Chiemela Uzunma Kalu

Twenty-one billion dollars. That is how much global capital flowed into educational technology last year. Venture firms, philanthropists, and governments placed their bets on the idea that technology can transform learning, close achievement gaps, and prepare the next generation for an AI-driven economy.

I believe in that idea. I have built my career around it.

But I also work inside the systems that underpin higher education every single day, and I want to ask an uncomfortable question that the investment headlines tend to skip: Are we funding the right layer of the problem?

Because from where I sit, a significant portion of that $21 billion is being deployed based on a fundamental misunderstanding of where the real problems within educational institutions exist—and what it will actually take to solve them.

The Layer Nobody Is Funding

Here is what most EdTech investors do not see in a pitch deck: the degree-audit rules that determine whether a student is on track to graduate are configured manually, institution by institution, inside rule-based systems that require ongoing expert maintenance.

When a curriculum changes, when a transfer-equivalency agreement is updated, or when a new program launches, someone has to update those rules. If they do not, the system provides students with incorrect information—silently, confidently, and at scale.

I configure these systems for a living. I work with Ellucian Banner and DegreeWorks, enterprise student information platforms used by more than a thousand U.S. institutions. I write the Scribe logic that determines whether a student's completed coursework satisfies their program requirements.

When that logic is accurate, everything built on top of it—including the AI advising tools institutions are now deploying—can function reliably. When it is not, and at too many institutions it is not, those AI tools become sophisticated mechanisms for delivering incorrect answers to students who trust them.

"We are extraordinarily good at building beautiful tools on top of broken infrastructure. The $21 billion question is whether anyone is funding the infrastructure itself."

A 2024 EDUCAUSE survey found that the majority of AI advising tools deployed at U.S. institutions lack institution-specific knowledge grounding. They answer student questions based on generalized training data rather than the actual policies of the institution the student attends.

This is not a minor footnote. It is the central flaw in the current EdTech investment thesis: We are funding the application layer while systematically underfunding the data layer that makes any of it trustworthy.

Who Pays the Price

I want to be direct about who absorbs the cost of this misalignment because it is not random.

First-generation college students, transfer students, and students at under-resourced institutions are the most likely to rely on AI advising tools as a primary source of academic guidance. They are also the least likely to have the family experience, peer networks, or institutional navigation skills necessary to recognize and compensate for a wrong answer.

When an AI tool confidently tells a first-generation student that a transferred course satisfies a requirement that it does not actually satisfy, that student registers based on that information. They discover the mistake in their final semester. They pay for an extra term they cannot afford. They may not graduate on time at all.

This is not an edge case. It is the predictable operational consequence of deploying AI advising tools without first ensuring that the underlying institutional data is accurate.

And it means that the equity gap EdTech investment promises to close is, in many cases, being widened by the very tools funded to address it.

Where the Right Bets Actually Are

The highest-return investments in higher education technology right now are not in another consumer-facing chatbot. They are in the infrastructure layer that makes any chatbot trustworthy.

Accurate degree-audit configurations.

Maintained transfer-credit equivalency tables.

NLP systems that can automate exception processing and identify at-risk students earlier in their academic journeys.

Predictive transfer-credit mapping that reduces time to degree for the one in three U.S. college students who transfer between institutions.

Unified student-success dashboards that give institutions a real-time, accurate view of where their students actually stand.

This is the work I am building toward in my own research and practice: AI-augmented degree-audit logic that reduces manual advising intervention, automated exception processing that surfaces equity gaps earlier, and Banner ERP data integration that connects the institutional data layer to the student-success tools that depend on it.

Not because it is the most visible corner of EdTech, but because it is where the gap is largest and the impact is most direct.

The professionals doing this work—the systems analysts, registrar technologists, and degree-audit administrators—are among the most consequential and least celebrated people in higher education.

Their work is invisible when it is done correctly and catastrophic when it is not.

Yet they remain consistently underfunded relative to the AI products being layered on top of the systems they maintain.

The Question for 2026

Twenty-one billion dollars is a serious commitment to the idea that technology can transform education.

The question worth asking as that capital continues to flow is not whether the number is large enough.

It is whether it is aimed at the right problems.

The EdTech industry has spent a decade asking what new tools we can build for students. That instinct is right.

But the more consequential question for the next decade is what foundational infrastructure institutions need for any of those tools to actually work—for every student, not just the ones who were already going to be fine.

From where I sit, inside the systems that make everything else possible, we have more work to do on that question than the investment headlines suggest.

The money is there.

Now we need the wisdom to aim it better.