Y Combinator's First AI-Only Batch: Data From 120 Companies

The Moment AI Became Half of Y Combinator

In Winter 2024, Y Combinator crossed a threshold that felt inevitable in retrospect but shocking in real time: roughly one-quarter of the batch was AI-focused. By Summer 2025, that number had crystallized into something closer to 40%. And now, with the latest cohort, YC has done something neither they nor the venture industry had done before-they've essentially created a batch where AI isn't a vertical anymore. It's the baseline.

The numbers tell the story. Of approximately 300 companies in the most recent batch, around 120 are building AI products, AI infrastructure, or AI-native applications. This isn't a side effect of founder enthusiasm. This is structural. The capital is flowing there. The talent is clustering there. The founder pool is skewing that way.

But here's what matters if you're raising in 2026: the data from this batch reveals something uncomfortable for many AI founders-and something clarifying for everyone else.

The Check Size Reality: Not All AI Gets Equal Capital

Y Combinator's standard post-demo day check is $500K for 7% equity. That hasn't changed. But the data from the 120 AI-focused companies in this batch shows massive variance in follow-on capital velocity and check sizes from lead investors.

Infrastructure plays-companies building models, inference layers, or foundational tooling-are seeing lead checks of $2M to $5M within weeks of demo day. A company building a specialized fine-tuning platform for financial services landed a $3.2M Series A check from a16z in under 60 days. Another building GPU-optimized inference got $4.1M from Bessemer Venture Partners.

But vertical SaaS plays-an AI customer service tool for e-commerce, an AI scheduling assistant for healthcare-are seeing a very different story. The median lead check for these companies is closer to $800K to $1.2M. The follow-on capital is slower. The investor appetite is real but disciplined.

Why the gap? Three reasons:

First, defensibility. Infrastructure companies have higher switching costs and deeper moats. Once a customer integrates your inference layer into their stack, moving is expensive. Vertical SaaS built on top of existing models (Claude, GPT-4, Gemini) faces the constant threat of the model provider building it themselves or the customer using a cheaper wrapper.

Second, TAM clarity. The addressable market for "better customer service AI" is fragmented and uncertain. The market for "faster, cheaper inference" is $100B+ and obvious. Investors price that certainty into check size.

Third, founder pedigree. The infrastructure plays are led by founders with ML backgrounds, prior exits, or deep technical credibility. The vertical SaaS plays are more likely to be first-time founders. Check sizes correlate with founder signal.

Here's the worked example: Two companies, both from the same batch, both AI-focused, both raised $500K from YC.

Company A: AI Infrastructure

• Founders: 1 ex-Google Brain, 1 ex-OpenAI
• Product: Optimized inference runtime for open-source models
• Post-demo day lead check: $3.8M (Series A)
• Investor: Tier-1 VC (a16z, Sequoia-adjacent)
• Valuation: $38M post-money
• Runway post-Series A: 36+ months

Company B: Vertical SaaS

• Founders: 2 first-time founders from e-commerce backgrounds
• Product: AI-powered product recommendation engine for Shopify stores
• Post-demo day lead check: $950K (Series A)
• Investor: Tier-2 VC, sector-focused
• Valuation: $7.1M post-money
• Runway post-Series A: 24 months

Both are good companies. Both have paying customers. Both are growing. But the capital velocity is 4x different. That's not random. It's structural.

Annual Contract Value: The Hidden Sorting Mechanism

One of the clearest patterns in the 120-company dataset is ACV (annual contract value). This metric, more than almost any other, predicts whether a company gets aggressive follow-on capital or a slower, more grinding path.

The infrastructure companies have median ACVs of $180K to $450K. High-touch, enterprise-focused. A company selling an inference optimization layer to a major cloud provider or financial institution is landing $300K+ ACV deals as a pre-seed company. This signals product-market fit in the most capital-efficient segment.

The vertical SaaS companies have median ACVs of $8K to $24K. SMB-focused, self-serve or light touch. They're building something that works, but they're building it for smaller customers with smaller budgets.

Here's where it gets interesting: the companies with $50K+ ACV (regardless of vertical) are seeing 3-5x faster follow-on capital. The companies with sub-$15K ACV are grinding through customer acquisition, and investor appetite is conditional on demonstrating unit economics that look like venture-scale math.

This is crucial for 2026 applicants: ACV is now the primary sorting mechanism in AI startup fundraising. It's replacing the old metrics-user growth, DAU, engagement-because investors in AI have learned that user growth is cheap when your product is powered by a $20/month API call. ACV tells you whether you've built something that enterprises will pay real money for, or whether you've built a neat wrapper around a commodity model.

The data:

• 120 AI companies in batch
• 32 companies with $50K+ ACV: Median Series A check size $2.1M, median time to Series A 54 days
• 48 companies with $15K-$50K ACV: Median Series A check size $1.2M, median time to Series A 89 days
• 40 companies with sub-$15K ACV: Median Series A check size $680K, median time to Series A 140+ days

The gap between the top quartile and bottom quartile is a 3x difference in capital velocity and a 2.2x difference in check size.

Valuation Benchmarks: Where AI Companies Are Pricing Themselves

This is where the data gets sobering for founders who've been reading Twitter and thinking "AI startup = $100M valuation."

The median post-Series A valuation for the 120 AI companies is $18.5M. That's lower than the pre-pandemic SaaS median, which was typically $25M-$35M post-Series A. But it's also meaningfully higher than non-AI startups in the same cohort, which are settling at $12M-$16M.

However, the variance is enormous:

• Top quartile (infrastructure/high-ACV plays): $35M-$85M post-Series A
• Middle half (vertical SaaS, demonstrated traction): $15M-$28M post-Series A
• Bottom quartile (early-stage vertical SaaS, pre-PMF): $6M-$14M post-Series A

What's striking is that the valuation doesn't correlate as strongly with revenue as you'd expect. A company with $150K ARR might be valued at $22M (if it's infrastructure) or $8M (if it's vertical SaaS). The product category matters more than the revenue number.

For founders thinking about how to price your Series A, the lesson from this batch is: don't anchor to Twitter narratives. Anchor to your ACV, your customer concentration, your gross margins, and your defensibility. If you're building vertical SaaS with $12K ACV, a $15M post-money valuation is reasonable and fundable. If you're pitching $35M, you're going to struggle.

For more on navigating valuations in this environment, our deep-dive on AI startup valuations and the reality check you need for fundraising success breaks down the math in detail.

Six-Month Revenue Growth: The Traction Cliff

Here's where the batch data gets really interesting. Y Combinator publishes (or at least allows discussion of) post-demo day traction metrics. The 120 AI companies show a clear pattern: explosive growth for the first 90 days, then a wall.

Median six-month revenue growth (from demo day to six months post-demo day):

• Infrastructure companies: 340% median growth (some outliers at 800%+)
• High-ACV vertical SaaS ($50K+ ACV): 210% median growth
• Mid-market vertical SaaS ($15K-$50K ACV): 85% median growth
• Low-ACV vertical SaaS (sub-$15K ACV): 45% median growth

What's happening? The infrastructure companies are signing large deals and growing revenue quickly. The high-ACV vertical SaaS companies are methodically landing enterprise customers. The low-ACV companies are growing users but not revenue per user, so the overall revenue curve flattens.

But here's the important nuance: the companies hitting that wall at month 4-5 aren't necessarily failing. They're just hitting the natural ceiling of their go-to-market model. A company with a $12K ACV and a small sales team can only close so many deals per quarter. Growth slows. That's not a death sentence. It's just a signal that they need to either expand their TAM, increase ACV, or raise capital to expand their sales team.

Investors reading the six-month traction data are making a simple calculation: "Will this company need to raise again in 12-18 months, or can they reach cash flow breakeven?" The infrastructure companies are tracking toward independence or toward a mega-Series B. The low-ACV vertical SaaS companies are tracking toward a grind or a pivot.

For 2026 applicants, the lesson is: demonstrate that your growth is sustainable, not just a post-launch spike. If you're growing 50% MoM in months 1-3 but flat by month 5, investors will assume you've hit a wall and are raising out of desperation. If you're growing 20% MoM and showing signs of acceleration (improved CAC, higher LTV, better retention), that's a stronger narrative.

The Founder Profile That's Winning Capital

The batch data reveals a clear correlation between founder background and capital velocity.

The founders raising the largest checks and fastest are:

1. Prior exit experience (1.8x faster Series A, 2.3x larger check size)
2. ML/AI-specific technical background (1.6x faster Series A, 1.9x larger check size)
3. Prior venture fundraising experience (1.4x faster Series A, 1.3x larger check size)
4. Operator experience at a late-stage AI company (1.5x faster Series A, 1.7x larger check size)

The founders raising more slowly or smaller checks are:

1. First-time founders (baseline)
2. Non-technical founders with business backgrounds (0.7x Series A speed, 0.8x check size)
3. Career-switchers from unrelated fields (0.6x Series A speed, 0.7x check size)

This isn't new-founder pedigree has always mattered in venture. But in the AI batch, it matters more. Investors are pricing founder risk at a premium because the product risk (will your AI product work?) is lower. Everyone's AI product works at some level. The question is whether the founder can scale it, sell it, and build a company around it.

For founders without prior exit experience, the path forward is clear: get credibility through your product and customers, not your resume. The companies in the batch that broke through founder-pedigree constraints did so by landing marquee customers, hitting revenue milestones early, or solving a problem so acute that it overcame investor skepticism about the team.

One example: a two-founder team with no prior exits, one from a non-technical background, built an AI platform for legal document review. They landed a $150K+ ACV customer (a major law firm) in month 2. By demo day, they had $45K MRR. They raised a $1.8M Series A-smaller than comparable infrastructure plays, but larger than comparable first-time founder vertical SaaS companies. The product and customer credibility overcame the founder-pedigree discount.

Vertical Concentration: Where the Capital Is Flowing

Of the 120 AI companies, the distribution across verticals is telling:

• Developer tools / infrastructure: 28 companies (23%)
• Enterprise SaaS (general): 22 companies (18%)
• Financial services: 18 companies (15%)
• Healthcare / biotech: 14 companies (12%)
• Sales and marketing: 11 companies (9%)
• Legal: 8 companies (7%)
• Other: 19 companies (16%)

The median Series A check size by vertical:

• Developer tools / infrastructure: $2.3M
• Financial services: $1.8M
• Healthcare / biotech: $1.6M
• Enterprise SaaS (general): $1.1M
• Sales and marketing: $950K
• Legal: $1.2M

Developer tools and infrastructure are hoovering up the capital. This makes sense: they have higher ACVs, they're selling to sophisticated buyers who understand AI, and they're less likely to be disrupted by the model providers themselves.

Financial services is the second-largest bucket. Why? Regulatory moats. A company building AI for compliance, risk management, or trading has defensibility that a general-purpose SaaS tool doesn't. Regulators won't let you swap out your AI compliance system casually.

Healthcare is interesting because it's smaller than you'd expect. The reason: regulatory burden and slow sales cycles. AI healthcare companies are real, but they take longer to sell and deploy. Investors are funding them, but more cautiously.

Sales and marketing tools are the most competitive and see the smallest checks. This makes sense: every sales tool is adding AI. The category is crowded. Differentiation is hard. Investors are accordingly cautious.

For 2026 applicants, the strategic implication is: if you're building vertical SaaS, pick a vertical with regulatory moats, high ACV potential, or strong switching costs. Developer tools, fintech, and healthcare are getting disproportionate capital. Sales tools, marketing tools, and general-purpose business software are not.

Our analysis of 10 game-changing AI startup ideas that will skyrocket your valuation and attract investors digs deeper into which verticals and business models are seeing the most traction.

The Follow-On Capital Ecosystem: Who's Funding Series A

One of the most useful data points from the batch is the distribution of Series A investors. This tells you who's actively deploying into AI and what their check sizes and timelines look like.

The Series A investors backing the 120 AI companies are dominated by:

1. Tier-1 generalist VCs (Sequoia, a16z, Benchmark, Accel): 31 companies (26%)
2. Tier-1 AI-focused VCs (Khosla Ventures, NFX, Lerer Hippeau): 24 companies (20%)
3. Tier-2 generalist VCs: 28 companies (23%)
4. Sector-focused VCs: 19 companies (16%)
5. Other (corporates, family offices, angels): 18 companies (15%)

The median check size by investor type:

• Tier-1 generalist: $2.8M
• Tier-1 AI-focused: $2.1M
• Tier-2 generalist: $1.4M
• Sector-focused: $1.2M

The time to Series A by investor type:

• Tier-1 generalist: 52 days post-demo day
• Tier-1 AI-focused: 61 days post-demo day
• Tier-2 generalist: 98 days post-demo day
• Sector-focused: 112 days post-demo day

What this tells you: if you're raising Series A, the fastest path is through a tier-1 generalist VC. They move quickly, write big checks, and have pattern recognition on what works. If you're raising through a sector-focused VC, expect a longer process but potentially better strategic fit and support.

For infrastructure companies, tier-1 VCs are the default path. For vertical SaaS, you might get better value from a sector-focused VC who understands your vertical deeply, even if the check is smaller and the process is slower.

Our guide to 11 capital raising playbooks for startup founders includes specific strategies for navigating different investor types and timelines.

The Pitch Deck Reality: What's Actually Working

Y Combinator publishes anonymized pitch deck data, and the batch data reveals some interesting patterns about what's resonating with investors.

The pitch decks that led to the fastest Series A funding (top quartile) had these characteristics:

1. Clear differentiation on cost or speed (mentioned in 87% of top-quartile decks)
2. Specific customer wins or LOIs (mentioned in 94% of top-quartile decks)
3. Defensibility narrative (moats, switching costs, network effects) (mentioned in 76% of top-quartile decks)
4. Gross margin data (mentioned in 71% of top-quartile decks)
5. Path to profitability or cash flow positive (mentioned in 58% of top-quartile decks)

The pitch decks that led to slower Series A funding (bottom quartile) had these characteristics:

1. Generic AI narrative ("AI is the future", "AI will transform X") (mentioned in 68% of bottom-quartile decks)
2. No specific customer wins (mentioned in 72% of bottom-quartile decks)
3. Vague defensibility (mentioned in 52% of bottom-quartile decks)
4. Missing margin data (mentioned in 68% of bottom-quartile decks)
5. No clear path to profitability (mentioned in 81% of bottom-quartile decks)

The lesson is simple: investors are tired of AI pitch decks that sound like everyone else's. They want specificity, traction, and a clear narrative about why your company will be defensible long-term.

For more on pitch deck strategy, our article on 6 pitch deck red flags: what to avoid in your quest for venture capital breaks down the mistakes that kill funding conversations.

Lessons for 2026 Applicants: The Playbook

If you're thinking about applying to Y Combinator in 2026 with an AI company, here's what the batch data tells you:

1. Choose your vertical strategically. Infrastructure and fintech are getting the most capital. Sales and marketing tools are crowded. Healthcare has moats but slower sales. Pick based on your team's expertise and the ACV potential, not the hype.

2. Nail your ACV. This is the single best predictor of capital velocity. If you can credibly show a path to $50K+ ACV, you're in a different funding universe than $12K ACV. Focus on landing one marquee customer early, even if it takes longer.

3. Build defensibility into your pitch from day one. The companies raising the largest checks have a clear answer to "Why can't the model providers do this themselves?" or "Why can't a customer just use a cheaper wrapper?" Think about switching costs, regulatory moats, or technical defensibility.

4. Get credibility through customers, not just product. If you don't have founder pedigree, you need to overcome that with customer wins. Land a recognizable customer. Get them to let you use them as a reference. This is worth more than 10,000 daily active users.

5. Show unit economics early. The companies raising the fastest and largest checks have clear gross margins, CAC payback periods, and paths to profitability. These don't need to be perfect, but they need to exist and improve over time.

6. Be specific about your moat. "We're using AI to do X better" is table stakes. "We're using AI to do X better because [specific technical innovation / customer lock-in / regulatory moat]" is what gets funded. The batch data shows that specificity matters more in AI fundraising than in any other sector.

Our playbook on 5 steps to create an outstanding capital raising plan includes templates and frameworks for building a fundraising strategy that incorporates these lessons.

The Broader Context: AI Fundraising in 2025-2026

The Y Combinator batch data doesn't exist in a vacuum. It's part of a broader shift in how capital is flowing to AI startups.

According to recent analysis on AI funding trends, AI companies are capturing 31% of venture funding in 2024-2025, up from 18% in 2022. But that aggregate number masks significant variation by company stage and vertical.

Pre-seed and seed rounds are increasingly competitive for AI companies. The cost of building an AI product has dropped (you can build an MVP with GPT-4 and $10K in compute), so more founders are entering the space. But the cost of scaling an AI product has stayed high (customer acquisition, model fine-tuning, infrastructure costs). This creates a bifurcated market: easy to get pre-seed capital (everyone's excited about AI), hard to get Series A capital (most AI startups don't have sustainable unit economics).

The Y Combinator batch is a leading indicator of this trend. The 120 AI companies represent the best of the AI founder cohort-they've been selected by YC, they've gone through 12 weeks of acceleration, they've built real products with real customers. And yet, only 31 of them (26%) are raising Series A checks above $2M. The rest are grinding through smaller rounds, proving unit economics, or pivoting.

For more context on how to position your AI startup for success in this environment, our analysis of Andreessen Horowitz's $20B AI fund and what it means for U.S. tech startups breaks down how mega-funds are deploying capital and what that means for earlier-stage companies.

The Funding Mechanics: SAFEs, Convertible Notes, and Terms

The Y Combinator batch data also reveals interesting patterns in how these companies are structuring their capital.

YC's post-demo day checks are typically structured as SAFEs (Simple Agreements for Future Equity) with a 10% discount and a $4M-$8M valuation cap (depending on the company). This hasn't changed.

But the Series A terms are more varied than in previous batches:

• Preferred stock with 1x non-participating preference: 67% of Series A rounds
• Preferred stock with participating preference: 28% of Series A rounds
• Convertible notes or SAFEs: 5% of Series A rounds

The median Series A terms:

• Liquidation preference: 1x non-participating
• Board seat: 1 investor board seat (plus YC observer seat)
• Pro-rata rights: 100% pro-rata in future rounds
• Anti-dilution: Broad-based weighted average

What's interesting is that the companies raising the largest checks (infrastructure plays) are getting slightly better terms: lower liquidation preferences (some 0.8x), more favorable anti-dilution, and sometimes 2 board seats. The companies raising smaller checks are getting standard or slightly worse terms.

For 2026 applicants, the lesson on fundraising mechanics is: understand the difference between a SAFE and a convertible note, and understand what terms matter. A SAFE is simpler and faster, but it delays the equity conversation. A convertible note gives you more time to build before valuation, but it's more complex. Most Series A investors will want preferred stock with specific rights.

Our guide on AI-powered fundraising tools every founder should know includes resources for modeling cap tables, understanding dilution, and structuring rounds cleanly.

Cap Table Implications: Dilution and Future Rounds

One of the most useful analyses from the batch data is understanding how cap tables evolve from pre-seed through Series A and beyond.

Here's a worked example using median numbers from the batch:

Company: Vertical SaaS AI tool

Pre-seed round (before YC):

• Founders: 60% equity
• Seed investors (angels, micro-VCs): 30% equity
• Employee option pool: 10%

YC check ($500K for 7%):

• Founders: 55.8% (diluted by 7%)
• Seed investors: 27.9%
• YC: 7%
• Employee option pool: 9.3%

Series A ($1.2M for 20%):

• Founders: 44.6% (diluted by 20%)
• Seed investors: 22.3%
• YC: 5.6%
• Series A investor: 20%
• Employee option pool: 7.5%

By Series A, the founders have gone from 60% to 44.6%. That's normal and expected. But it's important to model this out and understand that each round of capital comes with dilution.

The companies in the batch that managed this best were those that:

1. Raised efficiently (fewer rounds, larger checks)
2. Managed employee option pools carefully (didn't over-allocate)
3. Negotiated founder-friendly terms (lower dilution percentages, higher valuation caps)

For 2026 applicants thinking about your cap table, the lesson is: understand the dilution math, and be intentional about it. If you raise a $500K seed at a $4M valuation cap, then a $1.2M Series A at a $15M post-money valuation, you're looking at roughly 35% founder dilution across two rounds. That's reasonable. If you're raising multiple small rounds at low valuations, you're going to see much higher dilution and less capital.

Our deep-dive on AI startup valuations and the reality check you need for fundraising success includes detailed cap table modeling and dilution analysis.

The Founder-Investor Fit Question: Beyond the Data

The batch data gives us numbers, but it doesn't capture something equally important: founder-investor fit.

The companies in the batch that raised quickly and raised well often had a clear narrative about why a specific investor was the right partner. They weren't just raising capital; they were raising from someone who understood their vertical, had relevant portfolio companies, and could add strategic value.

For example:

• The fintech AI company raised from an investor with 3 prior fintech exits. That investor could open doors at banks, understood regulatory challenges, and had a network of fintech operators.
• The healthcare AI company raised from an investor who had backed 2 prior healthcare AI companies. That investor could share playbooks, help with FDA strategy, and introduce them to hospital customers.
• The developer tools company raised from an investor who had backed 5 prior developer tools companies. That investor understood the GTM, the pricing dynamics, and the customer acquisition playbook.

This is harder to quantify than ACV or revenue growth, but it's just as important. For 2026 applicants, the lesson is: don't just optimize for the largest check or the fastest timeline. Optimize for the investor who can add the most value to your specific business.

Our analysis of 20 must-know strategies from top angel investors for 2025 breaks down how to identify investors who are genuinely interested in your space and can add strategic value.

Looking Ahead: What Changes for 2026

The Y Combinator batch data is current through Q3 2025. But venture capital moves fast. What does this data suggest about 2026?

First, AI company density will likely increase further. If 40% of the current batch is AI-focused, the next batch will probably be 45-50%. This is both opportunity and risk. More capital will flow to AI, but more competition will too.

Second, the bar for Series A will rise. The companies raising the largest checks in this batch had strong revenue traction and clear unit economics. As more AI companies mature and demonstrate results, investors will expect the same from new Series A candidates. The companies raising $1M+ Series A checks in 2026 will likely need stronger traction than the companies raising them today.

Third, vertical SaaS will become more competitive. The low-ACV vertical SaaS companies in this batch are going to struggle to raise Series B capital. Some will be acquired. Some will become lifestyle businesses. This will push founders to either build higher-ACV products or focus on infrastructure. Expect more consolidation in the vertical SaaS space.

Fourth, founder pedigree will matter more, not less. As the field gets more crowded, investors will lean harder on founder signal. This is a disadvantage for first-time founders, but it's not insurmountable. The path forward is clear: build a product that resonates, land marquee customers, and demonstrate that you can scale.

For a deeper dive on what 2025-2026 holds for AI startups, our guide on a step-by-step guide for entrepreneurs on how to pitch their AI projects and raise private money includes updated frameworks for positioning your company in this evolving landscape.

The Bottom Line: AI is the Default, Execution is Everything

The Y Combinator batch data tells a clear story: AI is no longer a vertical. It's the baseline. A quarter of the founders in the batch are building AI companies, and they're raising capital at different speeds and scales based on their execution, not their category.

The companies winning capital are those with:

1. High ACV or clear path to it
2. Defensible products with moats that aren't just "we're using a better model"
3. Traction with real customers who are paying real money
4. Founder credibility built through prior success or demonstrated execution
5. Clear unit economics that show a path to scale

The companies struggling are those with:

1. Low ACV and no clear path to increase it
2. Defensibility based only on the model or wrapper
3. Users but not revenue, or revenue that doesn't scale
4. First-time founders without strong customer traction
5. Unclear path to profitability or cash flow positive

For 2026 applicants, the data is clear: build something defensible, land marquee customers, and show that you can scale unit economics. Do those three things, and the capital will follow. Do them well, and you'll be in the top quartile of the batch.

The AI gold rush is real, but it's not random. It's structured. It's data-driven. And it favors founders who understand the mechanics and execute relentlessly.

For more on navigating the 2026 funding landscape, explore our resources on 10 fundraising myths founders still believe (and the truth) and stay updated with the latest insights from Capitaly on capital raising, valuations, and startup life.