# Cp · Capital — element 56 of 58

> Fuel, when you want it. Turns a working wedge into a funded expansion.

- **Group:** 12 · Human Elements
- **Necessity:** Optional
- **Price band:** Free
- **Maturity:** Stable
- **Edition:** v2026.Q3 · verified 2026-09-13

## Leading tools (v2026.Q3)

- **Customers** — revenue: zero dilution
- **Accelerators** — brand, network, 7%
- **Pre-seed & seed rounds (angels + AI funds)** — the ~20% dilution round
- **Revenue-based financing & venture debt** — cash against revenue
- **Equity crowdfunding (Reg CF)** — your users as investors

## Our take

The 2026 twist: default-alive is achievable solo, so capital is a choice, not a requirement. Choose deliberately.

## Combines with

Eq, Mn


## The top 5 — deep dossier (verified 2026-09-13)

Customers, if you can get them — in 2026 revenue is the only capital with zero dilution, no dilution ceiling, and no permission required, and it is now the gate to every other route on this list (Founderpath wants $10K MRR; venture debt wants $1M ARR; YC converts applicants with revenue at 3–5x the pre-product rate). Accelerators win when the network and the forcing function are worth 7% — YC's $500K/7% is still the benchmark, but Neo's $750K uncapped SAFE (0.75% dilution at a $100M round, Feb 2026) is now the founder-friendliest headline terms in the category. A priced seed round wins when you need $3M+ to buy compute or a team, and you accept ~20% dilution to get it. Non-dilutive debt — Founderpath, Lighter Capital, venture debt at 8–15% — wins when you have predictable recurring revenue and refuse to sell equity. Equity crowdfunding wins last and rarely: 7.9–9% in platform fees for a median raise around $350K, and it only works if you already have a crowd.

1. **Customers (revenue as capital)** (your users — via pre-sales, deposits, design-partner contracts, annual-prepay) — 0% dilution · 0% APR · cost is time and the discount you give early customers (annual prepay typically 10–20% off). Best for: Software with a working wedge and a buyer who feels the pain now — the default route for AI-led startups small enough that agents replace headcount. Why: The 2026 math changed: capital concentrated so hard that non-frontier startups can't count on it — OpenAI and Anthropic alone took $217B, or 43% of all H1 2026 startup funding (Crunchbase). Global Q2 seed was $12B, but it barbelled: $2.8B went into seed rounds of $100M+ while $5B went into rounds of $10M and under. Revenue is also the key to every other door here — Founderpath's floor is $10K MRR, Lighter's is $200K ARR, venture debt's is roughly $1M ARR plus a Series A. Watch: Slowest route by far, and it caps what you can attempt — you cannot pre-sell your way into a frontier training run or a two-year research bet. Customer money comes with real strings: annual prepay discounts cut effective price, and design-partner contracts can quietly turn your roadmap into their roadmap. Concentration risk is the failure mode nobody prices. [https://www.paulgraham.com/aord.html](https://www.paulgraham.com/aord.html)
2. **Accelerators** (Y Combinator · Neo · a16z Speedrun · HF0 · South Park Commons · EF · Antler) — YC $500K = $125K for 7% post-money + $375K uncapped MFN · Neo $750K uncapped SAFE, up to 5% in next round (0.75% at a $100M round) · a16z Speedrun up to $1M + $8M+ credits · HF0 $1M uncapped for 5% or $500K + 3% · SPC $400K for 7% + $600K guaranteed next round · EF US $250K for 8% · Antler US $250K for 9.09%. Best for: A working wedge that needs a network, a deadline, and a brand stamp more than it needs the money — and founders who can absorb 5–9% dilution for it. Why: YC's terms have not moved since January 11, 2022 — $125K for a fixed 7% plus $375K uncapped MFN — and its scarcity is the product: roughly 1.5–2% acceptance from 12,000–15,000 applications per batch. But 2026 broke the price cartel. Neo launched $750K on an uncapped SAFE in February 2026, capping participation at 5% of the next round — meaning a founder who raises at $100M gives up 0.75%, roughly a tenth of YC's take. HF0 stays deliberately tiny at 10 teams per batch and its S25 cohort averaged $82M post-demo-day valuations with a top team at $20M annualized revenue. Watch: The 7% is real money — at a $100M next round that is $7M for a 12-week program you may not need. YC's own edge is now heavily brand and batch network rather than terms, and Neo/Speedrun/HF0 all beat it on dilution. Selectivity cuts both ways: a rejection consumes months of founder attention, and the batch calendar imposes a demo-day narrative that can push companies to raise before they should. Antler's US terms ($250K/9.09%) are the most expensive on the list and date from an October 2023 announcement — confirm they still stand before applying. [https://www.ycombinator.com/apply](https://www.ycombinator.com/apply)
3. **Pre-seed & seed rounds (angels + AI funds)** (seed funds, solo GPs, angel syndicates — priced rounds and post-money SAFEs) — ~20% founder dilution at seed (Carta, 2026) · median seed post-money $24M, ~$16M pre, $3.1M median round · AI seed valuations run ~40%+ above non-AI peers. Best for: Teams that need $2–5M at once — GPU spend, a founding engineering team, or an 18–24 month runway to $1M ARR — and are willing to sell a fifth of the company for it. Why: This is still the largest single route by dollars and the only one that funds a company that isn't yet default-alive. The instrument standardized around YC's post-money SAFE (uncapped, capped, or discount variants — plus Canada, Cayman and Singapore versions), which makes dilution computable on day one. The AI premium is real and quantified: Carta puts seed post-money median at $24M in 2026, with AI companies raising roughly 1.3x larger rounds at valuations ~40%+ above non-AI peers, and AI foundational-model Series A medians near $300M against ~$55M for non-AI. Watch: The premium is not evenly distributed — it accrues to the frontier and to credentialed teams. Application-layer AI startups increasingly get priced as ordinary software, and the $5B of Q2 seed that went into rounds of $10M-and-under is where most founders actually live. ~20% dilution at seed compounds: a seed plus a Series A puts most founders under 50% before the company is proven. And post-money SAFEs stack — three uncapped notes and a priced round can dilute far past what any single term sheet implied. [https://www.ycombinator.com/documents](https://www.ycombinator.com/documents)
4. **Revenue-based financing & venture debt** (Founderpath · Lighter Capital · Stripe Capital · Hercules / TriplePoint (venture debt)) — Founderpath: 7% flat discount on revenue financing (a $200K advance repays $214K), term loans from 15%, MCA from 5% of monthly revenue, no closing or prepayment fees · Lighter Capital: up to $10M, rates undisclosed · Venture debt: 8–15% APR (SOFR ~4.5% + 6–9% spread) plus 1–5% warrant coverage. Best for: Predictable recurring revenue you want to convert into cash without selling equity — funding a sales hire, bridging a seasonal gap, or extending runway between priced rounds. Why: The only route that buys you money without buying your company. Founderpath has deployed $271M across 742 companies since 2021 at an average deal near $600–750K, funds in 24 hours to 2 weeks off a data connection, and takes no equity, board seat, warrants, or personal guarantee. Lighter Capital goes to $10M with a $200K ARR floor across US, Canada and Australia. Venture debt scales further — typically 25–35% of your last equity round, over 1–4 years with 6–12 months interest-only — but only after a Series A. Watch: The category shrank hard. Pipe wound down its direct SaaS revenue-trading marketplace back in early 2024 and is now embedded Capital-as-a-Service delivered only through partners like UberEats, GoCardless, Boulevard and Housecall Pro; it cut roughly 50% of staff in November 2025 and changed CEO in December. Capchase left too — its $200M+ raise in May 2026 funds B2B buy-now-pay-later vendor financing, not founder RBF. What remains is smaller and stricter. And a flat 7% discount is not 7% APR: repaid over 12 months it is roughly a low-teens effective annual rate, and worse on short terms. Venture debt adds warrants (so it is not truly non-dilutive) plus covenants that can trigger exactly when you're weakest. [https://founderpath.com](https://founderpath.com)
5. **Equity crowdfunding (Reg CF)** (Wefunder · Republic · StartEngine) — Wefunder 7.9% of a successful Reg CF raise + $1,000/yr admin, no upfront fee · Republic 7% cash + 2% securities = 9% · Reg CF cap $5M per 12 months · plus real legal, audit and marketing costs on top. Best for: Consumer products and community-native companies whose users would genuinely want to own a piece — and founders who want capital without a lead investor's governance. Why: It is the only route on this list where your customers become your cap table, and it is the only one with no gatekeeper: no partner meeting, no acceptance rate, no Series A prerequisite. Wefunder is the scale player at $826M in cumulative volume across ~3.29M investors, with roughly 75% of campaigns hitting their minimum and a typical raise near $350K. The $5M annual Reg CF ceiling is enough to be a real pre-seed or bridge. Watch: Ranked last for good reason. The 7.9–9% platform fee is a genuine cost of capital, and it is only 30–40% of the true expense once legal, audited financials and campaign marketing are counted. Volumes are tiny and consolidating — Republic did $15.6M in US Reg CF in 2024, about 6% of Wefunder's $99.4M, and has since pivoted toward accredited-only deal rooms, SPVs and tokenization after buying INX for $60M in April 2025. A messy cap table of hundreds of small holders can complicate later priced rounds, and B2B AI startups rarely have the crowd that makes this work at all. [https://wefunder.com](https://wefunder.com)

### How to choose
- If You can reach $10K MRR before you need money → Do that first, then take non-dilutive capital. $10K MRR is Founderpath's floor and $200K ARR is Lighter's; both take 0% equity, no board seat, no warrants. Every dollar of revenue makes every other route on this list cheaper.
- If You want an accelerator and you're optimizing for dilution rather than brand → Apply to Neo before YC. Neo's $750K uncapped SAFE caps participation at 5% of the next round — 0.75% if you raise at $100M — against YC's fixed 7%. Take YC anyway if the batch network and the stamp are the actual product you're buying.
- If You need $3M+ at once for compute or a founding team, and you are not default-alive → Raise a priced seed and budget ~20% dilution (Carta 2026 median: $24M post, $3.1M round). Do not stack three uncapped SAFEs hoping to avoid the conversation — post-money notes compound and the reckoning arrives at the priced round.
- If You are an application-layer AI startup and everyone tells you funding is easy right now → Assume it isn't for you. AI took >70% of Q2 2026 capital but OpenAI and Anthropic alone took 43% of the H1 total; the $5B of Q2 seed that landed in rounds of $10M-and-under is the real market. Plan to be default-alive.
- If Someone pitches you revenue-based financing at 'only 7%' → Convert it to an annual rate before signing. A 7% flat discount repaid over 12 months is roughly a low-teens effective APR and materially worse on shorter terms — still often cheaper than 20% equity, but not the number on the tin.
- If You're a B2B AI startup considering equity crowdfunding → Skip it. Reg CF works when you already have a crowd; 7.9–9% in platform fees plus legal and audit against a ~$350K median raise is a bad trade without one. Revisit only if your users would genuinely want to own the thing.

### The field (25 more)

Y Combinator, Neo Residency, a16z Speedrun, HF0, South Park Commons, Entrepreneur First, Antler, Neo Scholars, Techstars (fading), Founderpath, Lighter Capital, Stripe Capital, Pipe (renamed), Capchase (renamed), Venture debt (Hercules, TriplePoint, Bridge Bank), Mercury, Silicon Valley Bank (acquired), Wefunder, Republic, StartEngine, AngelList, YC post-money SAFE, Carta Data Desk, Antler Elevate, Compute credits programs

Full dossier data: https://elems.ai/e/cp.json

---
Source: [elems.ai](https://elems.ai/e/cp.html) — the periodic table of the AI-led startup. Data: https://elems.ai/elements.json (CC BY 4.0, cite elems.ai).