A $457 Million Supercomputer at Morehouse and a Bill to Reserve AI Institutes for HBCUs. Is This the Tipping Point?

Morehouse College — a historically Black liberal arts college in Atlanta with 2,100 students and no engineering school — just landed a piece of a $457 million National Science Foundation project to build one of the most powerful academic supercomputers in the southeastern United States.

The machine, called Horizon, will sit on Morehouse's campus. It will be used for artificial intelligence, climate modeling, machine learning, and biomedical research. The initial NSF allocation to Morehouse's Center for Broadening Participation in Computing is $5 million, with additional funding expected as the project scales. The Texas Advanced Computing Center at UT Austin leads the overall project, but Morehouse is not a token partner — it will physically host the infrastructure and run programs designed to build research capacity at institutions that have historically been shut out of big computing.

Three days before the Morehouse announcement, as Granted News reported, Representative Valerie Foushee introduced the HBCU Artificial Intelligence Research Leadership Act in Congress. The bill would require that 10 percent of the research institutes established through NSF's National Artificial Intelligence Research Institutes program be operated by, or in partnership with, a historically Black college or university.

Separately, either of these developments would be noteworthy. Together, they represent something more consequential: a coordinated push to ensure that the institutions which educate a disproportionate share of Black scientists and engineers are not left behind as AI research funding reshapes the American university system.

The Funding Gap Is Not New. The AI Moment Makes It Existential.

HBCUs have operated with structural funding disadvantages since their founding. The 107 historically Black colleges and universities in the United States enroll roughly 9 percent of all Black college students but produce 17 percent of Black bachelor's degree recipients, 24 percent of Black STEM bachelor's degrees, and a disproportionate share of Black doctoral scientists. They punch far above their weight in human capital production.

They do this with far less money. Average per-student endowment at HBCUs is a fraction of what predominantly white institutions hold. Federal research funding flows overwhelmingly to R1 research universities — institutions like Johns Hopkins, Stanford, MIT, and the University of Michigan that have spent decades building the grant infrastructure, preliminary data, and faculty networks that generate more grants. The system compounds: institutions that already have funding are better positioned to win more, while under-resourced institutions struggle to compete.

This dynamic has always been present. But the AI revolution makes it potentially catastrophic for HBCUs. Unlike many research fields where incremental funding can sustain incremental progress, AI research requires massive capital investment in computing infrastructure, GPU clusters, and data pipelines. A biology lab can produce meaningful research with a few hundred thousand dollars in equipment. An AI lab that cannot access modern computing infrastructure cannot meaningfully participate in the field.

The numbers tell the story. NSF's National AI Research Institutes program has awarded 29 institutes since its inception, with investments exceeding $500 million. These institutes, which bring together dozens of universities and research organizations, shape the direction of American AI research for years. But HBCU participation in these institutes has been minimal — typically as minor partners contributing specific workforce development or outreach components rather than leading the research agenda.

The Foushee bill targets this imbalance directly. If 10 percent of future AI Institutes must be operated by or partnered with HBCUs, it creates a structural floor — a guarantee that at least some of the nation's largest AI research investments will flow through institutions that serve communities historically excluded from the field.

What the Morehouse Supercomputer Actually Changes

The Horizon supercomputer project matters for reasons that go beyond raw computing power.

NSF's Leadership-Class Computing Facility program is designed to build computing infrastructure that serves broad academic communities, not just the host institution. When Horizon comes online at Morehouse, it will provide supercomputing access to researchers across the Southeast — including researchers at other HBCUs, minority-serving institutions, and non-R1 universities that have never had access to leadership-class computing.

This is architecturally different from the standard model, where supercomputers are housed at major research universities like TACC, NCSA at the University of Illinois, or the San Diego Supercomputer Center at UC San Diego. Those facilities are technically open to external researchers, but in practice, access is dominated by the host institution and its established research networks. Placing Horizon at Morehouse sends a signal — and creates practical pathways — for institutions outside the traditional supercomputing ecosystem.

Dr. Kinnis Gosha, the Morehouse principal investigator, has framed the project explicitly in terms of diversity. Approximately 62 percent of tech jobs in the United States are held by white Americans. The pipeline problem starts in education, and it starts with which institutions have access to the infrastructure needed to train the next generation of AI researchers.

The Morehouse project includes free summer enrichment programs for middle and high school boys, a postbaccalaureate artificial intelligence program, and faculty accelerator workshops. These are not peripheral add-ons. They are integral to the theory of change: build the infrastructure, train the faculty, recruit the students, and create a self-sustaining research ecosystem at an HBCU that can compete for AI funding on its own merits rather than relying on partnerships with wealthier institutions.

The Legislative Landscape

The HBCU Artificial Intelligence Research Leadership Act faces the political reality that any legislation requiring set-asides must navigate. The 10 percent reservation for HBCU-led or HBCU-partnered AI Institutes does not create new funding — it redirects a portion of existing NSF allocations.

This matters for how the bill is received. Supporters argue that the set-aside simply ensures equitable access to funding that HBCUs have been systematically disadvantaged in competing for. Skeptics may argue that earmarking institute slots by institution type compromises merit-based selection. The bill's co-sponsors have navigated this by requiring that HBCUs meet the same scientific merit standards as any other applicant — the 10 percent floor ensures they get reviewed, not that they automatically receive awards regardless of proposal quality.

The bill has endorsements from the Thurgood Marshall College Fund, Alabama A&M University, the Atlanta University Center Consortium, and the Universities Space Research Association's HBCU Science and Technology Council. Alabama A&M President Daniel Wims has called the legislation "essential to ensure American leadership in Artificial Intelligence globally" — framing HBCU inclusion not as an equity measure but as a competitiveness imperative. America cannot lead in AI if it ignores the talent pool that HBCUs cultivate.

A related bipartisan bill — the Expanding AI Voices Act, introduced by Representatives Foushee and Zach Nunn of Iowa — would codify and expand NSF's existing ExpandAI program, which supports AI research, education, and workforce development at institutions serving underrepresented communities. Together, these bills create a legislative framework for sustained investment in HBCU AI capacity.

The Infrastructure-First Argument

There is a persistent debate in research funding policy about whether to invest in people or infrastructure first. The traditional approach — fund individual researchers through competitive grants and let them build capacity at their home institutions — has served R1 universities well. It has not served HBCUs well, because institutions without existing infrastructure cannot attract the faculty who generate the preliminary data that win the grants that build the infrastructure.

The Morehouse supercomputer and the Foushee bill both take the infrastructure-first approach. Build the computing capacity. Reserve the institute slots. Create the conditions under which HBCU researchers can compete, and then let competition operate.

This approach has precedent. NSF's Established Program to Stimulate Competitive Research (EPSCoR) has operated on a similar logic for decades, directing funding to states that historically receive less federal research money. EPSCoR does not guarantee awards to under-resourced states — it ensures those states have the infrastructure and capacity to compete. The HBCU AI Research Leadership Act applies the same principle to institutional type rather than geography.

The question is whether infrastructure investment at HBCUs can overcome decades of cumulative disadvantage quickly enough to matter in a field that is evolving as fast as AI. The technology does not wait for equity initiatives to mature. The compute architectures, model training approaches, and research paradigms that define AI are being shaped right now, and the institutions that participate in shaping them will have outsized influence for decades.

What Grant Seekers at HBCUs Should Do Now

The Morehouse announcement and the Foushee bill create concrete opportunities for researchers and administrators at historically Black colleges and universities, but capturing those opportunities requires action.

Map your institution's AI research assets. Even HBCUs without formal AI programs often have faculty doing AI-adjacent work — in computational biology, data science, applied statistics, machine learning applications in social science. Identifying and connecting these researchers is the first step toward building the collaborative proposals that AI Institute applications require.

Build partnerships with R1 institutions — but demand substantive roles. The 10 percent HBCU set-aside in the Foushee bill requires that institutes be "operated by, or in partnership with" an HBCU. The "in partnership" language creates risk: a well-meaning R1 university could include an HBCU as a minor partner to satisfy the requirement without giving the HBCU meaningful research leadership. HBCU administrators should negotiate for co-PI roles, budget allocations commensurate with research contributions, and shared governance of institute operations.

Pursue the Horizon supercomputer access pathway. Morehouse's Center for Broadening Participation in Computing will facilitate access to the Horizon system for researchers across the HBCU community. Getting in early — establishing accounts, running pilot computations, training students on the platform — positions your institution as a proven user when larger competitive allocations become available.

Track the legislative timeline. The Foushee bill has bipartisan potential but is not law yet. The Expanding AI Voices Act, which would codify and expand the ExpandAI program, may move faster. Researchers should prepare proposals that align with both programs so they can respond quickly when solicitations appear.

Invest in grant infrastructure. The single largest barrier to HBCU competitiveness in federal research funding is not faculty talent — it is the administrative infrastructure needed to manage large, complex, multi-institutional awards. Sponsored programs offices, institutional review boards, compliance teams, and budget management systems all require investment. The cost of building this infrastructure is a fraction of the funding it enables, and HBCU presidents and provosts should treat it as a strategic investment rather than an overhead burden.

The Stakes

The United States faces a straightforward choice about who participates in the AI research enterprise. The current trajectory concentrates AI funding at a small number of wealthy, predominantly white research universities. This is not because those universities are the only ones capable of doing good AI research. It is because the funding system rewards existing capacity, and existing capacity reflects historical investment patterns that excluded HBCUs.

The Morehouse supercomputer and the Foushee bill represent the beginning of a structural correction. They do not guarantee equity. They do not solve the funding gap overnight. But they create the conditions under which equity becomes possible — by putting infrastructure where it has not been, by reserving institutional slots that have been effectively unavailable, and by building the pipeline of HBCU-trained AI researchers who will compete for funding on their own terms in the decades ahead.

For researchers and administrators at HBCUs, this is not a moment to wait and see. It is a moment to build — and tools like Granted can help you identify the AI research funding opportunities, foundation partnerships, and federal programs that align with your institution's emerging capabilities.