Sal Khan was one of the first people on Earth to see GPT-4. OpenAI called him in the summer of 2022, months before ChatGPT existed, and showed him what was coming. He couldn’t sleep that weekend. By March 2023, Khan Academy launched Khanmigo, an AI tutor built on GPT-4, the same day OpenAI unveiled the model to the public. They were a launch partner. While every other education company was figuring out what ChatGPT meant for them, Khan Academy had already been building for seven months. The “obsolete” platform now has 120 million yearly learners. Khanmigo, their AI tutor, grew 731% year over year in the 2024-25 school year, reaching 2 million users. In classrooms alone, adoption went from 40,000 students to 700,000 in a single year, with projections past 1 million for 2025-26. Their teacher tools are free in over 70 countries. In January 2026, Khan Academy signed a deal with Google to put Gemini (Google’s AI) into new Writing Coach and Reading Coach tools for middle and high schoolers. They’re now working with both OpenAI and Google. A peer-reviewed study published in PNAS (one of the top scientific journals in the world) in January 2026, with researchers from Stanford and the University of Toronto, found that more Khan Academy usage is directly linked to higher student test scores. Sal Khan wrote a whole book in 2024 called “Brave New Words” arguing AI would save education. Sam Altman wrote a blurb for it. His TED Talk making the same argument was one of the 10 most-watched of 2023. In October 2025, he was named TED’s “vision steward.” Khan Academy is now the AI education company. That 731% growth happened while students spent 7.7 billion minutes learning on the platform in 2025.

Part 2 on this, because the financial picture tells an even bigger story. In 1984, a University of Chicago psychologist named Benjamin Bloom ran a study that haunted education for 40 years. Students who got one-on-one tutoring outperformed 98% of students in normal classrooms. He called it the “2 Sigma Problem” because the obvious fix, giving every kid a personal tutor, was impossibly expensive. Human tutoring runs $50 to $150 an hour. Scaling that nationwide would cost over $15 billion a year. For four decades, nobody solved this. AI might. The AI tutoring market hit $1.6 billion in 2024 and is projected to reach $8 billion by 2030. The broader AI-in-education market is growing even faster, from $5.9 billion to $32 billion over the same period. But here’s what matters: the money is flowing toward companies that teach, not companies that give answers. Chegg is the clearest example. They charged students $15 to $20 a month for pre-written homework answers. When ChatGPT launched, students realized they could get the same thing for free. Chegg’s stock dropped 99% from its 2021 peak, erasing $14.5 billion in value. They lost over half a million paid subscribers. They’ve cut roughly half their workforce across two rounds of layoffs in 2025. Wall Street’s consensus is “reduce” as of March 2026, which is polite language for “get out.” Khan Academy went the opposite direction. Khanmigo, their AI tutor, is designed around a principle called Socratic questioning, where it refuses to give you the answer and instead asks you questions until you figure it out yourself. It costs $4 a month for families. It’s free for teachers in over 70 countries. A January 2026 study published in PNAS (one of the top peer-reviewed scientific journals) with researchers from Stanford and the University of Toronto found that increased Khan Academy usage leads to measurably higher test scores across 200,000 students. Khan Academy now works with both OpenAI (GPT-4 powers Khanmigo) and Google (Gemini powers their new Writing and Reading Coach tools). They’re a nonprofit doing $107 million a year in revenue with 84% going directly to educational programs, positioned at the center of a market that barely existed three years ago. The company that sold answers got destroyed. The company that taught understanding became the platform. Bloom’s 40-year-old problem might actually have a solution now, and it costs $4 a month.

What has died is not Khan’s work, but the limitation of his reach. In 2026, the value is not in “explaining” (which is already a commodity), but in designing the validation systems so that the student really learns. Sal Khan has gone from being a teacher to being an architect of mass learning.

@AnishA_Moonka Khanmigo still leaves a lot to be desired. Socratic AI tutors are the future but we’re not there yet.

@AnishA_Moonka What did he do differently that weekend versus other early previewers who saw the same thing and built nothing?

@AnishA_Moonka the seven month head start is what made it work. while everyone else was brainstorming use cases, khan academy was already testing with real students. that's the playbook for any startup building on new tech. get access early, ship before the hype cycle peaks.

@AnishA_Moonka Sal khan spent 20 years building the best free education platform on earth and ai didn't make it obsolete, it gave him the tool to finally solve the one problem he couldn't crack alone

@AnishA_Moonka But I would raise the point that Khanmigo is in danger since it is entirely dependent on the frontier models themselves. If OpenAI or Anthropic or Google choose to make a universal Edtech platform - it would wipe Khanmigo out.

@AnishA_Moonka Mr khan really got the first mover advantage here.

@AnishA_Moonka Khan Academy actually became _more_ valuable once tutoring could scale — the bottleneck was never the content, it was the 1:1 feedback loop

@AnishA_Moonka khan couldnt sleep because he saw education getting democratized at scale. meanwhile most people saw gpt-4 and thought cool chatbot

@AnishA_Moonka best AI education company in the world and the moat is a guy who made youtube videos for his cousin in 2006. distribution is everything

@AnishA_Moonka imagine seeing the future and your first thought is homework

@AnishA_Moonka Early access is an advantage. Execution speed is the only moat.

The partnership with both OpenAI and Google is the strategic detail most people will read past. Khan Academy isn't betting on one model provider winning. It's building the education layer that sits on top of whatever frontier model is best at any given moment. That's not a technology choice. It's a positioning choice that makes Khan Academy structurally defensible regardless of which lab wins the capability race.

@AnishA_Moonka Seven months of building while everyone else was still having the conversation. That's the whole advantage.

@AnishA_Moonka a good reminder that incumbents don’t always die in platform shifts. sometimes the one with the best data, trust and reach just absorbs the new layer first

@AnishA_Moonka Obsoleted? Bro’s out here breaking education records.

@AnishA_Moonka early access helped, but their fast classroom integration mattered way more

@AnishA_Moonka Not sure who is going to khan academy or it’s just forced thru school systems From what I can see kids use ChatGPT to study on their own by prompting in their own style

@AnishA_Moonka This is such a killer breakdown, Anish. Khan went from 'obsolete' narrative to literally solving Bloom's 2 Sigma problem at $4/mo while Chegg imploded. Head start + right philosophy wins.

@AnishA_Moonka Have you seen this video? youtube.com/watch?v=0xS68s… It is not clear to me that personalized tutors will solve all problems in education as it takes away a critical aspect that makes people learn: Peer influence.

@AnishA_Moonka Wow you managed to make the original post sadder

@AnishA_Moonka You didn't need to drop a nuke on the guy, jeez just use a flyswatter next time

@AnishA_Moonka While everyone else was panic-searching "Is AI cheating?", Sal Khan was busy building the future of how we learn.

The part of this that doesn't get enough attention: Khan Academy's 7-month head start wasn't just about moving fast — it was about having 20 years of behavioral data on how students actually learn. Khanmigo isn't effective because of GPT-4; it's effective because they knew exactly what GPT-4 needed to do and how to constrain it. That's the pattern in AI products that actually work: domain expertise came first, and the model became the execution layer. EdTech companies that tried to "add AI" without that foundation are still figuring out their use case while Khan Academy is closing 700K-student classroom deals. The 731% growth number isn't an AI story — it's a story about what happens when you have 20 years of accumulated judgment about what a learner needs, and suddenly have a tool to act on it at scale.

Wild numbers. That momentum is huge. But the real win is when teachers steer it, outcomes improve in real classrooms, and kids in every zip code benefit.

@AnishA_Moonka The moat in AI education is not early model access. It is whether students actually understand more, stay longer, and improve. That is why we are building @Teraaiguide around learning depth, not just impressive demos.

@AnishA_Moonka I’m glad they adapted quickly 🙌

@AnishA_Moonka Aren't they a non-profit?

@AnishA_Moonka Great example of how AI plays on in the real world of domain-expertise!