# What Is a Cap Table? Ownership Math for AI Founders URL: https://aistartupinsights.com/journal/what-is-a-cap-table Type: blog Locale: en Published: 2026-10-26 Updated: 2026-10-28 --- > A cap table records who owns what. See how a seed round, option pool and SAFEs change founder ownership, with worked math and 2026 benchmarks. A cap table (capitalization table) is the ledger of who owns what in a company: every share, option, warrant and convertible, tied to a holder and a percentage. What is a cap table in practice? The document that decides how much of the company you still own after each raise. Carta's July 2026 data puts the median seed round at 18% dilution and the median Series A at 18% too. That headline number hides the pool, the SAFEs and the bridge notes sitting underneath it. ## What does a cap table actually contain? Six fields per line, and nothing more is required: the holder, the security type (common, preferred, option, warrant, SAFE, note), the share count, the ownership percentage, the price paid and the amount invested. The [LTSE guide to capitalization tables](https://ltse.com/insights/what-is-a-capitalization-table) lists the same core columns, and more detailed versions add voting rights by class. Two views matter, and mixing them up is the most common founder error: - **Outstanding shares:** only stock that has actually been issued. Founders, early employees who exercised, investors who already hold shares. - **Fully diluted shares:** outstanding plus every option granted, every option reserved in the pool, and every warrant. Investors price rounds on this count. Quote ownership on a fully diluted basis or do not quote it. A founder who says "I own 60%" on an outstanding basis may own 48% once the pool and converted SAFEs are in. The delta compounds at every round. ![Whiteboard grid of colored sticky notes arranged by investor and employee groups](https://fdzlnqpwsaniezitwiuw.supabase.co/storage/v1/object/public/cms-media/aistartupinsights/2026-10/fef264-i3.webp) ## How do the securities on a cap table differ? Not every line carries the same weight, and the order of the lines is part of the story. Five instruments show up on almost every AI startup cap table: - **Common stock:** held by founders and exercised employee options. Last in line at exit. - **Preferred stock:** issued to investors in a priced round. Carries a liquidation preference and, often, protective provisions. - **Options:** the right to buy common at a fixed strike price once vested. They count toward fully diluted shares from the day they are reserved. - **SAFEs and convertible notes:** cash in, shares later. They do not appear as shares until the priced round converts them, which is exactly why they get forgotten. The mechanics are covered in [what is a SAFE note](/journal/what-is-a-safe-note). - **Warrants:** rights to buy shares, typically attached to venture debt or a bridge. Skip the habit of tracking SAFEs in a separate tab "until they convert." Model them as shares from the day the money lands. Your next round negotiation starts from the converted view, not the tidy one. ## How does a priced round change the cap table? A round does three things at once: it adds new preferred shares, it usually enlarges the option pool, and it converts any outstanding SAFEs or notes. The last two are where founders lose track. Take a concrete case. Two founders hold 4,000,000 shares each, 8,000,000 in total, with no pool yet. They raise $4M at a $20M post-money valuation, and the lead requires a 10% post-money option pool created before the money comes in. - **Investors:** 20% (that is $4M divided by $20M), or 2,285,714 shares. - **New option pool:** 10%, or 1,142,857 shares. - **Founders:** the remaining 70%, still 8,000,000 shares, now 35% each. - **Price per share:** $1.75. Pre-money fully diluted shares are 9,142,857, which is the founders plus the pool, valued at $16M. The headline says founders sold 20%. The real number is 30%, because the pool is carved out of the pre-money. This is the "option pool shuffle," and it is the single most negotiable term in a seed term sheet. Pushing the pool target from 10% down to 7%, or sizing it to the actual 18-month hiring plan, returns roughly 3 points of founder ownership. Our write-up of [what a term sheet really commits you to](/journal/what-is-a-term-sheet) covers where pool sizing sits among the other clauses. ![Paper pie divided into unequal colored slices to illustrate ownership dilution](https://fdzlnqpwsaniezitwiuw.supabase.co/storage/v1/object/public/cms-media/aistartupinsights/2026-10/05bbbc-i1.webp) ## What is a pro forma cap table, and when do you need one? A pro forma cap table is the post-transaction version of the ledger, built before the transaction closes. It shows what each holder owns if the round closes at a given valuation, check size and pool target. It becomes the real cap table the moment the round funds. Build one at three moments: - **Before you accept a term sheet,** to see the real founder dilution, not the headline. - **When comparing two offers,** because a higher valuation with a larger pool can leave you owning less. - **Before a hiring push,** to test whether the pool covers the next 18 months of grants. The point is not precision to the share. The point is seeing the delta between what an investor says you are selling and what you are selling. Investors run this math on their side before they send the term sheet. Run it on yours first. ## What changed on AI startup cap tables in 2026? Three structural shifts, all visible in Carta's 2026 benchmarks of software rounds. The seed median now sits at a $24.3M valuation on $4.1M raised, with, by Carta's own estimate, probably at least half of the sampled companies AI-native. Higher valuations at the same check size mean less dilution per round on paper. At the top of the range, the 95th percentile seed valuation in Q2 2026 reached $200.4M, against $72.2M a year earlier. That is a signal about concentration, not about how easy fundraising became: Carta also reports fewer rounds completed. For a founder, the practical effect is a wider spread between a clean cap table and a messy one. Investors paying a premium valuation read the cap table harder, not softer. AI companies also carry a cost that older SaaS cap tables did not: expensive technical hires. A pool that looked generous for a five-person software team looks thin when you are competing for research engineers. Size the pool from named roles and levels, not from a rule of thumb. ![Hands signing financing documents across a table during a funding round](https://fdzlnqpwsaniezitwiuw.supabase.co/storage/v1/object/public/cms-media/aistartupinsights/2026-10/8c8e60-i2.webp) ## Why does the cap table matter at exit, not only at fundraising? Percentages stop being the whole story at an exit. Preferred stock carries a liquidation preference, so in a sale below the last post-money valuation the investors can take their money back before common holders see anything. A founder holding 35% on paper can receive far less than 35% of the proceeds, and employees holding exercised or unexercised options sit behind both. This is why a cap table needs the preference terms attached to each class, not only share counts. A one-times non-participating preference is the market norm and rarely hurts founders in a good outcome. Participating preferred, or stacked preferences from several rounds, is where the waterfall bends against common. Model a sale at half of your last valuation and at double it. The gap between the two outputs tells you how much of your ownership is real in a downside, and how much only exists in the upside case. Employees deserve the same clarity. Options only convert to cash if the strike price is below the sale price and the grant has vested. A pool that looks large on a fully diluted basis can still pay out little to the team if the strike prices were set late in a high-valuation round. ## Which cap table mistakes cost the most? Five mistakes recur, ordered by how much damage they do at the next diligence: - **Verbal equity promises.** An advisor who was told "about 0.5%" with nothing signed is a diligence finding and a negotiation you did not plan for. - **Untracked SAFEs and notes.** Three SAFEs with different caps, one with a discount and one without, produce a conversion no one at the company can recompute on the spot. - **No vesting on founder shares.** Investors will impose it at the round. Better to set it up on your own terms at incorporation. - **Version drift.** A spreadsheet emailed between cofounder, lawyer and investor ends up in four versions with three different totals. - **Granting options before the board approves them.** The grant is not valid until the paperwork is, and the strike price depends on a valuation you may not have. The fix is boring: one source of truth, a log of every change, and signed documents behind every line. Board consents and grant agreements go through an e-signature tool so the trail is dated. ## Spreadsheet or software: which one should you run? A spreadsheet is fine until the first priced round. Before that, a cap table has a handful of holders and one or two SAFEs, and a clean sheet with a change log does the job. Worth the move to software the moment any of these is true: you have more than about ten option holders, you are about to close a priced round, or you need 409A valuations and board-approved grants in the same system. Software earns its price on the audit trail and on scenario modeling, not on the ledger itself. Carta is the default most lawyers and lead investors already use, which removes friction in diligence. We measured its entry tiers in our [Carta pricing review](/journal/carta-pricing). The trade-off is cost growth as holder counts rise, so check the next tier before you migrate. On the investor side, the cap table is a research input. Funds pull round history and prior ownership from a database such as PitchBook to see who else holds paper, and track relationships in a CRM such as Affinity. Reading your own cap table the way a fund analyst reads it is the cheapest preparation for a raise. ## What to do this week Open your cap table and answer four questions without opening another file: what is your fully diluted share count, what percentage does each founder own on that basis, how many shares are reserved and ungranted in the pool, and which SAFEs or notes are still unconverted. If any answer takes more than five minutes to find, that is the work for this week. Then build the pro forma for your next round at two valuations and one pool size, and compare the founder column. The signal to retain: the cap table is not a bookkeeping chore. It is the model of your company's ownership, and everyone across the table from you already has theirs. ## FAQ ### What is a cap table in simple terms? A cap table is the ledger of company ownership. Each line names a holder, the type of security they hold, how many shares, what they paid and the resulting ownership percentage, including options, SAFEs and warrants. ### What is the difference between outstanding and fully diluted shares? Outstanding shares are those already issued. Fully diluted shares add every granted option, every reserved but ungranted pool share and every warrant, and usually the shares SAFEs will convert into. Investors price rounds on the fully diluted count. ### What is a pro forma cap table? A pro forma cap table shows ownership after a planned transaction, such as a priced round, at a given valuation, check size and option pool target. Founders build it before signing a term sheet to see real dilution. ### How much dilution should a founder expect at seed? Carta's July 2026 data puts the median seed round at 18% dilution. Founders also absorb the option pool when it is created pre-money, so effective dilution is often several points higher than the headline. ### What is the option pool shuffle? The option pool shuffle is a pre-money option pool increase. The new pool shares are counted before the investment, so existing holders absorb the dilution and the effective pre-money valuation falls. Negotiating the pool size is a core term sheet lever. ### When should a startup move from a spreadsheet to cap table software? Usually before the first priced round, or once there are roughly ten or more option holders. Software adds an audit trail, scenario modeling and board-approved grant tracking that a shared spreadsheet handles poorly.