What are the main business models in EdTech?

Why are EdTech business models under so much pressure now?

EdTech business models are being stress-tested today because cheap capital disappeared just as generative AI made basic digital learning content much easier to reproduce.

HolonIQ's funding data shows how severe the reset has been. Global EdTech venture funding reached about $2.4 billion in 2024, the lowest level in a decade, after roughly $3 billion in 2023 and around $16 billion in 2021. Funding recovered slightly to $2.6 billion in 2025, but HolonIQ counted only about $1 billion during the first half of 2026. That is stabilization at a much lower level, rather than a return to the pandemic funding boom.

Investors have also become much more selective about what they fund. HolonIQ found that workforce training and K-12 support services attracted a large share of recent capital, while AI-enabled and career-focused companies dominated many of the bigger 2026 rounds. Money is still available, but companies increasingly need a credible answer to a simple question: what are customers paying for that ChatGPT, YouTube or a free course cannot easily give them?

That question cuts through most of EdTech today. Educational information has become abundant. Trusted credentials, human instruction, institutional software, learner attention, distribution and measurable career outcomes remain scarce. The best business models are increasingly built around those scarce pieces.

Is EdTech really one market?

EdTech is better understood as several very different markets because Duolingo, PowerSchool, Pearson VUE, Preply and Guild make money from completely different customer relationships.

A Duolingo subscriber can sign up in a few minutes and cancel just as quickly. PowerSchool sells software that can sit inside a school district for years. Pearson VUE charges candidates and institutions around high-stakes testing. Preply takes a cut when a learner books a tutor. Guild connects employer-funded education budgets with workers and education providers.

The products also have radically different cost structures. Software can usually serve another user cheaply. Tutoring requires another tutor-hour. Virtual schools need teachers and operational infrastructure. Assessment businesses spend heavily on security, recognition and test delivery. Marketplaces can stay relatively asset-light if independent instructors provide the teaching.

Asking whether "EdTech" has good margins or recurring revenue rarely tells us much. We first need to know which part of education the company has chosen to monetize.

Who actually pays for EdTech?

EdTech revenue mainly comes from five payer groups, and the payer usually tells us more about the economics than the learning product itself.

Consumers and parents can buy immediately, which makes consumer EdTech easy to distribute globally, but they can also cancel with almost no friction. Schools and universities move much more slowly, although successful vendors can keep contracts for years. Employers can bring hundreds or thousands of learners through one agreement. Governments can fund entire virtual-school populations or assessment programs. Education providers themselves may pay another EdTech company for technology, distribution or student recruitment.

The difference can be enormous. A consumer language app may earn around $100 a year from one subscriber. An enterprise learning contract can cover thousands of employees. A school software vendor may sell several products into the same district. A university program partner can participate in thousands of dollars of tuition for every enrolled student.

When we compare EdTech business models, following the money first prevents a lot of confusion.

Who pays?	Typical business model	Examples	What makes it attractive
Learners or parents	Subscription, course purchase, tutoring	Duolingo, Preply, Outschool	Huge addressable market and fast adoption
Schools and universities	SaaS licenses, courseware, platform contracts	Instructure, PowerSchool, Pearson	Recurring contracts and costly switching
Employers	Enterprise learning and education benefits	Coursera, Guild, Pearson	Large contracts tied to workforce budgets
Governments	Assessments, virtual schools, training	Pearson	Large funded populations and long contracts
Education providers	Marketplace commissions, platform fees, tuition sharing	Udemy, Coursera, OPM providers	Monetization tied to provider distribution or enrollment

Are consumer EdTech subscriptions still a great business model?

Consumer EdTech subscriptions can still be excellent businesses today, but the subscription works only when the free product creates enough habit and value to keep people coming back.

Duolingo is currently the clearest example. In its latest Q2 2026 results, the company reported 58.7 million daily active users, up 23% year over year, alongside 12.7 million paid subscribers, up 17%. Quarterly revenue reached $298.5 million, up 18%.

Those numbers reveal why the model works. Duolingo does not need every learner to pay. Tens of millions use the free product, creating word of mouth, product data and a massive conversion funnel. A relatively small portion of that audience can then pay for Super or Max subscriptions while the free tier keeps bringing new people into the system.

The combined Coursera-Udemy business shows a different version of the same shift. Under the combined company's new reporting method, paid subscribers reached 1.66 million in Q2 2026, up 44% from the comparable combined figure a year earlier.

The hard part, then, is not putting educational content behind a recurring payment. Consumer subscriptions become powerful when the product already gives people a reason to return several times a week. Without that behavior, monthly billing simply makes cancellation easier to measure.

Why is Duolingo growing while Chegg is collapsing?

Duolingo is growing while Chegg is collapsing because Duolingo monetizes an ongoing learning habit, while Chegg depended much more heavily on paid access to academic answers that AI can now provide elsewhere.

The latest numbers make the contrast difficult to ignore. As we saw above, Duolingo's Q2 2026 revenue grew 18% and daily active users increased 23%. Chegg reported $51.8 million of revenue in the same quarter, down 51% year over year. Its Academic Services business has been shrinking so quickly that Chegg stopped emphasizing separate revenue guidance for it and is redirecting investment toward workforce skilling.

Chegg's skilling revenue actually grew 2% in the latest quarter, which makes the comparison more interesting. The company is not suffering from a general collapse in demand for education. The part closest to traditional homework help is shrinking, while the part connected to employability is holding up better.

AI explains a large part of that divergence. A student who wants an explanation of an economics problem can now ask ChatGPT, Gemini or another model immediately. Replacing a language-learning routine with streaks, progress history, personalized exercises and social motivation requires much more than generating a correct answer.

The lesson is blunt: recurring billing helps only when recurring behavior comes first.

Latest comparison	Duolingo	Chegg
Latest quarterly revenue growth	+18%	-51%
Core user behavior	Frequent, repeated learning sessions	Academic help when a student needs an answer
Main AI effect	Adds features and lowers content-production costs	Recreates a large part of the traditional value proposition
Current business direction	Expanding its learning ecosystem	Moving resources toward skilling and employability

If you want more recent data on this point, please see our latest EdTech market report.

Are one-off online course sales dying?

One-off online course sales are clearly losing ground to subscriptions at the biggest horizontal learning platforms, although they are unlikely to disappear completely.

Udemy gives us unusually clean evidence. In 2025, subscriptions generated $566 million, or 72% of total company revenue, up from 66% a year earlier. Consumer subscription revenue grew 44%, while overall Consumer revenue fell 9% because transactional purchases were declining.

The trend continued immediately before Udemy combined with Coursera. During Q1 2026, Udemy's consumer subscription revenue rose from $9.3 million to $14.4 million year over year, an increase of roughly 55%. Transactional and other consumer revenue went the opposite way, falling from $63.3 million to $44.1 million, roughly 30%.

That shift makes sense for a large catalog. If someone wants to take one specific photography course, buying it once still works perfectly well. But when a platform has thousands of relevant courses, coding exercises, AI tutors and constantly changing professional content, access to the library can become more useful than ownership of one course.

Udemy itself had already decided to prioritize recurring offerings before its combination with Coursera. The numbers suggest that this was more than a pricing experiment. The economics of horizontal online learning are steadily moving from "sell another course" toward "keep the learner inside the platform."

How do EdTech marketplaces make money?

EdTech marketplaces make money by connecting learners with instructors and keeping part of the transaction, with the highest take rates usually appearing when the platform itself brings the customer.

Udemy's current revenue-sharing model makes that visible. When an instructor generates a course sale through their own coupon or referral link, the instructor receives 97% of net revenue. When the learner comes through Udemy's marketplace or advertising, the instructor receives 37%. Udemy captures far more value when it owns the demand.

Preply uses a similar principle for live tutoring. The platform currently takes 100% of the value of a new student's trial lesson, then charges tutors between 18% and 33% on subsequent lessons, with the commission falling as tutors accumulate teaching hours.

Outschool's model is simpler. Its latest teacher policy sets the teacher fee at 30% of the class price, leaving 70% for the teacher. Outschool also charges parents a separate marketplace fee.

Those percentages can look high until we consider what the platform is selling to the instructor. Payments are the easy part. The valuable piece is access to learners. A teacher who can independently attract 100 paying students has little reason to hand a platform 30% of revenue. A teacher who would otherwise attract ten may happily do it.

Distribution is the real product on the supply side of many EdTech marketplaces.

Platform	Current economics for educators	What the platform is really charging for
Udemy	Instructor gets 97% on self-generated sales and 37% on platform-generated sales	Learner acquisition and marketplace demand
Preply	100% of first trial lesson, then 18% to 33% commission	Matching, demand, payments and repeat booking
Outschool	30% teacher fee	Parent acquisition, trust and class infrastructure

If you want more recent data on this point, please see our latest EdTech market report.

Why are EdTech companies chasing enterprise customers?

EdTech companies are chasing enterprise customers because one employer can bring thousands of learners, recurring contracts and much higher switching friction than individual course purchases.

Udemy had already become primarily an enterprise-learning company before combining with Coursera. Udemy Business generated $524 million in 2025, around two-thirds of total revenue, and ended the year with $540 million of annual recurring revenue.

The newly combined Coursera business makes the scale clearer today. In Q2 2026, enterprise revenue reached $140 million and enterprise gross margin was 79.3%. Under the unified post-merger reporting method, the business had 12,107 enterprise customers.

There is still pressure inside the model. Enterprise net retention was 91%, down from a comparable 95% a year earlier. Companies do cut learning budgets, consolidate vendors and negotiate pricing. Large contracts are more durable than consumer purchases, but they are hardly automatic renewals.

What employers buy has also changed. A library of videos alone is increasingly easy to replace. Companies want skill mapping, AI practice, analytics, role-specific pathways, integrations and proof that employees actually learned something useful.

The enterprise opportunity favors platforms that can become part of how a company manages skills, rather than another content subscription sitting in the HR budget.

Why is school software worth billions?

School software can be worth billions because once a product becomes part of a district or university's daily workflow, replacing it can be expensive, disruptive and politically painful.

The acquisition market gave us a useful reality check after the pandemic EdTech boom ended. Bain Capital agreed to acquire PowerSchool in a transaction valuing the company at $5.6 billion. KKR and Dragoneer acquired Instructure, the company behind Canvas, at an enterprise value of roughly $4.8 billion. Together, buyers committed more than $10 billion to two institutional education-software companies.

These products sit much deeper inside education than a study app. PowerSchool touches areas such as student records, enrollment, attendance, grades and communications. Canvas can become the place where faculty distribute coursework, manage classes, collect assignments and connect other university systems.

Selling into schools is painful. Procurement takes time, data privacy matters, implementation requires support, and budgets can be rigid. Yet those same obstacles help incumbents after a contract is won. A university replacing its learning-management system may need to migrate thousands of courses, retrain faculty, reconnect other software and move years of data.

Institutional EdTech often behaves like vertical SaaS. Growth can be slower than in consumer apps, but a vendor that becomes part of the operating system of a school can build unusually sticky revenue. Not glamorous, but sticky.

If you want more recent data on this point, please see our latest EdTech market report.

Can textbook publishers still make money online?

Digital publishers can still make substantial money in EdTech, but the strongest model today increasingly gets the product into a course automatically instead of hoping each student chooses to buy it.

Pearson's 2025 results show the transition. Its Higher Education division generated £775 million of sales, while US Inclusive Access grew 19%. Inclusive Access puts digital course materials directly into the institution's course experience, usually making them available from the start of class.

Cengage has been moving in the same direction through Inclusive Access and Cengage Unlimited Institutional. Its institutional model bundles content into a university relationship instead of relying entirely on individual students shopping for a textbook.

That changes the economics. A standalone digital textbook competes with used books, piracy, open educational resources, online summaries and increasingly capable AI systems. Once the material is selected by the institution, integrated into the learning-management system and distributed across an entire class, revenue becomes much more predictable.

Publishers are becoming more interesting when we view them as course infrastructure businesses. The content still matters, but distribution through the institution now does much more of the economic work.

How do online tutoring platforms make money?

Online tutoring platforms usually take a commission or sell recurring memberships, but human teaching costs stop them from enjoying the same scalability as pure software.

Preply and Outschool use marketplace economics, leaving most teaching capacity with independent instructors. That keeps the platform relatively asset-light while letting supply expand across countries, languages and subjects.

Varsity Tutors has pushed further toward recurring membership. Nerdy's latest full-year results showed that Learning Memberships generated $41.6 million in Q4 2025, representing 85% of company revenue. The company ended the year with 33,200 active members, and average monthly revenue per member reached $364, up 21% year over year.

The catch appears once usage increases. A Duolingo subscriber can complete another exercise at almost no incremental labor cost. A tutoring member who consumes another hour of live instruction needs another hour from a tutor. That puts a ceiling on how software-like the margins can become.

AI could change the mix. A learner may eventually spend most of the week practicing with software and use a human tutor only for difficult concepts, feedback or motivation. If platforms can deliver more learning without increasing tutor-hours at the same rate, membership economics improve substantially.

For now, tutoring remains a hybrid business. Recurring payments make revenue more predictable, while the amount of human teaching inside the product still determines much of the margin.

Why are testing and certification such valuable EdTech businesses?

Testing and certification are among the strongest EdTech models because customers are paying for trusted proof of knowledge, and trusted proof remains much scarcer than educational content.

Pearson makes the scale visible. Its Assessment & Qualifications division generated £1.60 billion of sales in 2025, around 45% of Pearson's £3.58 billion group revenue. That division includes professional testing, student assessments, clinical assessments and formal qualifications.

The attraction goes beyond size. A professional certification or admissions test becomes useful when universities, employers, governments or regulators recognize it. Once that recognition exists, candidates have a reason to pay for the test even if the learning material required to prepare for it is available cheaply elsewhere.

The Duolingo English Test shows how an EdTech company can build the same model from the other direction. Duolingo first built a massive language-learning audience and then created an English proficiency test accepted by thousands of education programs.

Generative AI can produce an unlimited number of practice questions. It cannot simply declare that a score is trusted by 5,000 universities. Identity checks, security, psychometrics, institutional acceptance and regulatory relationships take years to build.

Credentialing looks particularly interesting now. AI keeps lowering the cost of learning information while potentially increasing the value of reliable evidence that someone actually knows something.

If you want more recent data on this point, please see our latest EdTech market report.

How do virtual schools make money?

Virtual-school EdTech companies make money from student enrollment and public education funding, giving them far more revenue per learner than a typical education app but also much heavier operating costs.

Pearson's Connections Academy is a good example. Pearson partners with public schools and helps provide curriculum, teachers, enrollment systems, student support and other parts of the virtual-school operation.

Pearson's Virtual Learning division generated £511 million in 2025. Underlying sales grew 8%, helped by a 13% increase in enrollment for the 2025/26 school year. Pearson later reported enrollment growth reaching 15% as demand remained strong.

The key unit here is closer to a student-year than an app subscription. When enrollment rises, Pearson participates in education funding attached to that student. The revenue available per learner can therefore be many times larger than the annual price of a consumer learning app.

Costs rise too. Teachers, curriculum, support and school operations remain real expenses, so virtual-school economics will never look exactly like software margins.

Still, the model shows how large EdTech can become when a company moves from selling a tool used in education to operating part of education itself.

Why do employers pay for workers' education?

Employers pay for EdTech when education becomes cheaper than continually recruiting new people, losing existing workers or leaving important jobs unfilled.

Guild has built an entire B2B2C model around that idea. Employers fund degrees, certificates and career programs for workers, while Guild connects the employer, learner and education provider and helps handle pathways, payments and support.

Guild says more than 90% of learners using its Learning Marketplace currently incur no personal cost for tuition, textbooks or fees. Its internal data also says participating learners are 3.5 times more likely to move into another role inside their company than comparable employees who do not use the benefit.

That tells us why the employer can justify spending far more than a consumer normally would. A $5,000 education benefit looks expensive beside a $100 learning subscription. It can look cheap beside replacing an employee, hiring a nurse, filling a technical position or building an internal promotion pipeline.

The model is particularly well suited to sectors with persistent talent shortages. Healthcare, frontline management and technical occupations give employers a concrete reason to connect education spending with specific jobs.

Guild monetizes the infrastructure around workforce mobility more than the lessons themselves. That is a powerful position because the budget comes from labor economics, one of the largest cost bases inside most big companies.

Why did the 2U-style online-degree model break?

The classic 2U online program management model broke because it combined long-term tuition sharing with huge student-acquisition costs and heavy upfront investment, leaving little room for error when online-degree growth slowed.

The original pitch was attractive to universities. An OPM could help build an online degree, provide technology, recruit students, run marketing and offer support without requiring the university to fund everything upfront. In return, the OPM could keep a large percentage of tuition over many years.

At scale, however, acquiring students became brutally expensive. In 2023, 2U generated about $946 million of revenue and spent $372 million on marketing and sales alone, equivalent to roughly 39 cents for every dollar of revenue. The company also spent on curriculum, technology, student support and administration.

Its balance sheet had little tolerance for disappointment. 2U ended 2023 with roughly $897 million of long-term debt and reported a $318 million net loss. The company entered Chapter 11 restructuring in 2024.

The lesson is broader than one failed company. Revenue sharing works when the partner can acquire demand efficiently and keep programs full. If marketing costs keep rising while the provider has already financed program creation and taken on debt, the arrangement becomes very unforgiving, very quickly.

Universities still need online-program technology, recruitment and services. The market has simply become much less enthusiastic about providers financing almost everything upfront in exchange for decades of tuition share.

If you want more recent data on this point, please see our latest EdTech market report.

Is AI creating a new EdTech business model?

AI is changing EdTech products incredibly quickly, but most AI education companies currently make money through familiar freemium, subscription and institutional software models.

MagicSchool is a good example. Teachers can use a free version, individual users can upgrade to Plus for $8.33 per month when billed annually, and schools or districts can buy custom-priced Enterprise plans. The enterprise product adds administration, single sign-on, SIS and LMS integrations, privacy controls and district-wide management.

That progression is revealing. The AI generation itself gets teachers into the product. The school pays for the institutional layer around it.

Coursera is making an even larger AI bet. Shortly after completing the Udemy combination, the company announced a $100 million strategic investment in LearnVector, Andrew Ng's new AI-native learning company. The aim is to build more personalized and adaptive learning experiences across the combined platform.

AI can clearly raise willingness to pay for tutoring, feedback, teacher productivity and personalization. It can also raise costs because every interaction may consume model inference, while traditional educational content can be reused almost indefinitely.

More importantly, general AI models are formidable competitors. An EdTech startup whose entire proposition is "ask an AI tutor a question" has a difficult position when ChatGPT, Gemini and Claude can already do something similar.

The better businesses will connect AI with assets that are harder to copy: student history, curriculum, teacher workflow, institutional distribution, assessments, credentials or a strong existing learning habit. For now, AI is reshaping how EdTech is delivered much faster than it is inventing entirely new ways for EdTech companies to get paid.

Which EdTech business models are actually strongest today?

The strongest EdTech business models today are institutional SaaS, habit-driven consumer subscriptions, assessment and certification, enterprise learning, and marketplaces that genuinely own learner demand.

The common thread is fairly simple. Strong EdTech companies charge for something that remains scarce even when educational content becomes abundant.

School software owns workflow. Assessment companies own trusted proof. Duolingo owns a repeated consumer habit and enormous organic distribution. Enterprise platforms sit inside employer skill systems. Tutoring marketplaces control access to learners. Employer-funded education companies connect learning with career mobility and large corporate budgets.

Several other models still work, but the economics are harder. Virtual schools can produce large revenue per learner but require substantial operations. Tutoring memberships improve retention but still carry human-delivery costs. Digital publishers are becoming stronger where institutions control distribution. Traditional OPM revenue-sharing can still make sense in specific programs, although 2U showed how dangerous the model becomes when student acquisition is expensive and debt is high.

We would be much more cautious today about any EdTech company whose main asset is simply a large library of explanations, lessons or answers. Generative AI keeps making that layer cheaper.

So the winners increasingly monetize recurring access to scarce distribution, workflow, trust, human expertise or outcomes. The technology changes quickly. Those economic advantages change much more slowly.

EdTech business model	How the company gets paid	Examples	Current strength
Consumer freemium + subscription	Free users convert into recurring paid plans	Duolingo	Very strong when usage becomes habitual
Institutional SaaS	Schools pay recurring software contracts	Instructure, PowerSchool	Among the strongest and stickiest models
Enterprise learning	Employers pay for workforce learning access	Coursera, Pearson	Strong, especially when tied to skills infrastructure
Assessment and certification	Candidates or institutions pay for trusted tests and credentials	Pearson VUE, Duolingo English Test	Highly defensible
Course subscription	Learners pay for recurring catalog access	Coursera, Udemy	Increasingly replacing one-off course purchases
Course marketplace	Platform keeps part of instructor-generated revenue	Udemy	Strong when the platform owns demand
Tutoring marketplace	Commission on teacher or tutor revenue	Preply, Outschool	Attractive but less scalable than software
Tutoring membership	Recurring fee for ongoing learning support	Varsity Tutors	Better retention, with meaningful delivery costs
Digital courseware	Institutions or students pay for assigned digital materials	Pearson, Cengage	Strongest when integrated directly into courses
Virtual schooling	Public funding tied to student enrollment	Pearson Connections Academy	Large revenue opportunity, operationally heavy
Employer education benefits	Employers fund employee education and career paths	Guild	Attractive where learning solves workforce problems
Online program management	University fees or tuition revenue share	2U-style OPMs	Much less attractive when acquisition costs are high
AI education software	Freemium, premium subscriptions and institutional licenses	MagicSchool, AI learning platforms	Fast-growing product layer, with business models still familiar

Who is the author of this content?

NEW MARKET PITCH TEAM

We track new markets so founders and investors can move faster

We build living "market pitch" documents for emerging markets: AI, synthetic biology, new proteins, and more. Instead of outdated PDFs or hallucinated LLM answers, our clients get a clean, visual, always-updated view of what's really happening: key players, deals, regulations, and signals that matter. Learn more about us.