Andreessen Horowitz is backing a new alternative to college built around a radically different model.
The Horowitz Andreessen Academy will heavily emphasize proof of work in admissions, building rather than traditional coursework, and hands-on experience with companies shaping emerging technology. Its first class will be small, and companies including OpenAI, Anthropic, Google, Meta, NVIDIA, and others are helping shape the curriculum and provide access to their networks and resources.
The initial Founding Class Fellowship will not award a degree or college credit. The Academy plans to pursue approvals for a two-year program beginning later.
It would be easy to look at this and conclude that AI is finally coming for the university.
Except we’ve heard versions of this before.
Universities have survived disruption before
MOOCs were going to unbundle higher education. Online learning was going to make campuses unnecessary. Once anyone could learn from the world’s best professors online, why pay for a traditional degree?
Universities survived.
Partly because those predictions underestimated what else universities provide.
They don’t just transfer knowledge. They provide credentials and signaling, networks and community, mentorship, research, recruiting pipelines, and experiences that are much harder to replicate.
Many previous attempts to disrupt higher education attacked one part of the bundle, particularly the transfer of knowledge, without replacing the rest of it.
Making a world-class lecture available online is very different from replacing the credential, the network, the relationships, the recruiting pipeline, and the experience surrounding it.
Universities also have something most organizations don’t: an entire system built around them that makes them remarkably difficult to displace.
But that may buy universities time. It doesn’t eliminate the need to rethink the model.
The bigger threat may be unbundling
I think unbundling will put enormous pressure on many universities.
Because the question isn’t whether people will continue to need education, credentials, relationships, and opportunities to develop.
Of course they will.
The more interesting question is whether all of those things still need to come from the same place.
What happens when education comes from one source, credentials from another, experience from employers, networks from communities, and AI provides increasingly personalized access to knowledge and expertise?
The university doesn’t necessarily disappear.
But the bundle starts coming apart.
And once that happens, the question changes.
Will people continue to buy all of those things, bundled together, from the same institution, in roughly the same way, at the same price?
Industries can survive while value moves elsewhere
That points to something much bigger than higher education.
AI may make some banking activities unnecessary without making banks unnecessary.
Agents may weaken Amazon’s search interface without eliminating the enormous value of its fulfillment network.
AI may automate significant portions of consulting without eliminating the need for judgment, trust, and helping groups of people make consequential choices.
In each case, the underlying need can remain while where and how value is created changes.
That’s an important distinction.
The survival of an industry does not guarantee the position of the organizations currently leading it.
Kodak is a useful example.
Kodak understood that photography was going digital. It invested heavily in digital technology and, in 2004 and 2005, actually led the U.S. digital-camera market. But the economics of digital photography were dramatically different from the high-margin film business Kodak had dominated.
People didn’t stop taking pictures. We take more than ever.
Photography thrived. That didn’t preserve Kodak’s position.
The value moved.
“How should we use AI?” is too small a question
That’s why I think “How should we use AI?” is far too small a question for leadership teams.
It tends to lead toward productivity:
Where can we automate? Where can we move faster? Where can we reduce cost? Where can AI make people more effective?
Those are important questions.
But they are questions about improving the organization we already have.
Even “How should we adapt?” may be too small if adaptation means doing today’s work substantially better while the assumptions underneath the business are changing.
The harder questions are:
What is becoming less valuable?
Where is value becoming more important, and who will capture it?
What can now be created that wasn’t possible before?
Those questions force a different conversation. They move the leadership team from asking how AI fits into the existing organization to asking what the organization should become.
Because improving your organization is not the same as rethinking it.
Your organization can get much better at what it does today and still lose.
The question for every leadership team
Universities may have more time than most organizations to rethink their models.
But the question they face is increasingly one every leadership team needs to confront.
Are we improving the organization we have, or rethinking the organization we will need?