Y Combinator Demo Day is the public thermometer of which sectors attract top-tier capital. The 2024 batches (W24, S24) show clear patterns: vertical AI dominates (~60% of startups), defence tech is growing, climate is holding steady, and consumer is recovering. This article analyses trends for European founders and investors, with honesty about what the hype obscures.

Key takeaways

  • Vertical AI went from representing ~15% of pre-ChatGPT batches to dominating at ~55-60%: the fastest compositional shift in YC’s history.

  • The YC differential remains quality filtering and the network: the alumni community is the hardest asset to replicate.

  • For European founders: YC is open to non-Americans; Delaware C-Corp incorporation and dual EU+US HQ are the standard pattern.

  • Metrics that matter to investors: 15-30% MoM revenue growth at seed, ARR, cohort retention, and a moat beyond the LLM wrapper.

  • Post-YC seed valuations rose after the 2022-2023 "winter": median ~$15-20M; hot AI companies at $50-100M.

2024 batch composition

Approximate W24/S24 distribution:

  • Vertical AI/ML: ~55%.

  • B2B SaaS: ~15%.

  • Fintech: ~10%.

  • Climate: ~8%.

  • Defence / dual-use: ~5%.

  • Healthcare: ~5%.

  • Other: ~2%.

The massive shift toward AI versus pre-ChatGPT batches (where AI was ~15%) is the most important structural change.

Vertical AI: the dominant

AI for specific professions

Products with deep specialisation in a vertical:

  • Legal: contract review, automated discovery.

  • Sales: SDR automation, RevOps.

  • Medicine: diagnostic support, clinical documentation.

  • Accounting: automated bookkeeping.

  • Engineering: CAD, simulations.

Pattern: domain expert + AI capabilities. AI without domain is a commodity; domain without AI is slow.

AI infrastructure

Tools to build proprietary AI systems:

  • Vector databases.

  • LLM routers and proxies (see LLM proxies with LiteLLM for the market state).

  • Fine-tuning platforms.

  • Evaluation frameworks.

These compete with LangChain, LlamaIndex, and established players. Sustainable competitive advantage requires more than API wrapping.

AI agents

Companies building autonomous agents:

  • Customer support.

  • Sales outreach.

  • Code reviewers.

  • Business process automation.

The hype is real but execution varies enormously. See CrewAI: agent teams for the real state of the multi-agent pattern.

Defence and dual-use

A sector growing since the Ukraine invasion. YC, which previously explicitly avoided the sector, now includes it explicitly (its first defense investment, Ares Industries, was announced in August 2024):

  • Drones and autonomous systems.

  • Satellite imagery analysis.

  • Government cybersecurity.

  • Defence manufacturing.

For European founders, this sector has specific regulatory implications (see EU AI Act for the high-risk systems framework).

Metrics that attract capital

YC investors look for:

  • Revenue growth: 15-30% MoM at seed is the ideal.

  • ARR: $100K-1M 6-12 months post-YC.

  • Retention: DAU/MAU, cohort retention.

  • LTV/CAC: improving direction.

  • AI moat: proprietary data, user-generated data flywheel.

Just having a "revolutionary idea" is no longer sufficient. The market demands metrics from day one.

Founder patterns

Common traits in 2024 batches:

  • Domain expert + technical: mixed teams with real vertical experience.

  • AI-specialty PhDs: increasing versus previous years.

  • Second-time founders: more frequent and better valued.

  • Age distribution: primarily 25-35.

  • Geographic: 70%+ US, but with notable international growth.

Lessons for European founders

Observations

  • The EU-US funding gap persists in AI, though it is narrowing.

  • The YC model (cohort + demo day) inspires European accelerators: Seedcamp, Antler, Techstars.

  • Remote-first startups are more common; physical US presence is not as mandatory as before.

  • European founders in YC: some batches have 10-15% non-Americans.

Practical tactics

  • Apply to YC: the programme is open to non-Americans.

  • Delaware C-Corp incorporation: the standard for raising US capital.

  • Dual HQ: some startups maintain EU operations + US incorporation.

  • Raise US: US funds pay higher valuations, but require US presence or nexus.

European alternatives

Similar-model accelerators in EU:

  • Seedcamp[1]: London.

  • Antler[2]: global + EU.

  • Techstars[3]: multiple European programmes.

  • Y Combinator Reach: remote option for non-US founders.

For EU founders who cannot or do not want to relocate: valid, but with a smaller network.

Criticisms and considerations

The "YC bubble"

  • Many startups are "thin wrappers" on GPT-4 without real differentiation.

  • Valuations possibly reflect cycle optimism more than fundamental value.

  • Focus on fast exit may incentivise shortcuts on complex engineering.

The counter-argument

  • Abundant capital seeks deployment in the best opportunities.

  • Vertical AI first-movers capture markets with genuine network effects.

  • YC’s filter selects real quality. Failory’s analysis of YC’s full track record puts recent failure at ~13%, well below the ~90% typical of early-stage venture. Older cohorts rise to ~40%, since they have had more time to fail. The upside is also larger.

The market will decide. The real signal will come in 2-3 years when retention rates and sustainable Series A rounds become visible.

Historical post-YC predictions

Aggregate data on the full history of YC alumni (all cohorts, not just 2024), per Failory’s analysis of YC startups[4]:

  • ~87% of YC startups are still operating: a recent failure rate of ~13%, well below the ~90% typical of early-stage venture. In the first 17 cohorts, which have had more years to fail, inactivity rises to ~40%.

  • ~4-5.5% of YC startups become unicorns (a $1B+ valuation), versus ~2.5% across venture-backed seed-stage startups as a whole.

  • YC has produced 82 unicorns and 17 IPOs, with over $600B in combined valuation across its alumni.

  • Few reach the level of Airbnb or Stripe.

Return concentration, with few winners defining fund performance, is the rule and not the exception.

Conclusion

YC Demo Day 2024 reflects the current state of tech: AI dominant, capital abundant for the best ideas, valuations elevated post-winter. For founders, the lessons are clear: choose a vertical with real depth, build a moat beyond the LLM wrapper, and measure from day one. For European founders, YC remains the gold-standard of training: worth applying. For investors, the batch is a signal but demands discernment: not all that glitters is a unicorn, and the market will take time to separate solid projects from those merely surfing the cycle.

English version of this article: also read it in Spanish.

Sources:

  1. Y Combinator: official Demo Day[5]
  2. Crunchbase News: 50% of YC’s W24 batch is built with AI[6]
  3. TechCrunch: YC backs its first defense startup, Ares Industries[7]
  4. Failory: Y Combinator startup failure and survival analysis[4]

Frequently asked questions

Can a European founder get into Y Combinator without relocating to the US?

Yes, the programme is open to non-Americans and some batches have 10-15% non-US founders; remote-first startups are more common and physical US presence is no longer as mandatory. The usual pattern is incorporating as a Delaware C-Corp and keeping a dual HQ, with EU operations and US incorporation. For those who do not want to relocate there is Y Combinator Reach, a valid remote option, though its network is smaller. Raising from US funds, however, usually requires US presence or nexus.

Which metrics do investors expect from a startup after YC?

Revenue growth of 15-30% month over month at seed, ARR of $100K-1M within 6-12 months post-YC, retention measured by DAU/MAU and by cohort, and an improving LTV/CAC. In AI, add a real moat: proprietary data or a user-generated data flywheel, beyond the wrapper on an LLM. Just having a revolutionary idea is no longer enough. Post-YC seed valuations sit around a median of $15-20M, with hot AI companies reaching $50-100M.

What share of YC startups fail or become unicorns?

Per Failory's analysis of the full alumni track record, about 87% of YC startups are still operating. The recent failure rate is ~13%, against the ~90% typical of early-stage venture, and it rises to ~40% in the first 17 cohorts. Between 4% and 5.5% reach unicorn valuation ($1B or more), versus ~2.5% across venture-backed seed-stage startups. YC totals 82 unicorns, 17 IPOs and over $600B in combined valuation.

Sources

  1. Seedcamp
  2. Antler
  3. Techstars
  4. Failory’s analysis of YC startups
  5. Y Combinator: official Demo Day
  6. Crunchbase News: 50% of YC’s W24 batch is built with AI
  7. TechCrunch: YC backs its first defense startup, Ares Industries