AI has changed what is defensible in Education. For years, many EdTech products built their value around access to content, homework support, tutoring, practice, feedback or study help. Those were meaningful use cases, but they are now much easier for foundation models and large platforms to reproduce, bundle or give away.

Chegg shows how quickly that pressure can appear. In Q1 2025, the company reported a 31% year-on-year decline in subscribers and said the trends affecting the business would worsen before they got better. Later that year, it announced another major restructuring as AI continued to pressure its core homework-help model.

The same pressure is moving into institutions. AI tools are being bundled into the environments schools, universities and companies already use. The front end of learning is becoming easier to generate, personalise and distribute.

That does not mean Education companies have no moat. It means the moat is moving. The defensible product is less likely to be generic content delivery or undifferentiated tutoring. It is more likely to sit in the system of record, the assessment layer, proprietary curriculum, workflow, institutional data, compliance, teacher practice or the tools administrators rely on day to day.

For scale-ups, the question is no longer only how to make learning more engaging. It is what part of the education system the company is becoming hard to replace inside.

The second pattern is the movement of buying power away from individual users and toward institutions. The teacher with a credit card still matters. So does the student, learner or manager who loves the product. But in more categories, the decision is being made higher up: by district leaders, university finance teams, procurement offices, CTOs, multi-academy trusts, CHROs or enterprise software owners.

That changes the commercial motion. In the UK, the Department for Education has been funding AI tools for marking and feedback through national programmes, including a £1 million initiative for developers. Its AI Tools for Education Phase 2 framework is explicitly evidence-led and focused on tools that support assessment and feedback in real educational settings.

In the US, the end of pandemic-era ESSER funding has put further pressure on school budgets and vendor renewals. When emergency money disappears, products that once grew through discretionary adoption have to defend themselves in more formal budget reviews.

The private market is moving in the same direction. Large transactions such as Bain’s acquisition of PowerSchool, Workday’s agreement to acquire Sana for approximately $1.1 billion, and other platform-level moves show how education and learning tools are being pulled into larger enterprise and institutional systems.

For Education scale-ups, this means the buyer is changing faster than the user. A product may still be loved by teachers, learners or employees, but the renewal may be decided by someone asking a different question: does this tool reduce cost, improve outcomes, consolidate the stack, support compliance or prove its role in the system?The companies that adapt well are the ones that stop selling only to enthusiasm and start selling to institutional logic.

The third pattern is the shift from engagement to evidence. Education companies have often grown by showing that people use the product: minutes spent, lessons completed, questions answered, modules finished, learners retained. Those metrics still matter, but they are no longer enough on their own. Buyers are asking what changed because the product was used.

In K-12, that may mean evidence of learning gain, teacher time saved, assessment quality or intervention effectiveness. In higher education, it may mean retention, attainment, employability or student support. In workforce learning, it may mean productivity, skills mobility, internal hiring, compliance or measurable performance improvement.

The logic is visible across the market. Pearson has increasingly emphasised assessment, virtual schools and complex services. Workforce-learning platforms are shifting their language from completion data to business outcomes. Enterprise buyers are asking whether learning products can show productivity, not just participation.

That creates a different evidence burden. A company can have strong engagement and still struggle to renew if it cannot show the outcome the buyer cares about. The evidence does not have to look the same in every segment, but it has to be credible to the decision-maker controlling the budget.

For scale-ups, this means evidence is not a marketing asset added at the end. It becomes part of product, customer success, implementation, pricing and renewal strategy. The companies most likely to survive this phase are the ones that can connect usage to outcomes the buyer already has to defend.