Why does GPU access matter so much?
Because modern AI research requires large amounts of parallel computation. Without compute, researchers cannot produce results; without results, they cannot win the grants that count as research expenditure; and expenditure plus doctorates is what determines classification.
Does donated cloud compute count toward the $50 million?
That is the accounting question worth asking. Research expenditure generally tracks money the institution spends, so in-kind compute may advance the research without moving the expenditure figure proportionally. An institution’s research office can answer specifically.
Why should a parent of a younger child care?
Because research classification determines whether undergraduates can get paid research positions, and that is often what converts an interested teenager into a scientist. Checking a university’s designation is a concrete, two-minute piece of college planning.
Is relying on one company’s cloud a problem?
It is a trade-off that almost every university now accepts, because operating your own cluster is more expensive and depreciates quickly. The questions that matter are the term length, ownership of resulting work and what happens to in-progress research if terms change.
About the author
Ricky Flores is the founder of HiWave Makers and an electrical engineer with 15+ years of experience building consumer technology at Apple, Samsung, and Texas Instruments. He writes about how kids learn to build, think, and create in a tech-saturated world. Read more at hiwavemakers.com.
Sources
- Pursuit. (2026). “AI in Education: News, Policies, Innovations” (entry dated September 23, 2026). https://www.pursuit.us/news/ai-in-education-news-policies-innovations
- American Council on Education. (2025). “2025 Research Activity Designations.” Carnegie Classification of Institutions of Higher Education. https://carnegieclassifications.acenet.edu/carnegie-classification/classification-methodology/2025-research-activity-designations/
- Morgan State University. (2026). “Morgan In The News” and events listing, including the Google HBCU RISE Summit of October 5, 2026. https://www.morgan.edu/news
- Sidoti, O., Park, E., & Gottfried, J. (2025, January 15). “About a quarter of U.S. teens have used ChatGPT for schoolwork — double the share in 2023.” Pew Research Center. https://www.pewresearch.org/short-reads/2025/01/15/about-a-quarter-of-us-teens-have-used-chatgpt-for-schoolwork-double-the-share-in-2023/
- Nickow, A., Oreopoulos, P., & Quan, V. (2020, July). “The Impressive Effects of Tutoring on PreK-12 Learning: A Systematic Review and Meta-Analysis of the Experimental Evidence.” NBER Working Paper 27476. https://www.nber.org/papers/w27476
- UNESCO. (2024). “AI Competency Framework for Students.” https://www.unesco.org/en/articles/ai-competency-framework-students
- National Center for Education Statistics. (2025). NAEP Long-Term Trend Assessment results. The Nation’s Report Card. https://www.nationsreportcard.gov/