EdTech deals rose 11% while the median funding round shrank 31%

Preply raised $150 million, the most of any EdTech company in the past 12 months, according to NewMarketPitch. That round stands out in a market where the typical round got a lot smaller. The firm tracked 108 deals worth at least $907 million in the last 12 months, against 97 deals and at least $1.1 billion in the 12 months before. Deal activity rose 11%, while disclosed funding fell 19% and the median round dropped 31%.

Taking out the giant rounds doesn't change the picture. Leave aside everything above $50 million and startups still raised at least $587 million, down 25%. The money is spread across more companies, and each one gets less of it.

Startup funding sounds like an investor story. It lands on the person who signs the software contract: a district technology director, a curriculum lead, a corporate learning manager. Over the next few budget cycles, these are the vendors that buyer will be sizing up.

First checks up, Series A down

Investors got more willing to back companies right at the start. Among rounds where NewMarketPitch could identify the financing type, first financings almost doubled. Most of those first checks were small. Seed activity grew, while Series A rounds, usually the first big check that pays for a sales team, implementation staff and product hardening, became less common.

EdTech now has 37 first financings where it had 19, and a thinner Series A pipeline behind them: more products are starting, and fewer are getting the money to grow up.

The recent deal table shows the pattern. Medly AI raised an $8 million seed round led by Felix Capital for AI exam-prep tutoring. Aristotle took a $5 million seed from True Ventures for AI tutoring across school subjects, and Ed.ai raised $5.8 million from Bpifrance Digital Venture for AI grading and feedback. Series A rounds still happen, but they tend to be specific bets: Arivihan's $10 million from Accel and Prosus Ventures for personalized AI tutoring, and Bites' $40 million round led by PeakSpan Capital for mobile employee training.

A second tracker points the same way. HolonIQ recorded about $1 billion of global EdTech venture investment in the first half of 2026, down 26% from $1.35 billion in the same period in 2025. According to NewMarketPitch, deal volumes in that half-year report remained relatively stable while total capital declined. The two firms count different things over different periods, so the totals can't be compared directly. The direction matches, though: steady or rising deal counts, smaller checks.

Tutoring and workforce skills draw the money

In NewMarketPitch's breakdown, tutoring, test prep and adaptive learning took the most disclosed money, and employers kept funding software that teaches staff skills they use on the job. AI runs through a lot of this but doesn't explain all of it. The tracker points to Coursera's bet on Andrew Ng's LearnVector as a play on a new kind of personal instruction. It also notes that established tutoring and workforce businesses are raising money to expand products people already use.

Deal counts tell a slightly different story from dollar totals. HolonIQ's 2025 review, as compiled by BlueTree, splits deal volume by segment. Workforce training leads, with K-12 close behind.

These shares count transactions, not dollars, so one large round can make a segment look bigger by value than by activity. For corporate learning buyers, the useful point is depth: workforce training had the most deals of any segment. That suggests a crowded field of tools competing for the same training budget. More competitors usually give a buyer more bargaining power, as long as the buyer can tell a durable product from a well-funded demo.

Districts are asking for proof of use

On the buying side, Mordor Intelligence estimates US K-12 EdTech spending at $31.24 billion in 2026. It projects $53.08 billion by 2032, a 9.24% annual growth rate. The firm says districts are moving from emergency, pandemic-era purchases to planned procurement as ESSER relief funding runs out. They're looking harder at evidence, student data protection, cybersecurity and interoperability. Title I districts also need documented justification under ESSA evidence rules when they pick or renew software, and many are swapping single-purpose tools bought in 2020-2022 for integrated platforms. One example is Renaissance linking Star Assessments with Savvas Realize in August 2025.

Amira Learning shows how the funding and procurement sides can meet. Georgia picked Amira as its free, state-approved universal reading screener for the 2025-2026 school year, according to Mordor. NewMarketPitch's deal table lists a $5 million deal for Amira Learning in September 2026, with the stage marked as unknown. A state contract gives a company a distribution route that a seed or Series A round can't buy.

Proof also covers whether students use the thing. Two August 2026 NBER working papers by Philip Oreopoulos and co-authors, both run in Tennessee's Hamilton County Schools, make the point with unusual precision. In a two-year randomized trial across 18 middle schools, students assigned to Khan Academy with its Khanmigo tutor gained about 1.3 national percentile ranks per term, or 0.06 to 0.08 standard deviations over a school year. That's about the same as Khan Academy practice without AI. Nearly every student, 96%, tried Khanmigo. The median student messaged it on only a third of practice days, and in just 17% of sessions where they got something wrong.

The companion study, with more than 6,000 students on the NUMI practice platform, found that AI support changed what happened after a mistake. Students with AI got more next attempts right and needed fewer tries to get back to a correct answer. The best delayed-test results showed up when the AI was built into a mastery workflow, and even those gains were only marginally significant. Both papers are working papers, not yet peer reviewed, and both cover middle-school math. They suggest the same thing for buyers: how the product is designed and how often students engage with it matter more than whether an AI feature exists.

Customer spending is still rising

The strongest objection to worrying about young vendors is that customers keep spending. Grand View Research, cited in NewMarketPitch's October analysis, estimates global EdTech revenue rose from $187 billion in 2025 to $213.2 billion in 2026. A startup that sells into that demand can grow on revenue without a big Series A. Workforce products show it: Docebo's average enterprise contract value rose 27%, and Stride's career-learning programs grew faster than its general education business.

Consolidation is the other route. PowerSchool and Instructure were acquired in deals worth about $10.4 billion combined, which reflects how much buyers value software that's already embedded in institutions. For a district, a small vendor building on a major platform's integrations could end up more durable than its funding stage suggests. The thin-funding concern matters most for standalone tools that would have to carry the whole integration and support load themselves.

Questions for the next contract

1. What stage is the company at, and how many school or employer customers of our size does it support today?
2. Which rostering and data standards does the product already support, and which integrations are still on the roadmap?
3. What share of students in a comparable deployment used the core feature each week, and what did a comparison group show on learning outcomes?

The signal to watch is the Series A count in NewMarketPitch's next quarterly updates. If it recovers, this year's crop of seed-funded tutoring and grading tools will have the money to build what districts are asking for. If it keeps shrinking, more of those 37 first-time companies will need a state contract, a platform partner or a buyer willing to wait.