Edtech: capital has returned, but it is no longer buying the same thing
Edtech: capital has returned, but it is no longer buying the same thing
Edtech: capital has returned, but it is no longer buying the same thing
Share
On September 1, 2026, Unacademy, one of India's most highly funded online education platforms, was sold to its competitor upGrad for $206 million in stock. In 2021, it was valued at 3.44 billion. Its Chief Executive Officer, Gaurav Munjal, summarized the deal in one sentence: "We raised at a peak, but sold at a fraction." We raised at the peak, we sold at a fraction.
Two weeks later, on September 16, Brighteye Ventures, Europe's leading fund specializing in learning and work, announced an initial closing of $72 million for its third fund, after measuring that Europe had invested 1.4 billion euros in the sector in six months—nearly 90% of the entire year 2025, which was itself twice that of 2024.
These two news stories describe the Edtech of 2026 better than any report. Capital has returned, massively. But it is no longer funding the same companies or the same promises. For a leader, an advisor, or a fund committed to education, the question is no longer whether the sector has a future, but what type of leadership this new cycle demands.
Key Figures to Know
Indicator | Value | Source |
|---|---|---|
Investments in Europe in learning and work, 2025 | 1.6 billion euros, compared to 710 million in 2024, at the level of the 2021 peak | Brighteye Ventures, 2026 report |
Investments in Europe, first half of 2026 | 1.4 billion euros, approximately 90% of the 2025 total | Brighteye Ventures, H1 2026 edition |
Global Edtech venture capital in the strict sense, first half of 2026 | $1 billion, down 26% year-on-year | HolonIQ, July 14, 2026 |
Private Edtech Unicorns | 14, for a cumulative valuation of $34.2 billion | HolonIQ, May 18, 2026 |
Revenue of the French Edtech sector, 2025 | 1.8 billion euros, 550 companies, 16,000 jobs | EY-Parthenon and EdTech France, June 2026 |
Edtech fundraising in France, 2025 | 80 to 90 million euros, compared to 409 million in 2021 | EY-Parthenon and EdTech France, June 2026 |
First obstacle cited by French Edtechs | Slowness of public purchasing cycles, 73% of respondents | EY-Parthenon and EdTech France, April 2026 |
Preply, Series D | $150 million at 1.2 billion, positive EBITDA for twelve months | TechCrunch, January 21, 2026 |
Unacademy, sale to upGrad | $206 million, 94% below the 2021 peak | TechCrunch, September 1, 2026 |
AI Tutor in Nigeria, randomized trial | Gain of 0.3 standard deviations in six weeks | World Bank, January 9, 2025 |
Contribution of online training in France | 130 million euros in public revenue per year for 80 million in public cost | Asterès for EdTech France, August 2026 |
Why This Subject Matters Now
Three series of events overlapped between January and September 2026, and it is their combination that makes this topic urgent.
Capital has returned to Europe at an unprecedented level. 1.6 billion euros in 2025, the level of the 2021 peak, followed by 1.4 billion over the first six months of 2026. And for the first time, according to Brighteye, late-stage rounds outnumbered seed and Series A rounds combined. Capital has not only started flowing again; it has changed recipients.
Major players have consolidated. Coursera closed its merger with Udemy on May 11, 2026, creating an entity valued at $2.5 billion, whose second-quarter revenue grew by 60%. Workday paid approximately $1.1 billion for Sana, an enterprise learning platform. Pearson acquired two skills assessment companies within ten days: ITS on September 18 and Workera on September 29, 2026.
The valuations of 2021 have finished correcting. Unacademy sold at 94% below its peak. Byju's, valued at $22 billion in March 2022, was described as "worth zero" by its own founder in October 2024, and that founder was sentenced to six months in prison for contempt by a Singapore court in May 2026. Chegg, whose quarterly revenue fell by 51% in the second quarter of 2026, following two waves of layoffs of 22% and then 45% of its workforce in 2025.
Two false interpretations are circulating. The first concludes that Edtech is a burned-out sector and that investors are turning away. Brighteye's numbers say the opposite. The second concludes that everything is starting up again just as before. Those of HolonIQ, which measures Edtech in the strict sense and notes a 26% drop in global venture capital in the first half of the year, indicate that the money is going elsewhere than in 2021. It is neither an exit nor a return. It is a shift.
What Capital is Buying in 2026
The divergence between the two main observers of the sector is the subject. HolonIQ counts conventional Edtech, tools for schools and universities: $2.6 billion globally in 2025, 1 billion in the first half of 2026. Brighteye counts learning and work, including corporate training: 9 billion euros in 2025, a high since 2018. In Europe, of the 1.6 billion in 2025, conventional Edtech accounts for 471 million and corporate training for 601 million.
What Capital Funded in 2021 | What Capital Funds in 2026 |
|---|---|
Consumer online course platforms with high user growth | Already profitable or near-profitable companies: Preply, EBITDA positive for twelve months before its Series D; Multiverse, first quarter of positive cash flow before its round |
Numerous seed rounds and rapid Series A rounds | Late-stage rounds outnumbering seed and Series A rounds combined |
School education and student support | Corporate skills and employee assessment: Sana acquired by Workday, Workera and ITS by Pearson, LearnVector funded by Coursera |
The promise of a tutor for every student | Measured evidence of impact, and the governance that goes with it |
Acquiring audience at all costs | Consolidation: seventeen private equity transactions in the first half of 2026 compared to ten a year earlier |
Three lessons emerge.
Capital has chosen the employee over the student. This is not a judgment on the social utility of the two segments; it is a reading of economic models. Corporate training is sold to a single, solvent buyer, with a short cycle and proof of return on investment that platforms know how to produce. Schooling is sold to institutions whose purchasing cycle is the first obstacle cited by 73% of French Edtechs. Capital goes where evidence is built fastest.
Capital rewards profitability over growth. Preply raised $150 million at a $1.2 billion valuation in January 2026, with positive EBITDA for a year. Its CEO, Kirill Bigai, summarizes the thesis: "The future of learning is going to be human-guided and amplified by AI." Multiverse announced its first quarter of positive cash flow before raising $70 million. In both cases, the raise comes after the proof.
Capital is concentrating. Brighteye speaks of a "structural shift" toward category leaders. For the others, the consequence is described bluntly by a French entrepreneur quoted in the EY-Parthenon study: "There are going to be quick sales, sometimes forced, because not everyone has the cash to hold out."
What AI Has Proven, and What It Has Not Yet Proven
Technology serving education has measured results, and it is on these results that value is now being built.
In Nigeria, a randomized trial conducted with the World Bank over six weeks in 2024 measured a gain of 0.3 standard deviations for students supported by a tutor based on a large language model, a result that exceeds that of 80% of educational interventions evaluated by randomized trials in developing countries. At Harvard, an AI tutor designed for a physics course roughly doubled the learning gains of 194 students compared to a traditional active learning class. At Stanford, the Tutor CoPilot tool, tested on 900 tutors and 1,800 students, improved student mastery by 4 percentage points, and by 9 percentage points for those whose tutors were the least experienced, at a cost of $20 per tutor per year.
These three studies have one thing in common. AI did not replace a human. It made a teacher or tutor more effective, and the effect is strongest where the human was least equipped. This is the thesis of Ben Wirz, founder of Brighteye, when announcing his third fund: "The defining opportunity of the AI era is not replacing humans with machines. It is expanding what humans can do with machines."
Conversely, the limits are just as well documented. On August 27, 2026, Sal Khan, founder of Khan Academy, wrote that the first version of Khanmigo "did not change student learning as much as many of us hoped." A study by the MIT Media Lab on 54 participants observed the lowest brain connectivity during writing among users of language models. In the UK, 12% of students directly insert text generated by an AI into graded assignments, up from 3% in 2024, according to the Higher Education Policy Institute. And Anthropic's analysis of student usage of Claude shows that about 47% of exchanges involve a request for a direct answer.
What these results outline is a sector where proof of impact has become the primary asset. An Edtech that can show a randomized trial is worth more than an Edtech that shows a number of users. And this proof is generated under a pedagogical and scientific direction that, in most organizations in the sector, does not exist at the executive committee level.
Selling to the Government is Not Selling Software
The second transformation relates to buyers. In France, according to the EY-Parthenon survey for EdTech France of 114 companies in April 2026, 73% cite the slowness of public purchasing cycles as the primary obstacle, 69% cite budget pressure on institutions, and 56% cite the complexity of bidding processes. In the school sector, 48% of revenues come from public buyers; in professional training outside of enterprises, 78% go through intermediary bodies.
The regulator has entered the income statement. In two years, four decisions have altered the economic models of training in France: the co-payment for the personal training account (CPF), set at 100 euros in May 2024 and raised to 150 euros on April 2, 2026; funding ceilings by category in the 2026 budget law; a 20% reduction in apprenticeship funding when 80% or more of the training is conducted remotely, confirmed by the Conseil d'État on October 24, 2025; and the withdrawal of 50,000 teacher licenses for AI tools at the start of the 2025 school year. France compétences, whose state funding fell from 4 billion euros in 2022 to 850 million in 2025, recorded the first drop in apprenticeship starts since 2018.
This context does not doom the sector. Asterès' August 2026 study of 5,559 online learners shows that 69% would not have taken the training in person, an associated salary increase of 5.6%, and 130 million euros in public revenue per year for 80 million in public cost. Online training is a good investment for society. But its rules change every year, and a leader who cannot read a decree as quickly as a sales dashboard does not control their main risk.
This is where the difference in profile comes into play. A leader trained in enterprise software knows how to sell to a chief information officer with a cycle of a few months and an annual contract. A leader who succeeds in schools or state-funded training knows how to build an institutional relationship over several years, anticipate a reform, and model its effect on cash flow. These are two different professions, and a company must know which one it actually practices.
What This Demands of Leadership Roles
The return of capital, the primacy of proof, and the weight of the regulator are shifting the content of six roles.
Role | What It Used to Do | What is Now Demanded |
|---|---|---|
Chief Executive Officer | Grow a user base and raise funds at the right time | Achieve profitability before raising funds, choose between school and corporate markets, and lead consolidation as an acquirer or target |
Chief Financial Officer | Manage growth and cash for a venture-backed company | Model regulatory risk as a market risk, and prepare for an exit to a strategic buyer or private equity fund |
Product Leadership | Integrate generative AI features | Choose between open or closed models, sovereign or non-sovereign hosting, and make the product compliant with regulations that classify learning assessment as high-risk |
Pedagogical & Impact Leadership | Design content | Produce measurable proof of effectiveness acceptable to a public buyer, and champion it at the executive committee level |
Institutional Sales Leadership | Respond to calls for tender | Build multi-year relationships with local authorities, regional education authorities, and universities, in a cycle that 73% of companies in the sector find too slow |
Board of Directors | Validate a growth trajectory | Bring together financial, regulatory, and AI expertise, and decide between a fundraise, a sale, or a merger |
Two of these roles deserve closer examination.
Pedagogical and impact leadership is no longer a support function. Among players raising funds in 2026, proof of effectiveness is the primary sales and valuation argument. The results from Nigeria, Harvard, and Stanford were produced with researchers using randomized protocols and published. An Edtech wanting to sell to a school district or university must be able to produce this type of proof, which requires a scientific profile on the executive committee, not just a content manager.
The board of directors is changing. Pearson appointed three non-executive directors on October 1, 2026: a former CFO with private equity experience, a former Edtech CEO, and a representative of its largest shareholder. Duolingo added a former Citigroup CFO to its audit, risk, and compliance committee in August 2026. The board of an education company targeting an exit in 2026 combines finance, regulation, and artificial intelligence, not just pedagogical expertise.
The Case of the French Sector and Funds
France is one of the five countries that concentrate more than 80% of European transactions, alongside the UK, Germany, Spain, and the Netherlands. Its sector represents 1.8 billion euros in revenue in 2025, 550 companies, and 16,000 jobs, with growth slowing to 6% per year between 2023 and 2025, down from 11% per year in the previous period. Its fundraising fell from 409 million euros in 2021 to 35 to 40 million in 2024, before rising to 80 to 90 million in 2025: Wooclap 25 million, Edflex 15 million, Didask 10 million, Wilgo and Filiz 6 million, Edumapper and EMMA 5 million.
While Europe was returning to its 2021 level, France remained at one-fifth of hers, weighed down by public purchasing in its model and unstable funding rules. Orianne Ledroit, General Delegate of EdTech France, poses the question in these terms in the preface to the EY-Parthenon study: "The question is whether we collectively give ourselves the means to build European champions rather than undergo consolidation coming from elsewhere."
For a fund holding a French Edtech, three questions arise in 2026. The segment: a company whose revenue depends 48% on public buyers has neither the same risk profile nor the same leader profile as a company selling skills to large corporations. Sovereignty: 58% of French Edtechs host in France or Europe and Mistral is the third most used model, but OpenAI remains first at 55%; the trust argument matters to public buyers, provided it doesn't cut them off from the most powerful models. The exit: in a market that concentrates capital on leaders, a portfolio company that is not a category leader must prepare its sale or merger before cash flow forces it to do so.
Common Mistakes
Reading 2021 figures as a benchmark. 2026 capital does not fund the same companies or segments. A valuation that compares itself to five years ago is setting itself up for a correction.
Confusing usage with effectiveness. User count is no longer proof. Public buyers and funds demand impact measurements, and published results show that the effect depends on teacher guidance.
Hiring a software executive to sell to the state. Institutional selling is a different profession, with longer cycles and regulatory risks that enterprise software does not experience.
Treating regulation as a compliance matter. Four public decisions in two years have changed the economic models of the sector in France. The topic belongs to the CFO and the board.
Keeping pedagogical leadership off the executive committee. Proof of effectiveness has become the asset that builds value. The function that produces it must sit where product and sales strategy are decided.
Waiting for the next funding round instead of preparing the exit. In a market that concentrates capital on leaders, a mid-sized company must know which strategic buyer or peer it could merge with before the question arises under duress.
How to Evaluate an Edtech Leader for This Cycle
Did they achieve profitability before raising, or raise to achieve it? The answer reveals whether they experienced the 2021 cycle or the 2026 cycle. The market no longer finances the latter.
Do they know how to produce proof of impact that a public buyer would accept? Ask for the protocol, sample size, who conducted it, and what changed in the product. A leader who answers with user count does not understand what builds value.
Have they ever managed the impact of a regulatory decision on their revenue? What they did in the subsequent three months reveals whether they manage risk or merely suffer from it.
What is their reading of the boundary between school and corporate markets? A leader who talks about both markets as one has not measured the difference in cycles, buyers, and evidence. One who can say which they practice and why has a strategy.
How do they position themselves on the question of exit? In a consolidating market, a leader who has no view on potential acquirers or peers to merge with will leave the board to figure this out at the worst possible moment.
Frequently Asked Questions
How much is Edtech raising in 2026? It depends on the scope. Brighteye Ventures, which counts learning and work, measured 1.4 billion euros in Europe in the first half of 2026, nearly 90% of the 2025 total (1.6 billion, compared to 710 million in 2024). HolonIQ, which counts Edtech in the strict sense, measured $1 billion globally, down 26% year-on-year. Capital is shifting toward corporate skills and category leaders.
Is AI replacing teachers? Published results say otherwise. In Nigeria, Harvard, and Stanford, measured gains come from AI making a teacher or tutor more effective, with the maximum effect seen where the human was least experienced. Sal Khan acknowledged in August 2026 that the first version of Khanmigo did not improve learning as much as hoped. The AI that creates value in education is the one that equips humans.
Why did Byju's fall? Valued at $22 billion in March 2022, the Indian company was described as "worth zero" by its founder in October 2024, and he was sentenced to six months in prison for contempt in Singapore in May 2026. The case illustrates what happens when a valuation is validated without the income statement being able to support it, and without a credible succession plan.
What is the weight of Edtech in France? 1.8 billion euros in revenue in 2025, 550 companies, 16,000 jobs, according to EY-Parthenon and EdTech France. The top twenty companies generate half of the revenue, and fundraising reached 80 to 90 million euros in 2025, compared to 409 million in 2021.
What changed in training funding in France in 2026? The co-payment for the personal training account (CPF) rose to 150 euros on April 2, 2026, the budget law set ceilings by category, apprenticeship funding is reduced by 20% when training is 80% or more remote, and state funding to France compétences was cut to 850 million euros, down from 4 billion in 2022.
What profile of leader are Edtechs looking for in 2026? It depends on the segment. To sell skills to businesses, organizations seek leaders from software and platforms, like Coursera with an Amazon veteran in 2025. To sell to schools or state-funded training, they need leaders who know how to build institutional relationships and model regulatory risk. In both cases, achieving profitability before raising and producing proof of impact has become decisive.
Key Takeaways
Capital has returned to European Edtech: 1.6 billion euros in 2025, 1.4 billion in the first half of 2026, at the level of the 2021 peak.
It is no longer funding the same companies: fewer, larger, later-stage rounds focused on corporate skills and already profitable companies.
The cost of 2021 was paid in 2026: Unacademy sold at 94% below its peak, Byju's at zero, Chegg halved.
Proof of impact has become the primary asset. Published results show that AI creates value when it equips teachers, not when it replaces them.
In France, the regulator has entered the income statement, and selling to the state is a different job than selling enterprise software. The mandate of the Edtech leader has changed with it.
Laroze Partners' Perspective
Education technology is one of the rare sectors where we can measure, with scientific protocols, what technology brings to human beings. A tutor based on a language model helped Nigerian students progress more than most educational interventions ever evaluated. Online training in France brings more to society than it costs. This sector deserves the capital returning to it.
But this capital has learned. It saw what 2021 valuations cost, and it now buys profitability before growth, proof before promises, and leadership teams capable of managing a public buyer, a regulator, and a fund at the same time. The difficulty is not finding a leader who knows education or a leader who knows software. It is finding the one who has already led a company to profitability in a market whose rules change every year. The Laroze Pattern®, our strategic method for reading trajectories, leadership behaviors, and performance dynamics, serves precisely to distinguish, within a career path, what proves this capability from what merely suggests it.
The Edtechs that navigate this cycle will not be those that integrated AI the fastest. They will be those that put on the executive committee and the board the people capable of proving what the company brings, and selling it to those who make the decisions.
Sources
Brighteye Ventures, European Learning & Work Funding Report 2026 and H1 2026 edition · HolonIQ, July 14, 2026 and unicorn list as of May 18, 2026 · EY-Parthenon and EdTech France, La filière EdTech française 2026, June 2026 · Asterès for EdTech France, August 2026 · TechCrunch, January 21 and September 1, 2026 · Coursera, Q2 2026 results · Pearson, press releases of September 18 and 29, 2026 · Workday, September 16, 2025 · World Bank, January 9, 2025 · Harvard Gazette, September 2024 · Stanford, Tutor CoPilot (arXiv) · Khan Academy, August 27, 2026 · Higher Education Policy Institute, March 12, 2026 · Anthropic Education Report, April 8, 2025 · MIT Media Lab (arXiv) · service-public.gouv.fr · France compétences, 2025 activity report · Entrackr, May 2026.
On September 1, 2026, Unacademy, one of India's most highly funded online education platforms, was sold to its competitor upGrad for $206 million in stock. In 2021, it was valued at 3.44 billion. Its Chief Executive Officer, Gaurav Munjal, summarized the deal in one sentence: "We raised at a peak, but sold at a fraction." We raised at the peak, we sold at a fraction.
Two weeks later, on September 16, Brighteye Ventures, Europe's leading fund specializing in learning and work, announced an initial closing of $72 million for its third fund, after measuring that Europe had invested 1.4 billion euros in the sector in six months—nearly 90% of the entire year 2025, which was itself twice that of 2024.
These two news stories describe the Edtech of 2026 better than any report. Capital has returned, massively. But it is no longer funding the same companies or the same promises. For a leader, an advisor, or a fund committed to education, the question is no longer whether the sector has a future, but what type of leadership this new cycle demands.
Key Figures to Know
Indicator | Value | Source |
|---|---|---|
Investments in Europe in learning and work, 2025 | 1.6 billion euros, compared to 710 million in 2024, at the level of the 2021 peak | Brighteye Ventures, 2026 report |
Investments in Europe, first half of 2026 | 1.4 billion euros, approximately 90% of the 2025 total | Brighteye Ventures, H1 2026 edition |
Global Edtech venture capital in the strict sense, first half of 2026 | $1 billion, down 26% year-on-year | HolonIQ, July 14, 2026 |
Private Edtech Unicorns | 14, for a cumulative valuation of $34.2 billion | HolonIQ, May 18, 2026 |
Revenue of the French Edtech sector, 2025 | 1.8 billion euros, 550 companies, 16,000 jobs | EY-Parthenon and EdTech France, June 2026 |
Edtech fundraising in France, 2025 | 80 to 90 million euros, compared to 409 million in 2021 | EY-Parthenon and EdTech France, June 2026 |
First obstacle cited by French Edtechs | Slowness of public purchasing cycles, 73% of respondents | EY-Parthenon and EdTech France, April 2026 |
Preply, Series D | $150 million at 1.2 billion, positive EBITDA for twelve months | TechCrunch, January 21, 2026 |
Unacademy, sale to upGrad | $206 million, 94% below the 2021 peak | TechCrunch, September 1, 2026 |
AI Tutor in Nigeria, randomized trial | Gain of 0.3 standard deviations in six weeks | World Bank, January 9, 2025 |
Contribution of online training in France | 130 million euros in public revenue per year for 80 million in public cost | Asterès for EdTech France, August 2026 |
Why This Subject Matters Now
Three series of events overlapped between January and September 2026, and it is their combination that makes this topic urgent.
Capital has returned to Europe at an unprecedented level. 1.6 billion euros in 2025, the level of the 2021 peak, followed by 1.4 billion over the first six months of 2026. And for the first time, according to Brighteye, late-stage rounds outnumbered seed and Series A rounds combined. Capital has not only started flowing again; it has changed recipients.
Major players have consolidated. Coursera closed its merger with Udemy on May 11, 2026, creating an entity valued at $2.5 billion, whose second-quarter revenue grew by 60%. Workday paid approximately $1.1 billion for Sana, an enterprise learning platform. Pearson acquired two skills assessment companies within ten days: ITS on September 18 and Workera on September 29, 2026.
The valuations of 2021 have finished correcting. Unacademy sold at 94% below its peak. Byju's, valued at $22 billion in March 2022, was described as "worth zero" by its own founder in October 2024, and that founder was sentenced to six months in prison for contempt by a Singapore court in May 2026. Chegg, whose quarterly revenue fell by 51% in the second quarter of 2026, following two waves of layoffs of 22% and then 45% of its workforce in 2025.
Two false interpretations are circulating. The first concludes that Edtech is a burned-out sector and that investors are turning away. Brighteye's numbers say the opposite. The second concludes that everything is starting up again just as before. Those of HolonIQ, which measures Edtech in the strict sense and notes a 26% drop in global venture capital in the first half of the year, indicate that the money is going elsewhere than in 2021. It is neither an exit nor a return. It is a shift.
What Capital is Buying in 2026
The divergence between the two main observers of the sector is the subject. HolonIQ counts conventional Edtech, tools for schools and universities: $2.6 billion globally in 2025, 1 billion in the first half of 2026. Brighteye counts learning and work, including corporate training: 9 billion euros in 2025, a high since 2018. In Europe, of the 1.6 billion in 2025, conventional Edtech accounts for 471 million and corporate training for 601 million.
What Capital Funded in 2021 | What Capital Funds in 2026 |
|---|---|
Consumer online course platforms with high user growth | Already profitable or near-profitable companies: Preply, EBITDA positive for twelve months before its Series D; Multiverse, first quarter of positive cash flow before its round |
Numerous seed rounds and rapid Series A rounds | Late-stage rounds outnumbering seed and Series A rounds combined |
School education and student support | Corporate skills and employee assessment: Sana acquired by Workday, Workera and ITS by Pearson, LearnVector funded by Coursera |
The promise of a tutor for every student | Measured evidence of impact, and the governance that goes with it |
Acquiring audience at all costs | Consolidation: seventeen private equity transactions in the first half of 2026 compared to ten a year earlier |
Three lessons emerge.
Capital has chosen the employee over the student. This is not a judgment on the social utility of the two segments; it is a reading of economic models. Corporate training is sold to a single, solvent buyer, with a short cycle and proof of return on investment that platforms know how to produce. Schooling is sold to institutions whose purchasing cycle is the first obstacle cited by 73% of French Edtechs. Capital goes where evidence is built fastest.
Capital rewards profitability over growth. Preply raised $150 million at a $1.2 billion valuation in January 2026, with positive EBITDA for a year. Its CEO, Kirill Bigai, summarizes the thesis: "The future of learning is going to be human-guided and amplified by AI." Multiverse announced its first quarter of positive cash flow before raising $70 million. In both cases, the raise comes after the proof.
Capital is concentrating. Brighteye speaks of a "structural shift" toward category leaders. For the others, the consequence is described bluntly by a French entrepreneur quoted in the EY-Parthenon study: "There are going to be quick sales, sometimes forced, because not everyone has the cash to hold out."
What AI Has Proven, and What It Has Not Yet Proven
Technology serving education has measured results, and it is on these results that value is now being built.
In Nigeria, a randomized trial conducted with the World Bank over six weeks in 2024 measured a gain of 0.3 standard deviations for students supported by a tutor based on a large language model, a result that exceeds that of 80% of educational interventions evaluated by randomized trials in developing countries. At Harvard, an AI tutor designed for a physics course roughly doubled the learning gains of 194 students compared to a traditional active learning class. At Stanford, the Tutor CoPilot tool, tested on 900 tutors and 1,800 students, improved student mastery by 4 percentage points, and by 9 percentage points for those whose tutors were the least experienced, at a cost of $20 per tutor per year.
These three studies have one thing in common. AI did not replace a human. It made a teacher or tutor more effective, and the effect is strongest where the human was least equipped. This is the thesis of Ben Wirz, founder of Brighteye, when announcing his third fund: "The defining opportunity of the AI era is not replacing humans with machines. It is expanding what humans can do with machines."
Conversely, the limits are just as well documented. On August 27, 2026, Sal Khan, founder of Khan Academy, wrote that the first version of Khanmigo "did not change student learning as much as many of us hoped." A study by the MIT Media Lab on 54 participants observed the lowest brain connectivity during writing among users of language models. In the UK, 12% of students directly insert text generated by an AI into graded assignments, up from 3% in 2024, according to the Higher Education Policy Institute. And Anthropic's analysis of student usage of Claude shows that about 47% of exchanges involve a request for a direct answer.
What these results outline is a sector where proof of impact has become the primary asset. An Edtech that can show a randomized trial is worth more than an Edtech that shows a number of users. And this proof is generated under a pedagogical and scientific direction that, in most organizations in the sector, does not exist at the executive committee level.
Selling to the Government is Not Selling Software
The second transformation relates to buyers. In France, according to the EY-Parthenon survey for EdTech France of 114 companies in April 2026, 73% cite the slowness of public purchasing cycles as the primary obstacle, 69% cite budget pressure on institutions, and 56% cite the complexity of bidding processes. In the school sector, 48% of revenues come from public buyers; in professional training outside of enterprises, 78% go through intermediary bodies.
The regulator has entered the income statement. In two years, four decisions have altered the economic models of training in France: the co-payment for the personal training account (CPF), set at 100 euros in May 2024 and raised to 150 euros on April 2, 2026; funding ceilings by category in the 2026 budget law; a 20% reduction in apprenticeship funding when 80% or more of the training is conducted remotely, confirmed by the Conseil d'État on October 24, 2025; and the withdrawal of 50,000 teacher licenses for AI tools at the start of the 2025 school year. France compétences, whose state funding fell from 4 billion euros in 2022 to 850 million in 2025, recorded the first drop in apprenticeship starts since 2018.
This context does not doom the sector. Asterès' August 2026 study of 5,559 online learners shows that 69% would not have taken the training in person, an associated salary increase of 5.6%, and 130 million euros in public revenue per year for 80 million in public cost. Online training is a good investment for society. But its rules change every year, and a leader who cannot read a decree as quickly as a sales dashboard does not control their main risk.
This is where the difference in profile comes into play. A leader trained in enterprise software knows how to sell to a chief information officer with a cycle of a few months and an annual contract. A leader who succeeds in schools or state-funded training knows how to build an institutional relationship over several years, anticipate a reform, and model its effect on cash flow. These are two different professions, and a company must know which one it actually practices.
What This Demands of Leadership Roles
The return of capital, the primacy of proof, and the weight of the regulator are shifting the content of six roles.
Role | What It Used to Do | What is Now Demanded |
|---|---|---|
Chief Executive Officer | Grow a user base and raise funds at the right time | Achieve profitability before raising funds, choose between school and corporate markets, and lead consolidation as an acquirer or target |
Chief Financial Officer | Manage growth and cash for a venture-backed company | Model regulatory risk as a market risk, and prepare for an exit to a strategic buyer or private equity fund |
Product Leadership | Integrate generative AI features | Choose between open or closed models, sovereign or non-sovereign hosting, and make the product compliant with regulations that classify learning assessment as high-risk |
Pedagogical & Impact Leadership | Design content | Produce measurable proof of effectiveness acceptable to a public buyer, and champion it at the executive committee level |
Institutional Sales Leadership | Respond to calls for tender | Build multi-year relationships with local authorities, regional education authorities, and universities, in a cycle that 73% of companies in the sector find too slow |
Board of Directors | Validate a growth trajectory | Bring together financial, regulatory, and AI expertise, and decide between a fundraise, a sale, or a merger |
Two of these roles deserve closer examination.
Pedagogical and impact leadership is no longer a support function. Among players raising funds in 2026, proof of effectiveness is the primary sales and valuation argument. The results from Nigeria, Harvard, and Stanford were produced with researchers using randomized protocols and published. An Edtech wanting to sell to a school district or university must be able to produce this type of proof, which requires a scientific profile on the executive committee, not just a content manager.
The board of directors is changing. Pearson appointed three non-executive directors on October 1, 2026: a former CFO with private equity experience, a former Edtech CEO, and a representative of its largest shareholder. Duolingo added a former Citigroup CFO to its audit, risk, and compliance committee in August 2026. The board of an education company targeting an exit in 2026 combines finance, regulation, and artificial intelligence, not just pedagogical expertise.
The Case of the French Sector and Funds
France is one of the five countries that concentrate more than 80% of European transactions, alongside the UK, Germany, Spain, and the Netherlands. Its sector represents 1.8 billion euros in revenue in 2025, 550 companies, and 16,000 jobs, with growth slowing to 6% per year between 2023 and 2025, down from 11% per year in the previous period. Its fundraising fell from 409 million euros in 2021 to 35 to 40 million in 2024, before rising to 80 to 90 million in 2025: Wooclap 25 million, Edflex 15 million, Didask 10 million, Wilgo and Filiz 6 million, Edumapper and EMMA 5 million.
While Europe was returning to its 2021 level, France remained at one-fifth of hers, weighed down by public purchasing in its model and unstable funding rules. Orianne Ledroit, General Delegate of EdTech France, poses the question in these terms in the preface to the EY-Parthenon study: "The question is whether we collectively give ourselves the means to build European champions rather than undergo consolidation coming from elsewhere."
For a fund holding a French Edtech, three questions arise in 2026. The segment: a company whose revenue depends 48% on public buyers has neither the same risk profile nor the same leader profile as a company selling skills to large corporations. Sovereignty: 58% of French Edtechs host in France or Europe and Mistral is the third most used model, but OpenAI remains first at 55%; the trust argument matters to public buyers, provided it doesn't cut them off from the most powerful models. The exit: in a market that concentrates capital on leaders, a portfolio company that is not a category leader must prepare its sale or merger before cash flow forces it to do so.
Common Mistakes
Reading 2021 figures as a benchmark. 2026 capital does not fund the same companies or segments. A valuation that compares itself to five years ago is setting itself up for a correction.
Confusing usage with effectiveness. User count is no longer proof. Public buyers and funds demand impact measurements, and published results show that the effect depends on teacher guidance.
Hiring a software executive to sell to the state. Institutional selling is a different profession, with longer cycles and regulatory risks that enterprise software does not experience.
Treating regulation as a compliance matter. Four public decisions in two years have changed the economic models of the sector in France. The topic belongs to the CFO and the board.
Keeping pedagogical leadership off the executive committee. Proof of effectiveness has become the asset that builds value. The function that produces it must sit where product and sales strategy are decided.
Waiting for the next funding round instead of preparing the exit. In a market that concentrates capital on leaders, a mid-sized company must know which strategic buyer or peer it could merge with before the question arises under duress.
How to Evaluate an Edtech Leader for This Cycle
Did they achieve profitability before raising, or raise to achieve it? The answer reveals whether they experienced the 2021 cycle or the 2026 cycle. The market no longer finances the latter.
Do they know how to produce proof of impact that a public buyer would accept? Ask for the protocol, sample size, who conducted it, and what changed in the product. A leader who answers with user count does not understand what builds value.
Have they ever managed the impact of a regulatory decision on their revenue? What they did in the subsequent three months reveals whether they manage risk or merely suffer from it.
What is their reading of the boundary between school and corporate markets? A leader who talks about both markets as one has not measured the difference in cycles, buyers, and evidence. One who can say which they practice and why has a strategy.
How do they position themselves on the question of exit? In a consolidating market, a leader who has no view on potential acquirers or peers to merge with will leave the board to figure this out at the worst possible moment.
Frequently Asked Questions
How much is Edtech raising in 2026? It depends on the scope. Brighteye Ventures, which counts learning and work, measured 1.4 billion euros in Europe in the first half of 2026, nearly 90% of the 2025 total (1.6 billion, compared to 710 million in 2024). HolonIQ, which counts Edtech in the strict sense, measured $1 billion globally, down 26% year-on-year. Capital is shifting toward corporate skills and category leaders.
Is AI replacing teachers? Published results say otherwise. In Nigeria, Harvard, and Stanford, measured gains come from AI making a teacher or tutor more effective, with the maximum effect seen where the human was least experienced. Sal Khan acknowledged in August 2026 that the first version of Khanmigo did not improve learning as much as hoped. The AI that creates value in education is the one that equips humans.
Why did Byju's fall? Valued at $22 billion in March 2022, the Indian company was described as "worth zero" by its founder in October 2024, and he was sentenced to six months in prison for contempt in Singapore in May 2026. The case illustrates what happens when a valuation is validated without the income statement being able to support it, and without a credible succession plan.
What is the weight of Edtech in France? 1.8 billion euros in revenue in 2025, 550 companies, 16,000 jobs, according to EY-Parthenon and EdTech France. The top twenty companies generate half of the revenue, and fundraising reached 80 to 90 million euros in 2025, compared to 409 million in 2021.
What changed in training funding in France in 2026? The co-payment for the personal training account (CPF) rose to 150 euros on April 2, 2026, the budget law set ceilings by category, apprenticeship funding is reduced by 20% when training is 80% or more remote, and state funding to France compétences was cut to 850 million euros, down from 4 billion in 2022.
What profile of leader are Edtechs looking for in 2026? It depends on the segment. To sell skills to businesses, organizations seek leaders from software and platforms, like Coursera with an Amazon veteran in 2025. To sell to schools or state-funded training, they need leaders who know how to build institutional relationships and model regulatory risk. In both cases, achieving profitability before raising and producing proof of impact has become decisive.
Key Takeaways
Capital has returned to European Edtech: 1.6 billion euros in 2025, 1.4 billion in the first half of 2026, at the level of the 2021 peak.
It is no longer funding the same companies: fewer, larger, later-stage rounds focused on corporate skills and already profitable companies.
The cost of 2021 was paid in 2026: Unacademy sold at 94% below its peak, Byju's at zero, Chegg halved.
Proof of impact has become the primary asset. Published results show that AI creates value when it equips teachers, not when it replaces them.
In France, the regulator has entered the income statement, and selling to the state is a different job than selling enterprise software. The mandate of the Edtech leader has changed with it.
Laroze Partners' Perspective
Education technology is one of the rare sectors where we can measure, with scientific protocols, what technology brings to human beings. A tutor based on a language model helped Nigerian students progress more than most educational interventions ever evaluated. Online training in France brings more to society than it costs. This sector deserves the capital returning to it.
But this capital has learned. It saw what 2021 valuations cost, and it now buys profitability before growth, proof before promises, and leadership teams capable of managing a public buyer, a regulator, and a fund at the same time. The difficulty is not finding a leader who knows education or a leader who knows software. It is finding the one who has already led a company to profitability in a market whose rules change every year. The Laroze Pattern®, our strategic method for reading trajectories, leadership behaviors, and performance dynamics, serves precisely to distinguish, within a career path, what proves this capability from what merely suggests it.
The Edtechs that navigate this cycle will not be those that integrated AI the fastest. They will be those that put on the executive committee and the board the people capable of proving what the company brings, and selling it to those who make the decisions.
Sources
Brighteye Ventures, European Learning & Work Funding Report 2026 and H1 2026 edition · HolonIQ, July 14, 2026 and unicorn list as of May 18, 2026 · EY-Parthenon and EdTech France, La filière EdTech française 2026, June 2026 · Asterès for EdTech France, August 2026 · TechCrunch, January 21 and September 1, 2026 · Coursera, Q2 2026 results · Pearson, press releases of September 18 and 29, 2026 · Workday, September 16, 2025 · World Bank, January 9, 2025 · Harvard Gazette, September 2024 · Stanford, Tutor CoPilot (arXiv) · Khan Academy, August 27, 2026 · Higher Education Policy Institute, March 12, 2026 · Anthropic Education Report, April 8, 2025 · MIT Media Lab (arXiv) · service-public.gouv.fr · France compétences, 2025 activity report · Entrackr, May 2026.
On September 1, 2026, Unacademy, one of India's most highly funded online education platforms, was sold to its competitor upGrad for $206 million in stock. In 2021, it was valued at 3.44 billion. Its Chief Executive Officer, Gaurav Munjal, summarized the deal in one sentence: "We raised at a peak, but sold at a fraction." We raised at the peak, we sold at a fraction.
Two weeks later, on September 16, Brighteye Ventures, Europe's leading fund specializing in learning and work, announced an initial closing of $72 million for its third fund, after measuring that Europe had invested 1.4 billion euros in the sector in six months—nearly 90% of the entire year 2025, which was itself twice that of 2024.
These two news stories describe the Edtech of 2026 better than any report. Capital has returned, massively. But it is no longer funding the same companies or the same promises. For a leader, an advisor, or a fund committed to education, the question is no longer whether the sector has a future, but what type of leadership this new cycle demands.
Key Figures to Know
Indicator | Value | Source |
|---|---|---|
Investments in Europe in learning and work, 2025 | 1.6 billion euros, compared to 710 million in 2024, at the level of the 2021 peak | Brighteye Ventures, 2026 report |
Investments in Europe, first half of 2026 | 1.4 billion euros, approximately 90% of the 2025 total | Brighteye Ventures, H1 2026 edition |
Global Edtech venture capital in the strict sense, first half of 2026 | $1 billion, down 26% year-on-year | HolonIQ, July 14, 2026 |
Private Edtech Unicorns | 14, for a cumulative valuation of $34.2 billion | HolonIQ, May 18, 2026 |
Revenue of the French Edtech sector, 2025 | 1.8 billion euros, 550 companies, 16,000 jobs | EY-Parthenon and EdTech France, June 2026 |
Edtech fundraising in France, 2025 | 80 to 90 million euros, compared to 409 million in 2021 | EY-Parthenon and EdTech France, June 2026 |
First obstacle cited by French Edtechs | Slowness of public purchasing cycles, 73% of respondents | EY-Parthenon and EdTech France, April 2026 |
Preply, Series D | $150 million at 1.2 billion, positive EBITDA for twelve months | TechCrunch, January 21, 2026 |
Unacademy, sale to upGrad | $206 million, 94% below the 2021 peak | TechCrunch, September 1, 2026 |
AI Tutor in Nigeria, randomized trial | Gain of 0.3 standard deviations in six weeks | World Bank, January 9, 2025 |
Contribution of online training in France | 130 million euros in public revenue per year for 80 million in public cost | Asterès for EdTech France, August 2026 |
Why This Subject Matters Now
Three series of events overlapped between January and September 2026, and it is their combination that makes this topic urgent.
Capital has returned to Europe at an unprecedented level. 1.6 billion euros in 2025, the level of the 2021 peak, followed by 1.4 billion over the first six months of 2026. And for the first time, according to Brighteye, late-stage rounds outnumbered seed and Series A rounds combined. Capital has not only started flowing again; it has changed recipients.
Major players have consolidated. Coursera closed its merger with Udemy on May 11, 2026, creating an entity valued at $2.5 billion, whose second-quarter revenue grew by 60%. Workday paid approximately $1.1 billion for Sana, an enterprise learning platform. Pearson acquired two skills assessment companies within ten days: ITS on September 18 and Workera on September 29, 2026.
The valuations of 2021 have finished correcting. Unacademy sold at 94% below its peak. Byju's, valued at $22 billion in March 2022, was described as "worth zero" by its own founder in October 2024, and that founder was sentenced to six months in prison for contempt by a Singapore court in May 2026. Chegg, whose quarterly revenue fell by 51% in the second quarter of 2026, following two waves of layoffs of 22% and then 45% of its workforce in 2025.
Two false interpretations are circulating. The first concludes that Edtech is a burned-out sector and that investors are turning away. Brighteye's numbers say the opposite. The second concludes that everything is starting up again just as before. Those of HolonIQ, which measures Edtech in the strict sense and notes a 26% drop in global venture capital in the first half of the year, indicate that the money is going elsewhere than in 2021. It is neither an exit nor a return. It is a shift.
What Capital is Buying in 2026
The divergence between the two main observers of the sector is the subject. HolonIQ counts conventional Edtech, tools for schools and universities: $2.6 billion globally in 2025, 1 billion in the first half of 2026. Brighteye counts learning and work, including corporate training: 9 billion euros in 2025, a high since 2018. In Europe, of the 1.6 billion in 2025, conventional Edtech accounts for 471 million and corporate training for 601 million.
What Capital Funded in 2021 | What Capital Funds in 2026 |
|---|---|
Consumer online course platforms with high user growth | Already profitable or near-profitable companies: Preply, EBITDA positive for twelve months before its Series D; Multiverse, first quarter of positive cash flow before its round |
Numerous seed rounds and rapid Series A rounds | Late-stage rounds outnumbering seed and Series A rounds combined |
School education and student support | Corporate skills and employee assessment: Sana acquired by Workday, Workera and ITS by Pearson, LearnVector funded by Coursera |
The promise of a tutor for every student | Measured evidence of impact, and the governance that goes with it |
Acquiring audience at all costs | Consolidation: seventeen private equity transactions in the first half of 2026 compared to ten a year earlier |
Three lessons emerge.
Capital has chosen the employee over the student. This is not a judgment on the social utility of the two segments; it is a reading of economic models. Corporate training is sold to a single, solvent buyer, with a short cycle and proof of return on investment that platforms know how to produce. Schooling is sold to institutions whose purchasing cycle is the first obstacle cited by 73% of French Edtechs. Capital goes where evidence is built fastest.
Capital rewards profitability over growth. Preply raised $150 million at a $1.2 billion valuation in January 2026, with positive EBITDA for a year. Its CEO, Kirill Bigai, summarizes the thesis: "The future of learning is going to be human-guided and amplified by AI." Multiverse announced its first quarter of positive cash flow before raising $70 million. In both cases, the raise comes after the proof.
Capital is concentrating. Brighteye speaks of a "structural shift" toward category leaders. For the others, the consequence is described bluntly by a French entrepreneur quoted in the EY-Parthenon study: "There are going to be quick sales, sometimes forced, because not everyone has the cash to hold out."
What AI Has Proven, and What It Has Not Yet Proven
Technology serving education has measured results, and it is on these results that value is now being built.
In Nigeria, a randomized trial conducted with the World Bank over six weeks in 2024 measured a gain of 0.3 standard deviations for students supported by a tutor based on a large language model, a result that exceeds that of 80% of educational interventions evaluated by randomized trials in developing countries. At Harvard, an AI tutor designed for a physics course roughly doubled the learning gains of 194 students compared to a traditional active learning class. At Stanford, the Tutor CoPilot tool, tested on 900 tutors and 1,800 students, improved student mastery by 4 percentage points, and by 9 percentage points for those whose tutors were the least experienced, at a cost of $20 per tutor per year.
These three studies have one thing in common. AI did not replace a human. It made a teacher or tutor more effective, and the effect is strongest where the human was least equipped. This is the thesis of Ben Wirz, founder of Brighteye, when announcing his third fund: "The defining opportunity of the AI era is not replacing humans with machines. It is expanding what humans can do with machines."
Conversely, the limits are just as well documented. On August 27, 2026, Sal Khan, founder of Khan Academy, wrote that the first version of Khanmigo "did not change student learning as much as many of us hoped." A study by the MIT Media Lab on 54 participants observed the lowest brain connectivity during writing among users of language models. In the UK, 12% of students directly insert text generated by an AI into graded assignments, up from 3% in 2024, according to the Higher Education Policy Institute. And Anthropic's analysis of student usage of Claude shows that about 47% of exchanges involve a request for a direct answer.
What these results outline is a sector where proof of impact has become the primary asset. An Edtech that can show a randomized trial is worth more than an Edtech that shows a number of users. And this proof is generated under a pedagogical and scientific direction that, in most organizations in the sector, does not exist at the executive committee level.
Selling to the Government is Not Selling Software
The second transformation relates to buyers. In France, according to the EY-Parthenon survey for EdTech France of 114 companies in April 2026, 73% cite the slowness of public purchasing cycles as the primary obstacle, 69% cite budget pressure on institutions, and 56% cite the complexity of bidding processes. In the school sector, 48% of revenues come from public buyers; in professional training outside of enterprises, 78% go through intermediary bodies.
The regulator has entered the income statement. In two years, four decisions have altered the economic models of training in France: the co-payment for the personal training account (CPF), set at 100 euros in May 2024 and raised to 150 euros on April 2, 2026; funding ceilings by category in the 2026 budget law; a 20% reduction in apprenticeship funding when 80% or more of the training is conducted remotely, confirmed by the Conseil d'État on October 24, 2025; and the withdrawal of 50,000 teacher licenses for AI tools at the start of the 2025 school year. France compétences, whose state funding fell from 4 billion euros in 2022 to 850 million in 2025, recorded the first drop in apprenticeship starts since 2018.
This context does not doom the sector. Asterès' August 2026 study of 5,559 online learners shows that 69% would not have taken the training in person, an associated salary increase of 5.6%, and 130 million euros in public revenue per year for 80 million in public cost. Online training is a good investment for society. But its rules change every year, and a leader who cannot read a decree as quickly as a sales dashboard does not control their main risk.
This is where the difference in profile comes into play. A leader trained in enterprise software knows how to sell to a chief information officer with a cycle of a few months and an annual contract. A leader who succeeds in schools or state-funded training knows how to build an institutional relationship over several years, anticipate a reform, and model its effect on cash flow. These are two different professions, and a company must know which one it actually practices.
What This Demands of Leadership Roles
The return of capital, the primacy of proof, and the weight of the regulator are shifting the content of six roles.
Role | What It Used to Do | What is Now Demanded |
|---|---|---|
Chief Executive Officer | Grow a user base and raise funds at the right time | Achieve profitability before raising funds, choose between school and corporate markets, and lead consolidation as an acquirer or target |
Chief Financial Officer | Manage growth and cash for a venture-backed company | Model regulatory risk as a market risk, and prepare for an exit to a strategic buyer or private equity fund |
Product Leadership | Integrate generative AI features | Choose between open or closed models, sovereign or non-sovereign hosting, and make the product compliant with regulations that classify learning assessment as high-risk |
Pedagogical & Impact Leadership | Design content | Produce measurable proof of effectiveness acceptable to a public buyer, and champion it at the executive committee level |
Institutional Sales Leadership | Respond to calls for tender | Build multi-year relationships with local authorities, regional education authorities, and universities, in a cycle that 73% of companies in the sector find too slow |
Board of Directors | Validate a growth trajectory | Bring together financial, regulatory, and AI expertise, and decide between a fundraise, a sale, or a merger |
Two of these roles deserve closer examination.
Pedagogical and impact leadership is no longer a support function. Among players raising funds in 2026, proof of effectiveness is the primary sales and valuation argument. The results from Nigeria, Harvard, and Stanford were produced with researchers using randomized protocols and published. An Edtech wanting to sell to a school district or university must be able to produce this type of proof, which requires a scientific profile on the executive committee, not just a content manager.
The board of directors is changing. Pearson appointed three non-executive directors on October 1, 2026: a former CFO with private equity experience, a former Edtech CEO, and a representative of its largest shareholder. Duolingo added a former Citigroup CFO to its audit, risk, and compliance committee in August 2026. The board of an education company targeting an exit in 2026 combines finance, regulation, and artificial intelligence, not just pedagogical expertise.
The Case of the French Sector and Funds
France is one of the five countries that concentrate more than 80% of European transactions, alongside the UK, Germany, Spain, and the Netherlands. Its sector represents 1.8 billion euros in revenue in 2025, 550 companies, and 16,000 jobs, with growth slowing to 6% per year between 2023 and 2025, down from 11% per year in the previous period. Its fundraising fell from 409 million euros in 2021 to 35 to 40 million in 2024, before rising to 80 to 90 million in 2025: Wooclap 25 million, Edflex 15 million, Didask 10 million, Wilgo and Filiz 6 million, Edumapper and EMMA 5 million.
While Europe was returning to its 2021 level, France remained at one-fifth of hers, weighed down by public purchasing in its model and unstable funding rules. Orianne Ledroit, General Delegate of EdTech France, poses the question in these terms in the preface to the EY-Parthenon study: "The question is whether we collectively give ourselves the means to build European champions rather than undergo consolidation coming from elsewhere."
For a fund holding a French Edtech, three questions arise in 2026. The segment: a company whose revenue depends 48% on public buyers has neither the same risk profile nor the same leader profile as a company selling skills to large corporations. Sovereignty: 58% of French Edtechs host in France or Europe and Mistral is the third most used model, but OpenAI remains first at 55%; the trust argument matters to public buyers, provided it doesn't cut them off from the most powerful models. The exit: in a market that concentrates capital on leaders, a portfolio company that is not a category leader must prepare its sale or merger before cash flow forces it to do so.
Common Mistakes
Reading 2021 figures as a benchmark. 2026 capital does not fund the same companies or segments. A valuation that compares itself to five years ago is setting itself up for a correction.
Confusing usage with effectiveness. User count is no longer proof. Public buyers and funds demand impact measurements, and published results show that the effect depends on teacher guidance.
Hiring a software executive to sell to the state. Institutional selling is a different profession, with longer cycles and regulatory risks that enterprise software does not experience.
Treating regulation as a compliance matter. Four public decisions in two years have changed the economic models of the sector in France. The topic belongs to the CFO and the board.
Keeping pedagogical leadership off the executive committee. Proof of effectiveness has become the asset that builds value. The function that produces it must sit where product and sales strategy are decided.
Waiting for the next funding round instead of preparing the exit. In a market that concentrates capital on leaders, a mid-sized company must know which strategic buyer or peer it could merge with before the question arises under duress.
How to Evaluate an Edtech Leader for This Cycle
Did they achieve profitability before raising, or raise to achieve it? The answer reveals whether they experienced the 2021 cycle or the 2026 cycle. The market no longer finances the latter.
Do they know how to produce proof of impact that a public buyer would accept? Ask for the protocol, sample size, who conducted it, and what changed in the product. A leader who answers with user count does not understand what builds value.
Have they ever managed the impact of a regulatory decision on their revenue? What they did in the subsequent three months reveals whether they manage risk or merely suffer from it.
What is their reading of the boundary between school and corporate markets? A leader who talks about both markets as one has not measured the difference in cycles, buyers, and evidence. One who can say which they practice and why has a strategy.
How do they position themselves on the question of exit? In a consolidating market, a leader who has no view on potential acquirers or peers to merge with will leave the board to figure this out at the worst possible moment.
Frequently Asked Questions
How much is Edtech raising in 2026? It depends on the scope. Brighteye Ventures, which counts learning and work, measured 1.4 billion euros in Europe in the first half of 2026, nearly 90% of the 2025 total (1.6 billion, compared to 710 million in 2024). HolonIQ, which counts Edtech in the strict sense, measured $1 billion globally, down 26% year-on-year. Capital is shifting toward corporate skills and category leaders.
Is AI replacing teachers? Published results say otherwise. In Nigeria, Harvard, and Stanford, measured gains come from AI making a teacher or tutor more effective, with the maximum effect seen where the human was least experienced. Sal Khan acknowledged in August 2026 that the first version of Khanmigo did not improve learning as much as hoped. The AI that creates value in education is the one that equips humans.
Why did Byju's fall? Valued at $22 billion in March 2022, the Indian company was described as "worth zero" by its founder in October 2024, and he was sentenced to six months in prison for contempt in Singapore in May 2026. The case illustrates what happens when a valuation is validated without the income statement being able to support it, and without a credible succession plan.
What is the weight of Edtech in France? 1.8 billion euros in revenue in 2025, 550 companies, 16,000 jobs, according to EY-Parthenon and EdTech France. The top twenty companies generate half of the revenue, and fundraising reached 80 to 90 million euros in 2025, compared to 409 million in 2021.
What changed in training funding in France in 2026? The co-payment for the personal training account (CPF) rose to 150 euros on April 2, 2026, the budget law set ceilings by category, apprenticeship funding is reduced by 20% when training is 80% or more remote, and state funding to France compétences was cut to 850 million euros, down from 4 billion in 2022.
What profile of leader are Edtechs looking for in 2026? It depends on the segment. To sell skills to businesses, organizations seek leaders from software and platforms, like Coursera with an Amazon veteran in 2025. To sell to schools or state-funded training, they need leaders who know how to build institutional relationships and model regulatory risk. In both cases, achieving profitability before raising and producing proof of impact has become decisive.
Key Takeaways
Capital has returned to European Edtech: 1.6 billion euros in 2025, 1.4 billion in the first half of 2026, at the level of the 2021 peak.
It is no longer funding the same companies: fewer, larger, later-stage rounds focused on corporate skills and already profitable companies.
The cost of 2021 was paid in 2026: Unacademy sold at 94% below its peak, Byju's at zero, Chegg halved.
Proof of impact has become the primary asset. Published results show that AI creates value when it equips teachers, not when it replaces them.
In France, the regulator has entered the income statement, and selling to the state is a different job than selling enterprise software. The mandate of the Edtech leader has changed with it.
Laroze Partners' Perspective
Education technology is one of the rare sectors where we can measure, with scientific protocols, what technology brings to human beings. A tutor based on a language model helped Nigerian students progress more than most educational interventions ever evaluated. Online training in France brings more to society than it costs. This sector deserves the capital returning to it.
But this capital has learned. It saw what 2021 valuations cost, and it now buys profitability before growth, proof before promises, and leadership teams capable of managing a public buyer, a regulator, and a fund at the same time. The difficulty is not finding a leader who knows education or a leader who knows software. It is finding the one who has already led a company to profitability in a market whose rules change every year. The Laroze Pattern®, our strategic method for reading trajectories, leadership behaviors, and performance dynamics, serves precisely to distinguish, within a career path, what proves this capability from what merely suggests it.
The Edtechs that navigate this cycle will not be those that integrated AI the fastest. They will be those that put on the executive committee and the board the people capable of proving what the company brings, and selling it to those who make the decisions.
Sources
Brighteye Ventures, European Learning & Work Funding Report 2026 and H1 2026 edition · HolonIQ, July 14, 2026 and unicorn list as of May 18, 2026 · EY-Parthenon and EdTech France, La filière EdTech française 2026, June 2026 · Asterès for EdTech France, August 2026 · TechCrunch, January 21 and September 1, 2026 · Coursera, Q2 2026 results · Pearson, press releases of September 18 and 29, 2026 · Workday, September 16, 2025 · World Bank, January 9, 2025 · Harvard Gazette, September 2024 · Stanford, Tutor CoPilot (arXiv) · Khan Academy, August 27, 2026 · Higher Education Policy Institute, March 12, 2026 · Anthropic Education Report, April 8, 2025 · MIT Media Lab (arXiv) · service-public.gouv.fr · France compétences, 2025 activity report · Entrackr, May 2026.
See more
See more
SaaS Vendors: What Kind of Leader is Needed to Move Away from Per-User Licensing?
SaaS Vendors: What Kind of Leader is Needed to Move Away from Per-User Licensing?
AI Ethics in IT Consulting: Does the AI Act Delay Change the Target Recruitment Profile?
AI Ethics in IT Consulting: Does the AI Act Delay Change the Target Recruitment Profile?
Chief GenIA Officer: The leader of transformation through generative AI
Chief GenIA Officer: The leader of transformation through generative AI
AI and Consulting: Avoiding Groupthink in Executive Teams
AI and Consulting: Avoiding Groupthink in Executive Teams
Data centers in France: Is there a greater shortage of leaders than capital?
Data centers in France: Is there a greater shortage of leaders than capital?
CONTACT
Let's talk about your next recruitment
Outline your needs in a few lines. Your request will be treated with the strictest confidentiality.
CONTACT
Let's talk about your next recruitment
Outline your needs in a few lines. Your request will be treated with the strictest confidentiality.
CONTACT
Let's talk about your next recruitment
Outline your needs in a few lines. Your request will be treated with the strictest confidentiality.
