Startups.
From Paul Graham's garage to the unicorn parade — what venture capital actually looks for, why most fail, and the few patterns that recur.
What is a startup?
"A startup is a company designed to grow fast." — Paul Graham, "Startup = Growth," 2012
Not just new. A laundromat is new but doesn't scale. Three traits define it:
- Big market. Total addressable market in the billions, not the millions.
- Repeatable model. Each unit sold makes future units cheaper or easier.
- Speed before scale. Find the model first; pour fuel only when it works.
The accelerator model
Founded 2005 by Paul Graham, Jessica Livingston, Robert Morris, Trevor Blackwell. Twice-yearly batch of ~250 startups in Mountain View / online. Investment: $500K for ~7% (current terms).
Demo Day at the end. ~5,000 founders have come through. Aggregate value of YC alums exceeds $600B as of 2024.
Notable alumni
Airbnb (W09) · Stripe (S09) · Dropbox (S07) · Reddit (S05) · Coinbase (S12) · Instacart (S12) · DoorDash (S13) · Twitch (W07)
YC Maxims
- Make something people want.
- Talk to users.
- Do things that don't scale.
- Launch ugly. Launch now.
- Default alive, not default dead.
- Default rate: ~5% weekly growth.
Lean Startup
Eric Ries (2011), based on Steve Blank's customer development. Replace business plans with hypotheses. Replace launches with experiments.
Pivot = a structured course correction to test a new fundamental hypothesis. Slack pivoted from a multiplayer game (Glitch) to a chat tool. Twitter pivoted from podcasting (Odeo). Instagram pivoted from a check-in app called Burbn.
Product–Market Fit
"You can always feel product/market fit when it's happening. The customers are buying the product just as fast as you can make it." — Marc Andreessen
Sean Ellis Test
Survey users: "How would you feel if you could no longer use this product?" PMF benchmark: 40%+ "very disappointed."
Retention curves
The cohort retention curve must flatten above zero. Users keep coming back without paid prompting.
Word of mouth
Customers refer without being asked. The CAC drops. The sales cycle shortens. You can't make the product fast enough.
The funding ladder
| Stage | Typical Size | Lead | Dilution | What you sell |
|---|---|---|---|---|
| Pre-seed | $50k–$1M | Friends, angels, YC | 5–10% | A pitch deck and a working demo |
| Seed | $1–4M | Seed funds (FF, Initialized) | 15–25% | Early traction, $1M ARR target |
| Series A | $8–18M | Sequoia, A16z, Benchmark | 20–25% | PMF, repeatable sales motion |
| Series B | $25–50M | Tier-1 funds, growth arms | 15–20% | Scale, $5–15M ARR, expanding ICP |
| Series C+ | $50–500M+ | Tiger, Coatue, SoftBank | 10–15% | Market leadership, IPO proximity |
| Pre-IPO | $500M–$5B | Crossover, sovereign wealth | 5–10% | Brand, IPO bookbuild |
Rule of thumb: raise 18–24 months of runway. The next round always takes longer than you expect.
The unit economics that matter
Customer Acquisition Cost. All sales+marketing ÷ new customers in the period.
Lifetime Value. ARPU × gross margin ÷ churn rate. SaaS rule: LTV/CAC ≥ 3.
Payback Period. Months for gross profit to cover CAC. Healthy: < 12 months.
Net Revenue Retention. Top SaaS: 120%+. Existing customers grow faster than churn.
SaaS Rule of 40: Growth Rate % + Operating Margin % ≥ 40. The market will value you on this above all else once you cross $50M ARR.
The founding team
YC's data: solo founders fail more often than 2- or 3-person teams. 4+ team members raises coordination costs and reduces equity per founder.
The classic split: hacker × hustler. One ships product. One ships sales, fundraising, hiring. Sometimes a third "designer/product" sits between.
"Bad cofounders are a worse problem than not having a cofounder." — Paul Graham
Equity vesting
Standard: 4-year vest, 1-year cliff. After year 1, 25% vests; remainder monthly. Reverse-vesting protects everyone if a cofounder leaves in month 8.
Founder breakups
Noam Wasserman's data: 65% of high-potential startups fail because of founder conflict. Have the equity-split conversation early; revisit it.
Airbnb · 2007–2011
Brian Chesky and Joe Gebbia, broke roommates in San Francisco, rent out three air mattresses during a design conference. Charge $80/night. Three guests. The seed of an idea.
Original site: AirBedAndBreakfast.com. They financed runway by selling election-themed cereal boxes ("Obama O's") at $40/box — made $30K.
Rejected by VCs 7 times. Accepted to YC W09 with PG's note: "They're like cockroaches. They will not die."
The unlock: photographers. NYC hosts had bad photos. The founders flew there with a camera, shot listings themselves. Bookings doubled. Lesson: do things that don't scale.
By the numbers, today
$11B revenue (2024) · 5M+ hosts · 1.5B+ guest arrivals all-time · IPO Dec 2020 at $47B valuation, popped to $100B day one.
Stripe · 2010–today
Patrick & John Collison, brothers from Limerick, Ireland. Built a way to accept payments online with seven lines of code at a time when the alternatives required signing a contract with a "merchant acquirer" and waiting weeks.
Beachhead
Developers. Hard to charm, but if you do — they evangelize. Stripe's docs are still considered industry-best.
Expansion
From payments to billing, identity (Atlas, Radar), corporate cards (Issuing), tax. The "atomic concept" expanded into a financial OS.
Numbers
$1.4T processed in 2024 · ~$70B valuation · still private · profitable.
"We thought we were building a developer tool. Turns out we were building economic infrastructure." — Patrick Collison, 2021
Quibi · 2018–2020
Jeffrey Katzenberg + Meg Whitman. Raised $1.75B from Disney, NBC, Sony, Alibaba, JPMorgan. Premium short-form video, mobile-only, $5–8/month. Star talent. Original shows.
Launched April 2020. Shut down October 2020. Six months.
What went wrong
- Wrong moment. Mobile-only during a pandemic when everyone was home in front of a TV.
- No social loop. Couldn't share clips. People watch short-form to send to friends.
- Wrong target. Hollywood made shows for adults; phone-shape video lived on TikTok.
- Capital ≠ traction. No PMF, but launched at scale anyway. Lean Startup commandment broken.
The pitch deck
Sequoia's classic 10-slide template (still 90% of seed pitches in 2026):
| # | Slide | Test |
|---|---|---|
| 1 | Company purpose | One sentence. Could a stranger repeat it? |
| 2 | Problem | Pain customers feel today. Quantify if you can. |
| 3 | Solution | Your product. Not features — the value. |
| 4 | Why now? | What changed in tech, regulation, behavior? |
| 5 | Market size | TAM / SAM / SOM. Bottoms-up beats top-down. |
| 6 | Competition | Don't say "no competitors." Say where you sit. |
| 7 | Product | Screenshots. Demo if possible. |
| 8 | Business model | How you make money. Pricing. Unit econ. |
| 9 | Team | Why you. Why now. Domain credentials. |
| 10 | Financials & ask | How much, for what milestone, runway months. |
Why startups die
CB Insights' top reasons (post-mortem analysis of ~500 failed startups):
| Reason | % |
|---|---|
| Ran out of cash / failed to raise | 38% |
| No market need | 35% |
| Got outcompeted | 20% |
| Flawed business model | 19% |
| Regulatory / legal | 18% |
| Pricing / cost issues | 15% |
| Wrong team | 14% |
| Bad product | 8% |
Default Alive vs. Default Dead
PG's 2015 essay. Plot expense growth vs. revenue growth at current burn. Will you reach profitability before cash runs out?
If yes — you can decide whether to raise. If no — your decision is forced. Most founders discover too late they're default dead.
"Until you're default alive, you have no permission to think about anything else." — PG
Term sheet, demystified
| Term | What it really means |
|---|---|
| Pre-money valuation | What the company is worth before this check lands |
| Post-money valuation | Pre-money + new investment |
| Liquidation preference | Investor gets their money out first on exit. 1x non-participating is standard. |
| Pro-rata rights | Right to invest in future rounds to maintain ownership % |
| Anti-dilution | Down-round protection. Broad-based weighted average is fairest. |
| Board composition | Who controls the company. 2 founders / 2 investors / 1 indep is balanced. |
| Option pool | Set aside for future hires. Larger pool = more dilution to founders, not investors. |
| SAFE / Convertible note | Defer valuation to next priced round. SAFE has no maturity date or interest. |
How startups end
Acqui-hire
$1–25M. Talent acquisition. Common when product fails but team is strong.
Strategic acquisition
$50M–$5B. Instagram→Meta ($1B), GitHub→Microsoft ($7.5B), Slack→Salesforce ($27B).
IPO
The big one. Median age at IPO: 11 years. Only ~1% of VC-backed companies make it.
Direct listing
Spotify (2018), Slack (2019). No new shares issued; just existing ones become tradeable.
SPAC merger
2020–21 fad. Faster than IPO but worse outcomes — most de-SPACs trade below issue.
Failure
~75% of seed-stage VC-backed startups never return investor capital. The base rate.
The startup funnel
The seven sins of seed-stage founders
- Premature scaling. Hiring sales before you know what you're selling. Building the team you'll have at $10M ARR while you're at $1M.
- Skipping user research. Believing your own pitch deck. Steve Blank: "No facts exist inside the building. Get out."
- Vanity metrics. Pageviews, downloads, signups. Track activation, retention, and revenue.
- Raising too much, too early. $10M seed at $40M post means your Series A needs to be a $200M valuation. Few companies make that jump.
- Co-founder splits without conversation. 50/50 sounds fair until someone's working twice as hard.
- Building features instead of solving problems. Customers want milkshakes, not blenders.
- Confusing fundraising with success. Money is fuel. Fuel without an engine is a fire.
Where to learn more
Books
- Zero to One — Peter Thiel
- The Lean Startup — Eric Ries
- The Hard Thing About Hard Things — Ben Horowitz
- Founders at Work — Jessica Livingston
- The Mom Test — Rob Fitzpatrick
- Hooked — Nir Eyal
- Crossing the Chasm — Geoffrey Moore
YouTube
- Y Combinator — Startup School lectures, How to Start a Startup (Stanford CS183B)
- Stanford GSB — View From The Top, ETL series
- a16z — Andreessen Horowitz channel
- This Week in Startups — Jason Calacanis interviews founders
Essays
paulgraham.com/articles.html — start with "How to Start a Startup," "Do Things That Don't Scale," "Default Alive or Default Dead?"
The only thing that matters.
"Make something people want." — Y Combinator motto, since 2005
Every framework, every term sheet, every essay reduces to that. The hard part is figuring out who the people are and what the something is. Everything else — the cap table, the demo day, the press, the IPO — is the byproduct.
Press → for next slide, ← for previous.
Thank you.
Twenty slides on what makes startups grow, raise, exit, or die.