From Spark to Surge: How AI Dominates Y Combinator and Why Higher Ed Should Pay Attention

Sam Altman once joked that Y Combinator startup batch size rivals a small nation’s GDP plan. He might have been half-kidding, but the numbers suggest he was not far off, especially when you notice how many of those startups now have artificial intelligence stamped across their pitch decks.

This is the same accelerator that helped launch some of the most recognizable names in tech: Coinbase , Dropbox , Airbnb , DoorDash , and Stripe . These companies reshaped entire industries, from financial transactions to how we travel, work, and get dinner delivered. That track record makes it worth paying close attention to what is happening inside YC today, because its current crop of startups is overwhelmingly AI-driven.

Meanwhile, Sam Bhagwat's team, best known for co-founding Gatsby, has been quietly building Mastra, an open-source framework for launching autonomous AI agents. In their own ways, both Sams are betting on the infrastructure, the “picks and shovels,” that will power AI’s next gold rush.

For higher education, the kind of orchestration layer Bhagwat is building could be transformative. Universities operate on interconnected workflows that touch nearly every part of the institution, from admissions pipelines to personalized learning paths. These workflows often involve multiple steps, data sources, and decision points. A flexible AI framework like Mastra could become the backbone for automating these processes, allowing small teams to rapidly prototype and deploy tools that improve student services, faculty productivity, and administrative efficiency without reinventing the foundation each time.

What began as my attempt to track EdTech trends inside Y Combinator quickly turned into a bigger discovery: a near-total transformation of the startup landscape. While digging through data on more than 5,000 YC companies, I stumbled onto something dramatic. In Winter 2022, 63 percent of companies were AI-focused. By Winter 2025, that number had surged to 89 percent. 

The AI Gold Rush Playbook

Much like Levi Strauss selling denim to miners, today’s AI boom is powered by tools and infrastructure that help small teams operate at the scale of much larger ones. APIs from companies like OpenAI and Anthropic make advanced AI accessible without requiring a PhD in machine learning. Orchestration frameworks like Mastra chain multiple models together into something that feels seamless to end users. Memory systems address AI’s tendency to forget, while modern development platforms handle scaling automatically.

When these pieces come together, you get the defining trait of this era: lean founding teams that can deliver complex solutions quickly. In higher education, this means a two-person EdTech startup could realistically launch an AI-powered tutoring platform or an automated grading assistant without the staffing and overhead that used to make such projects out of reach.

The Numbers Behind the Shift

Nearly 60 percent of all YC companies to date are AI-focused, but in recent cohorts, the share is much higher, above 80 percent for the last three batches. In Winter 2025, almost nine out of every ten companies had AI at their core. This is a 40 percent increase from just three years earlier.

For EdTech, the footprint is smaller but steady. Around 12 percent of companies in each cohort work in education. Half of those are also AI-first. These are not generic chatbots that have been repurposed for classrooms. They include tutoring copilots that generate custom problem sets in seconds, grading systems that return detailed feedback instantly, and adaptive learning tools that adjust in real time based on student performance.

Higher Education’s Quiet AI Revolution

EdTech rarely makes top headlines, but the sector has been steadily adapting AI to solve long-standing challenges. Many of these startups begin small, with founding teams of three to five people, and focus on proving value early rather than chasing large funding rounds. In higher ed, that often means securing paid pilot programs with universities before even applying to YC.

Some AI-powered tutoring systems are already in pilots with community colleges, where budgets are limited but the need for scalable student support is high. Others focus on helping faculty, with tools that generate quizzes, lesson plans, and even personalized student outreach in the instructor’s own writing style. The common thread is simple: solving real problems in ways that fit the complexity of academic life.

The B2B Reality Check

Despite the buzz around consumer apps, YC data shows that nearly half of all companies are building for business markets. In higher ed, this B2B focus matters. Selling to universities involves navigating procurement, compliance, and integration with legacy systems. These are challenges that reward domain expertise as much as technical skill.

For AI-focused EdTech companies, this often means the technology itself is only half the equation. The other half is knowing how to integrate with learning management systems, meet accessibility requirements, and address governance concerns from faculty and administrators.

Lean Teams, Big Impact

One of the most striking findings in the YC data is how small many of these companies are at the start. Nearly a quarter launch with one or two people. Another 20 percent start with teams of three to five. With modern AI tools and cloud platforms, these small teams can release polished, scalable products in weeks.

For higher education, this lowers the barrier to innovation. A faculty member with an idea for an AI grading assistant no longer needs institutional funding to hire a development team. They can collaborate with a lean startup or even prototype their own solution using existing AI infrastructure.

Crossover Companies to Watch

Some of the most compelling education-related companies in recent YC batches are not traditional EdTech at all. Consider RevisionDojo (YC F24) , Eduvant, Eduflow, Miyagi Labs , GradeWiz (YC W25), and Edexia (YC W25). They are AI-first companies that happen to solve education problems exceptionally well. This includes assessment tools that can evaluate everything from math proofs to art portfolios, personalized learning agents that adjust pacing based on student progress, and orchestration frameworks that combine multiple educational workflows into a single system.

This is where Mastra’s relevance to higher education comes back into focus. Education workflows naturally follow multi-step processes: assess, personalize, deliver, evaluate, repeat. An orchestration layer designed for AI agents could tie these steps together, enabling institutions to create sophisticated, adaptive learning and administrative tools without rebuilding the infrastructure each time. It is the kind of behind-the-scenes capability that can make AI innovation faster, more consistent, and more cost-effective across an entire campus.

What Higher Ed Leaders Should Take Away

The surge from 63 percent AI-focused companies in 2022 to 89 percent in 2025 is more than a startup statistic. It signals that AI is becoming the default technology foundation for new ventures. For higher education, this means:

• Innovation will increasingly come from small, agile teams, not traditional enterprise vendors.

• Domain expertise will be the differentiator. AI tools are available to all, but the winners will be those who deeply understand higher ed challenges.

• The most effective AI solutions will be embedded into workflows, not bolted on as isolated features.

The Bigger Question

Looking across more than 5,000 YC companies worth billions in aggregate value, one thing is clear. This is not just another tech wave. It is the formation of a new startup paradigm where AI capabilities are as essential now as having a website was two decades ago.

The companies being built today will shape the next decade of technology, and higher education will be part of that transformation. Whether you are an investor, an EdTech founder, or a university CIO, the question is the same: will you be riding the upward curve of this ski jump or watching from the lodge?

In this gold rush, the winners will be those who identify a genuine problem in higher ed, solve it better than anyone else with AI, and work with institutions that are ready to embrace the change.

#highered #edtech #artificialintelligence

This analysis is based on comprehensive data from 5,175 Y Combinator companies across all batches through Winter 2025. The data reveals clear trends in AI adoption, industry focus, and team composition that every entrepreneur should understand.