Pre-Seed, Seed, and Series A

What changes between pre-seed, seed, and Series A?#

Three things move with the stage label, and it is worth being precise about which:

• The burden of proof. Pre-seed rests on a sharp problem, a hidden insight, and a credible plan for turning uncertainty into evidence. Seed asks what customers actually do—activation, usage, retention, paid pilots, conversion. Series A asks whether something is worth scaling, which means cohort quality, pipeline conversion, gross margin, churn, and CAC payback.
• The financing structure. Early rounds use convertible instruments because the company is hard to value and needs to close quickly. The SEC defines a convertible note as a loan that can convert into another security, and a SAFE as a promise of a future ownership interest upon specified triggers. A note is debt with interest and a maturity date; a SAFE is neither. Both create dilution when they convert. Institutional Series A financings are priced preferred-stock rounds that make valuation explicit and add a fuller package of economic, governance, and protective rights.
• Dilution and control. Instruments and deal terms determine dilution; labels do not. A Series A adds a board seat, protective provisions, pro rata rights, a liquidation preference, and an option-pool increase. Compare fully diluted post-closing ownership, not the headline valuation.

What these rounds look like in 2026#

Round sizes and graduation rates: Crunchbase, May 2026. Instruments and pre-seed figures: Carta, State of Pre-Seed Q2 2026 (Aug 2026) and SAFE valuation caps (Aug 2026). Founder ownership: Carta, Founder Ownership Report 2026 (Mar 2026), covering rounds raised 2021–2025.

Read the spread, not the median. In Q1 2026 the 90th-percentile seed valuation was roughly 4x the median, and the top-decile Series A roughly 5x the median. A median here is the midpoint between two markets that barely touch: foundational-model AI startups raised at a ~$300M median Series A valuation against ~$55M for non-AI startups. In Carta's own words, these "are not comparable markets." A founder benchmarking against a single number is benchmarking against a market that does not exist.

Pre-seed capital is concentrating, not growing. US startups on Carta raised $3.19B across more than 11,500 pre-seed instruments in Q2 2026, against $3.22B across 14,825 instruments a year earlier—flat dollars, roughly 22% fewer deals, and a record average instrument size of $276K (up 27% year over year). AI's share of pre-seed dollars held at 49% in H1 2026 against 50% for full-year 2025, which Carta reads as a plateau rather than a continued rise.

How long does a stage last, and what are the odds?#

Through 2020, most US startups raising a $1M+ seed went on to a later round or an exit—Crunchbase puts the typical graduation rate at 55% or higher. For the 2023 seed cohort it is 24%. For 2024, 16%. Part of that decline is a measurement artifact, because recent cohorts have had less time to advance. But the interval itself has stretched past two years, so the censoring and the deterioration compound rather than cancel.

Two consequences for how you size a round:

1. The "raise 18–24 months of runway" heuristic is now shorter than the median seed-to-Series-A interval. Following it is, arithmetically, a plan to run out of money mid-process. The buffer term in the formula below is not optional.
2. The Series A evidence bar moved. The $1M ARR convention is out of date. Andy McLoughlin of Uncork Capital—whose own average seed check grew from $2.5M or less to $4.5M over roughly 18 months—puts the current bar at $2M to $4M ARR (Crunchbase, May 2026).

How do you size a round around a risk-retiring milestone?#

1. Name the risk. Write one sentence: "This round exists to prove X." If the sentence only describes hiring or spending, it is not yet a milestone.

2. Define the proof. Choose the smallest set of metrics that would change a rational financing decision, using cohorts and unit economics where possible. The metric must fit the motion: activation and retention for product-led software; win rate, sales cycle, and CAC payback for sales-led.

3. Calculate the cash needed. Use cash timing, not accounting profit:

   Monthly net burn = cash operating outflows − cash operating inflows

   Funding need = (monthly net burn × months to milestone) + one-time milestone costs + cash buffer − current unrestricted cash − committed non-dilutive inflows

4. Map the instrument to the facts. Use a SAFE only after modeling its conversion and cumulative dilution; a note only if you understand maturity, interest, and repayment risk; a priced round when size, investor requirements, or cap-table complexity justify full documents.

5. Check the legal pathway. The round name is irrelevant to securities-law compliance—see the note on legal status below. Use qualified counsel.

What are the most common mistakes?#

• Treating the stage as a legal status. "Pre-seed," "seed," and "Series A" do not select a securities exemption, permit general solicitation, or remove disclosure and anti-fraud duties. Every offer and sale of securities in the U.S. must be registered or rely on an available exemption—Regulation D, Regulation Crowdfunding, and Regulation A each carry different limits and rules. The instrument and offering pathway matter; the label does not.
• Treating fundraising as the milestone. A financing event is an input, not customer or enterprise value. The useful milestone is the measurable value created with the capital.
• Ignoring cumulative SAFE dilution. The mechanism is specific: post-money SAFEs do not dilute each other. Each investor's ownership is fixed at signing, so every subsequent SAFE dilutes the founders and the option pool instead of prior holders. With 93% of pre-seed rounds structured as SAFEs and 91% of those post-money, this is the default case, not an edge case—and pre-seed rounds above $2.5M typically stack ten or more instruments, each a separate claim on your stake that stays invisible until the priced round. Median valuation caps on SAFEs above $2.5M reached $35M in Q2 2026, up 40% year over year, with the 90th percentile reaching $100M. A cap is a promise to grow into a number. See SAFE notes and dilution and cap tables for the conversion math.
• Optimizing only for valuation. Board structure, liquidation preference, option-pool treatment, and investor quality can matter more than a modest valuation difference.

When does the stage framework break down?#

• Capital-intensive and regulated companies. Biotech, climate, hardware, and infrastructure raise large rounds pre-revenue because technical, clinical, or regulatory proof is the relevant milestone; a software ARR threshold misclassifies them. The cost shows up in ownership: at Series A, founders in physical industries hold a median 30.5% of their company against 37.5% for digital industries (Carta, Founder Ownership Report 2026). Seven percentage points is not a rounding error, and round sizes alone do not explain it.
• Sector concentration. Benchmarks can reflect capital concentration rather than universal quality—see the AI-versus-non-AI Series A valuation gap above.
• Bootstrapped or profitable businesses. Revenue, prepayments, grants, or suitable debt may finance the plan with better founder economics. Venture stages are a funding path, not a maturity ladder every company must climb.
• Bridge rounds and extensions. A seed extension, SAFE bridge, or insider round may buy time without changing the company's true evidence stage. Renaming a round does not retire risk.
• A closed or selective capital market. Reaching a milestone does not guarantee financing; fund strategy, sentiment, and reserves can delay or prevent a round. Build enough runway to survive a slower process.

Not every cost should be financed with equity#

For a company that must buy, install, or operate physical assets, it prices two very different risks identically. Split the next round's spend into two buckets:

• Uncollateralizable—the physics, the algorithms, the first reference deployment, the team. Nobody will lend against this. Venture equity is correctly priced for it.
• Collateralizable—everything downstream of a signed offtake or a contracted deployment. This is an asset with a cash-flow profile a lender can underwrite. Paying venture prices for it is a category error.

Carta's Q1 2026 market review shows hardware as the quiet winner of the current cycle, observing that "non-dilutive debt is letting hardware founders build before they raise," arriving at institutional rounds with less dilution than prior generations.

The practical question is therefore "what fraction of my next 24 months of spend is uncollateralizable?"—then raise the smallest equity round that covers only that, and finance the rest against contracted cash flows.

Two honest caveats. A deployment becomes financeable only at a specific combination of contract length and counterparty credit quality; below that line the debt is not available at any sane price. And investor psychology cuts both ways—some read a smaller equity round plus structured debt as capital discipline, others read it as insufficient conviction.