Recently, the NSF released the FY 2024 Higher Education Research and Development (HERD) data. I was thinking of writing a post for the general public (and for non-US academics) to dispel some myths, but the more I looked at it, I felt that some of this information would be useful for everybody. Plus, a bunch of universities are boastposting their numbers, but comparisons are in order.
Disclaimer #1: Numbers shouldn't be used to interpret quality (though the correlation is not negligible)
Disclaimer #2: There is some level of funny counting within every university.
Disclaimer #3: Some numbers include subsidiary funding that doesn't quite go to the university. e.g. Johns Hopkins numbers include Applied Physics Laboratory. Also, some universities (e.g. Georgia Tech / GTRI) count some fraction from affiliated research institutes and some don't (e.g. MIT / Lincoln Labs). I am not quite sure about this.
Click on the pictures to enlarge them.
Overall numbers
U.S. universities spent $117.7 billion on R&D in FY 2024. That's nearly 4x what it was in 2000 ($30B) and almost 500x what it was in 1953 ($255M). Even adjusted for inflation, the growth is remarkable. But the composition of that funding has shifted dramatically, and that shift says something important;
The Federal Retreat
Federal funding peaked at 73% of total university R&D in 1968. That was the height of the Space Race and Cold War science investment. Sputnik had shocked the nation into action, and the government responded by pouring money into universities. The result was an unprecedented expansion of research capacity- new labs, new faculty lines, new PhD programs. By 2000, the federal share had dropped to 58%. By 2024, it's down to 55%.
Now, to be clear: federal funding in absolute terms has grown substantially.. from $17.5B in 2000 to $64.7B in 2024. The federal government is spending more on university research than ever before. But universities have grown their research enterprise even faster, and they've done it by diversifying their funding base.
But cost of doing research is also increasing faster than inflation. When I got to the University of Michigan in 2013, funding a PhD student cost me $70k/year. It is now closer to $110k. In any case, the question is: how did universities fill the gap?
The Rise of Institutional Funds
The biggest shift has been the growth of institutional funds- money that universities spend from their own coffers on research. This category has grown from 14% in 1953 to 20% in 2000 to 26% in 2024. That's $30.2 billion, roughly half of what the federal government provides.
What's in this bucket? Startup packages for new faculty. Bridge funding when grants run out. Cost-sharing on federal grants. Seed funding for new initiatives. Running institutes like MICDE, Internal research competitions. And yes, the unrecovered portion of indirect costs.
This is universities betting on themselves. Whether that's sustainable is another question—one that depends heavily on tuition revenue, state appropriations (for publics), endowment returns (more so for privates), and clinical revenue (for those with medical centers).
A Tale of 13 Universities
The aggregate numbers are useful, but the institutional variation is where it gets interesting. Here's a comparison of some major research universities
Harvard and Penn have the largest institutional shares (35% each). Large endowments enable substantial internal research investment. Harvard also leads in nonprofit funding (13%)—foundations and philanthropic research support are a significant piece.
Ohio State and U. Florida show the land-grant pattern: more substantial state/local funding (4-16%) alongside federal and institutional. These are states that still invest directly in university research.
MIT has a high level of funding from industry (17% of its total), a reflection of deep industry ties, though its absolute R&D, MIT's model is different: smaller but more focused, with substantial classified and Lincoln Lab work that may or may not show up in these numbers.
**In fact, U. Michigan has a gargantuan annual budget of $15.6 B out of which $373 M comes from the state (< 3%, but it is an incredibly important piece - not all money is the same)
----------
So...
The data mostly confirms what we already knew: federal funding matters enormously. Compared to industry funding and other sources, it tends to fund basic research, and emphasizes collaboration. But it is no longer the majority; universities are increasingly self-funding their research; and different institutions have carved out very different niches.
What I do think is worth watching: as AI and automation begin to reshape the research enterprise (a topic I've written about before), the capital requirements for cutting-edge research may shift. Compute is expensive. Robotic labs are expensive. Large AI models are expensive to train. If the bottleneck moves from "intelligence" to "infrastructure," the funding picture could look very different a decade from now. Universities are already struggling to prioritize research given the cost of doing research, and the recent assault on indirect costs could be fatal.
For now, though, the message is clear: $118 billion is a lot of money*, but it is a top heavy model, spread across 925 institutions with very different models. I also don't know how sustainable this is. Below is a view of some of the inequity. Note: I am certainly not advocating for equity in funding. Different universities serve different roles. In fact, every university trying to do everything is not workable.
* But also not a lot of money compared to what the big tech companies are spending. For comparison, the top 5 tech companies spent $229B on R&D last year, pretty much all on computing and AI. Our Top 50 universities spent a quarter of that amount on everything.
Comments
Post a Comment