A Silicon Valley plan to "fix" universities
Higher education under the venture capitalist "entrepreneurial lens"

"… there’s no way to replace them without letting them fail."
(Michael Roth, President of Wesleyan University)
If you believe the Trump administration, with its Nazi salutes and “very fine people,” is defunding universities out of a concern to protect Jewish welfare, then you probably also believe
Tariffs were imposed on Canada as a response to fentanyl smuggling;
Over 130,000 people in the Social Security databases are over 160 years old;
Ukraine started the war with Russia;
Trump won the 2020 election;
Immigrants are eating cats and dogs in Ohio.
Skeptics of such claims will have sought other explanations for the assault on higher education. Some argue that Trump, in his “calculated exploitation of antisemitism as a wedge for a creeping authoritarianism in the United States,” is following the playbook of Florida governor Ron de Santis, who notably purged the Board of Trustees at New College,1 as well as such established autocrats as Turkey’s Erdoğan or Hungary’s Orbán.
Donald Trump is coming for America’s universities and higher education workers for the same reason that Mussolini demanded loyalty oaths in 1931 and Viktor Orbán launched a “vicious smear campaign” against Hungary’s premier university in 2019: Our commitment to open inquiry over propaganda, to persuasion over coercion, to democracy over dictatorship provides a check on authoritarian rule.2
Others point to “anti D.E.I. crusader” (and recently appointed New College trustee) Christopher Rufo and his ambition
to adjust the formula of finances from the federal government to the universities in a way that puts them in an existential terror and have them say, Unless we change what we’re doing, we’re not going to be able to meet our budget for the year.3
And then there are the lobbyists inside the faculty, including a mainly anonymous group, mentioned in a recent post by my colleague Peter Woit, who apparently “are willing to destroy the university if they don’t get what they want.”
Each of these explanations reflects an important aspect of the situation, but the Trump coalition has another important component: the tech bros. Do they also have a vested interest in the collapse of higher education in this country?
Here’s what Marc Andreessen had to say about universities about an hour into a marathon Lex Fridman podcast, six days after Trump’s inauguration:
… we have a government funded and supported cartel that has gone… It’s just obvious now. It’s just gone sideways in basically any possible way it could go sideways…They’re just wrong in every possible way at this point. And it’s all on the federal taxpayer back. And there is no way, I mean, my opinion, there is no way to fix these things without replacing them. And there’s no way to replace them without letting them fail. And by the way, it’s like everything else in life. …this is the most obvious conclusion of all time … when a company does a bad job is they go bankrupt and another company takes its place, and that’s how you get progress. And of course, below that is what happens is this is the process of evolution. Why does anything ever get better? Things are tested and tried, and then the things that are good survive. And so these places, they’ve been allowed to cut themselves off, both from evolution of the institutional level and evolution of the individual level as shown by the just widespread abuse of tenure. And so we’ve just stalled out. We built an ossified system, an ossified, centralized, corrupt system, where we’re surprised by the results. They are not fixable in their current form.
It’s true that Andreessen is just one billionaire, an especially chatty one, among the more than 900 who call this country home. But since the firm he runs with Ben Horowitz “ranks first on the list of venture capital firms by assets under management”4, with investments mainly concentrated in the tech industry, Andreessen’s opinion can be taken as a plausible proxy for the dominant worldview in their corner of the Bay Area.
Even if you are steeped in that worldview, like the 1.4 million people who have watched the podcast quoted above, you may not think of higher education as a “cartel.” But you are likely to see it as an industry overdue for absorption by Silicon Valley, in the process that has seen one social institution after another — taxis, shopping, journalism, friendship, and much more — “disrupted” beyond recognition into a vehicle for wealth extraction. One year before Andreessen’s interview with Fridman, he and Horowitz devoted three full episodes of their (more or less) weekly two-hour “Ben and Marc Show” podcast to elaborating their vision of “Higher Ed in Crisis” and their prescriptions for “fixing” the crisis.
Bundles of bundles
…you know the way we try to always do this in business is to try to start with like okay before you figure out what you're going to do it's like okay what exactly are your goals um5 and then and then of course the next question you know from what are exactly your goals is sort of the question of what exactly are your goals you know for who right like who you know who who are you trying to satisfy who are the customers…
(Andreessen, Part II, 5'18'')
Many of the accusations the billiard ball-headed billionaire bros level against the American system of higher education are undeniably justified.6 Student debt, for example, really is unsustainable; the explosive growth in the number of administrative positions is wildly misaligned with the educational needs of the population.7 The problem is with who Marc and Ben are. They are venture capitalists not only in fact but in essence, and this means that they can only perceive a social structure as a business in decline, a startup, or an investment opportunity.
Thus education is the “product,” students are the “main customers,” and what universities offer is a “value proposition” … which is “getting very weak.”8
…so there's a lot of factors that lead to that but like at the end of the day like how the hell do you end up with a product that costs $300,000 and gets the average student a job that's worth you know like $50, how are they ever going to pay back $300,000 if they only make $50,000 a year like that's insane um and that's that's kind of where the I I think the product for the student has fallen apart…
(Horowitz, Part II, 11:50)
Time to fix! The VCs begin their anticipatory post mortem of higher education by breaking down the “product” as a “bundle” of a dozen functions, like the bundles sold by Cable TV companies. Alongside such familiar functions as “Credentialing Agency,” “Sports League,” “Dating Site,” and “Hedge Fund,” Marc’s Twelve Functions of the University includes more conventional entries like “Courses,” “Research Bureau,” “Policy Think Tank.” There’s also “Moral Instructor,” about which I’ll have more to say below. Predictably, Marc and Ben, from their Menlo Park, CA vantage point, see the university’s way of organizing these functions as antiquated.
it’s all on the internet and it’s all on ChatGPT like there is no knowledge that you can’t get from your smartphone very cheaply and easily and so why am I paying $300,000?9
(Horowitz, Part I, 0:44:53)
The precise form of disruption their “entrepreneurial lens” has in mind takes the form of unbundling. Why should the university’s customers — the students — need to pay $300,000 for the whole bundle when technology can provide the desired functions at a much more modest cost? Why, indeed, pay Comcast for ESPN when you only watch HBO, or vice versa?
In Silicon Valley, the solution is obvious:
…I'll tell you what Al is really good at: filling out forms, looking at forms, getting data out of forms, comparing that data to the what the data should be, figuring out things anomalies in the data, that no human could you know it's amazing at that and well like could AI do all these administrative tasks could you just get rid of like that whole thing or like whatever 95 % of it and then you know you're you're starting to get cost back where they are well…
(Horowitz, Part II, 0:13:48)
Then the professors could be replaced by one-on-one tutoring:
you could literally assign every student like a really good instructor to teach them all these subjects um full-time, like no other students.
(Horowitz, Part II, 0:19:17)10
In response to which Marc observed
Alexander the Great is kind of the apotheosis of that because his tutor was Aristotle, right?
(Andreessen, Part II, 0:20:03)
Besides,
…probably over 95% of college students are there to prepare themselves for the workforce, and a four year degree to do that um seems a little misplaced in a world where the rate of change is so high…
(Horowitz, Part III, 0:14:01)
So what is to be done?
…part of the answer is create new institutions um and you know maybe you figure out a way to get them accredited and so forth but or figure out another funding model or maybe they you know happen in other countries or whatever um and so you know could you start new competitive institutions that maybe are able to solve some of these problems if it's too hard to to uh reform the current ones and then the third and this is probably we should set up another session on this but you know then then there's also like the is there a potential is there basically potential for entrepreneurial opportunity…
(Andreessen, Part I, 1:50:38)
The more realistic “entrepreneurial opportunities” are unveiled toward the end of Part II.
…I think that with universities kind of there's such a huge opening um for you know different lengths of degrees so like the four-year degree is really something that doesn't make much sense to me um so to adopt that … given that people who aren't need kind of skills and then new skills and new skills and new skills I mean could you imagine if we were trying to do our jobs based on just what we learned in college … you know like that's one thing you might bring into question and then if you had a shorter degree where you could get just as high paying a job and this something Lambda school does of course then all of the sudden you have a value proposition that's starting to look really good oh maybe like 10 grand you know you lend me 10 grand for a year this is what Lambda school does11 and then you pay me back if and only if you get a job well okay that starts to sound pretty good so like I think there's things that would be much more attractive to students potentially …
(Horowitz, Part II, 1:11:06)
A bit later Horowitz allows himself to dream of “building a new full-out four year Ivy League of the future type thing”:
The dreams continue through the end of Part II. The segment on the pros and cons of starting new universities, quoted briefly above, leads to 40 minutes of unbundling. The “Credentialing” function could be
abstracted out and turned into its own thing… this may be the best um startup idea of everything in education… if somebody had an organization that aptitude personality tested people.
Andreessen has a more ambitious idea for unbundling the “Courses” function:
Finally, the vision for research
This brings us to the “Research Bureau” function, the one of most concern to mathematicians reading this newsletter. Andreessen and Horowitz have a very clear vision, based on conversations with an unnamed “very experienced guy who’s been in the sort of leadership positions across this entire spectrum,” of how research should be funded.
The idea is that in any given field “there are five institutions that are on the leading edge… and those five institutions generate you know essentially all of the useful output that actually moves the field forward.”
so I was like well why doesn’t the government just fund that… and he says that’s like 10% of the money and I was like well then why spend the other 90%?
(Andreessen, Part II, 1:40:07)
Once the VCs “figure out like what is the actual 10% of the useful work” they can arrange to have it “funded with philanthropic dollars.” But the “very experienced guy” wonders
look maybe it should be a venture capital model it should be for-profit and you just basically make money it's a long-dated you know Revenue thing where you're making money in the long run on on on Commercial product development and a patent licensing coming out the other side um and and you should actually just like apply you should actually apply a a VC mindset to that so anyway.
And this, I submit, is how some far-sighted Silicon Valley thought leaders expect mathematicians to be replaced, with or without AGI.
“Such a earthquake and shock to the system”
OpenAI has recently issued a plan to collaborate with higher education institutions in order to transform them from within:
Students are at the forefront of the AI revolution and eager to expand their skills, and employers are increasingly seeking an AI-ready workforce. However, too many students are left to discover and learn about AI independently, with limited institutional support, leading to uneven adoption across the US. OpenAI aims to collaborate with states, universities, colleges, and schools to advance research and better integrate AI into education, ensuring a strong foundation for both students and America’s economic future.
The above is taken from an OpenAI pamphlet12 entitled
In the pamphlet we learn that “Many college and university students are teaching themselves and their friends about AI without waiting for their institutions to provide formal AI education or clear policies about the technology’s use.” But “Whether or not academia is prepared for AI adoption, employers are increasingly prioritizing AI proficiency.” OpenAI has prepared “actionable proposals” to help “integrate AI into our educational infrastructure and ensure that our workforce is ready to both sustain and benefit from our future with AI.” Arizona State University and the California State University system are getting with the program: ASU is “offering ChatGPT Enterprise [its most expensive model, estimated13 at $720 per user per year, with a minimum of 150 users for a contract] to students and faculty.” CSU, meanwhile, is offering ChatGPT Edu to roughly half a million students and faculty — that’s the “affordable plan” based on GPT-4o: “Contact our team to learn more.”
Both ASU and the Sacramento CSU campus are on the Department of Education’s list of “60 Universities Under Investigation for Antisemitic Discrimination and Harassment,” published March 10 of this year. The OpenAI pamphlet is undated but can’t have been produced before February.
Andreesen: …I think even the large institutions are closer to the edge financially yeah um than than than a lot of people think um especially of course if there's any withdrawal Federal funding
Horowitz: yeah yeah that that's that for sure yeah that would man a withdrawal of federal funding would be such a earthquake and shock to the system it's a little scary to even think about because the knock on effects would be so it'd be so fast um…
(Part III, 1:28:12)
Will ASU and CSU be able to pay their ChatGPT bills? And if not, does OpenAI have a plan to market its services to the universities’ unbundled remnants?
“Moral instruction”
…institutions committed to freedom of inquiry, to the contestation of ideas through conversation and debate, to the formation of communities that gather and celebrate a diverse array of experiences and thought, and to individual flourishing achieved through diligent study.14
The familiar arguments for a liberal arts education are totally absent from Andreessen’s list of Twelve Functions. If they appear at all, it’s in vestigial form,15 under the heading of “Moral Instruction,” a function the two billionaires see as totally compromised in existing universities, with no moral values beyond DEI.16
This brings us back to the Trump administration’s claims to be concerned about antisemitism. A Guardian journalist decodes these claims — clearly with the help of a liberal arts education —
A liberal arts professor warns more generally of what happens when we lose the habit of “freedom of inquiry”:
The centrifugal intellectual motion that attends this absence [of “ruling ideas”] makes the world feel more insane than incoherent because the world is now divided into irreconcilable domains of fact—mutually exclusive ways of designating of reality—and there is no voice of authority to decide which of these is legitimate, thus actionable. Power, or mere strength of will, fills that vacuum.17
This observation also contains a lesson for mathematicians struggling to find coherent answers to tech industry claims about the place of AI in mathematics. Readers with inside knowledge are urged to share information about the tech industry’s thoughts — the more insane, the better — on unbundling higher education, in the form of internal documents, or startup pitches, or conversations overheard in Bay Area coffee bars.
…whose alumni include the extraordinary topologist William Thurston, whom readers have encountered here more than once, as well as University of Michigan mathematician Richard Canary.
Jonathan Feingold and Veena Dubal, “ Trump Is Revealing Our Higher Ed Crisis,” to appear in the May issue of Academe.
New York Times interview, March 7, 2025. It’s telling that the Academe article referred to concern with antisemitism as “the thinnest pretext,” while the Rufo interview didn’t mention antisemitism at all.
Automatic transcript generated by YouTube.
And many are outrageous. Horowitz’s interminable rant about DEI in Part II culminates in the suggestion that Black students should be encouraged to go into music rather than engineering (“different populations… [are] interested in different things”). My roommate at Harvard heard exactly the same message from his advisor Richard Herrnstein — co-author of The Bell Curve with Charles Murray — except that it was basketball rather than music. Twenty minutes later he’s still talking about talent and “political litmus tests,” by now focusing on gender rather than race.
Horowitz had already ranted about DEI for 15 minutes in Part I! It may be relevant that Horowitz is the son of neoconservative activist David Horowitz, author of many books including The Professors: The 101 Most Dangerous Academics in America, published in 2006. Nine of the 101 named there were at Columbia. See note 8.
Rana Foroohar offers prescriptions for addressing these anomalies from a liberal perspective in the Financial Times.
Horowitz, Part II, 2:00:11.
That’s for a four-year education. Based on recent trends, the VCs predict that the price tag will reach $1 million in the very near future. Horowitz observes (Part I, 0:45:55) that “Even if you work for us that’s hard to pay off.” At this point they say it would be more cost effective to return to one-on-one tutoring, something the hereditary aristocracy practiced until very recently.
This is the right place to explain to readers who are surprised at the actions of Columbia’s Board of Trustees over the past year or so that Ben Horowitz was himself a Columbia trustee for several years not so long ago.
From Wikipedia:
Bloom Institute of Technology, also known as BloomTech, is a coding bootcamp providing for-profit massive online course. Launched in 2017 under the name Lambda School, it gained attention for being a coding bootcamp that offered income share agreements as a method of financing. During this time, it deceived students about costs, made false claims about graduates’ hiring rates, and engaged in illegal lending.
Thanks to Margaret Wertheim for bringing this to my attention.
According to https://explodingtopics.com/blog/chatgpt-enterprise
Ronald J. Daniels, “What Universities Owe Democracy,” Democracy, Spring 2022, No. 64.
That’s not quite true. The humanities and social sciences get a brief mention, as “navel gazing” in the segment on the “replication crisis” in research:
…let’s both write our opinions right and publish them and call them papers… they generally have zero zero future citations… there may be something brilliant in them but they’re not leading anything there…
Not, in any case, a reason for anyone to go to the university.
Examination of the “Moral Instruction” function in each of Parts I and II sparks an interminable rant by Horowitz on DEI. .
James Livingston, “When the ruling class goes missing,” March 29, 2025. Later Livingston, in reference to “the dispersal or disintegration of a ruling class,” writes that “it has come of age in our own time, in the form of morons like Marc Andreessen who mistake their ownership of assets for the possession of intelligence.”
This reminds me so much of the late 1990's/early 2000's movement around MOOC's (massive open online courses) which were also going revolutionize education and replace in-person teaching. They too were said to be more democratic, more efficient, and would make great education available to All. David Noble critiqued this wave of techno-edutopianism is his terrific book "Digital Diploma Mills" (2003). So much of what he said then applies to SiValley edu-hype today. Here's a quote from a review of Noble's book in the socialist magazine Monthly Review:
"In the first book-length analysis of the meaning of the Internet for the future of higher education, David Noble cuts through the rhetorical claims that these developments will bring benefits for all. His analysis shows how university teachers are losing control over what they teach, how they teach and for what purpose. It shows how erosion of their intellectual property rights makes academic employment ever less secure. The academic workforce is reconfigured as administrators claim ownership of the course-designs and teaching materials developed by faculty, and try to lower labor costs in the marketing and delivery of courses.
Rather than new opportunities for students the online university represents new opportunities for investors to profit while shifting the burden of paying for education from the public purse to the individual consumer—who increasingly has to work long hours at poorly-paid jobs in order to afford the privilege. And this transformation of higher education is often brought about through secretive agreements between corporations and universities—including many which rely on public funding."
https://open.substack.com/pub/nadvice/p/peter-thiel-on-endurance-innovation?r=41dd5t&utm_campaign=post&utm_medium=web&showWelcomeOnShare=true