Convertible notes vs SAFEs: the differences that still matter

I price private companies for a living. More than a thousand of them, on deals up to roughly $26 billion, and when a company sends me its cap table, the SAFEs and the convertible notes arrive as the same kind of object: claims that are not shares yet and will be. For most of the last decade the argument over which one to use has been conducted loudly and settled decisively in favor of the SAFE. The market is not close. On Carta's pre-seed data, convertible notes fell to a record low of 7 percent of pre-seed rounds and 8 percent of pre-seed dollars in the first quarter of 2026. That is not a debate, that is a rout.

So why write about the loser? Because the 7 percent is not randomly distributed, and because of something I notice every time I model one of these instruments: when the round goes to plan, a SAFE and a convertible note produce almost the same cap table. Same cap, same discount, same conversion into the same shadow series at the same Series A. If the two instruments were only ever going to be tested by success, the choice would be trivia. They are not. Every difference that still matters between them shows up in a state of the world where the plan does not happen: the raise slips, the round is a bridge that never lands, the company sells for less than anyone hoped, or it winds down. That is the honest way to read this comparison, and it is not how the comparison is usually written.

What the market actually decided

Start with the numbers, because they are unambiguous. Carta's State of Pre-Seed data shows SAFEs at 93 percent of pre-priced deals in the fourth quarter of 2025 against 7 percent for convertible notes, and 89 percent of the dollars against 11 percent. A year earlier, notes were still around 10 percent of pre-seed rounds. By the first quarter of 2026 they had slipped to that record low. Y Combinator introduced the SAFE in 2013 and replaced it with the post-money version in 2018, and the post-money SAFE is now simply what a pre-seed round is made of.

The reasons are not mysterious. A SAFE is a handful of pages of standard-form document, it costs essentially nothing to paper, it has no maturity date to renegotiate, and its post-money version makes the ownership sold knowable on the day you sign it. A convertible note requires a note purchase agreement, an interest rate, a maturity date, and a set of amendment mechanics, all of which are negotiable and therefore all of which cost time. For a company raising $500,000 from five angels, that overhead buys nothing anybody wants.

Carta adds a detail that matters more than the headline: the residual note usage is concentrated, not scattered. The sectors with the highest representation of convertible notes are energy, biotech and pharma, and medical devices. Hold that thought, because by the end of this piece it will look less like a legacy habit and more like a rational choice by people who have thought hard about what happens if the science does not work.

The differences that stopped mattering

Most comparison tables you will find are accurate and useless, because they list the features both instruments share as though they were distinctions. Let me clear those out. Valuation cap: both. Discount: both. Most-favored-nation clause: both. Pro-rata rights: a side letter, available on both. Conversion into a shadow series of preferred at the next priced round rather than into common: standard on both. Counting in the fully diluted denominator before conversion: both, and if you are not counting them, read how to read a cap table before you sign anything. Most importantly, both instruments do the identical central thing: they let you take money now and set the price later. That deferral is the entire product, and neither one is better at it.

What remains after you strip out the shared features is one structural fact with five consequences. A convertible note is debt. A SAFE is not. Everything below follows from that sentence.

One: the maturity date is an option the investor owns

A note has a date on which principal and accrued interest become payable. Cooley GO's primer on convertible debt, which is the practitioner reference most startup lawyers work from, describes maturity as a function of stage, market, and round size, with the practical band running roughly one to three years and eighteen to twenty-four months as the common case. If no qualifying financing has closed by that date, the note is due.

The standard reassurance is that this never happens, and the reassurance is broadly true. Calling a note on a pre-revenue startup usually destroys the company and pays the investor nothing, so the overwhelmingly common outcome is an amendment that extends maturity, typically by another year. Founders hear that and conclude the maturity date is theater. It is not. The maturity date is an option, and the investor owns it. It converts a passive holder into a counterparty with a decision to make, and it schedules that decision for the precise moment your leverage is at its lowest: you have not raised, you are running out of money, and you need something from them.

Which is why the clause I read first in a note is not the interest rate. It is the amendment provision. If the notes can be amended by holders of a majority in interest, an extension is a conversation with the group and the group's incentives are broadly aligned with yours. If amendment requires unanimous consent, then every individual holder, including the angel who wrote $25,000 and has not answered an email in a year, holds a veto over your extension, and a veto is a thing that can be priced. I have watched a company give away real economics to buy a signature from a small holder who simply understood the leverage he had been handed. Read the amendment mechanics before you read the rate.

The flip side deserves saying, because it is the strongest argument for a note: sometimes a deadline is the feature. When the money is a genuine bridge between priced rounds, a maturity date is what forces a decision instead of allowing indefinite drift. On a bridge, I want the clock.

Two: interest is dilution with a schedule

Notes accrue interest, and that interest does not usually get paid in cash. It converts. Carta's data put the median convertible-note interest rate at 7 percent in the first quarter of 2025, down from a high of 8 percent in the second quarter of 2024, and Kruze Consulting's 2025 market read describes the working range as roughly 5 to 7 percent simple interest annually.

Here is what that does to a cap table. Take an illustrative $1,000,000 note at 7 percent simple interest that converts twenty-four months after it was signed, at a $10,000,000 cap. The principal alone would buy 10 percent of the company. Two years of interest adds about $140,000 to the converting amount, so $1,140,000 converts and the holder lands at roughly 11.4 percent. That extra 1.4 points of your company cost the investor nothing in additional cash. These figures are illustrative and the arithmetic on your own note will differ, but the shape does not.

A point and a half is not a catastrophe. What makes it worth naming is the correlation. Interest accrues fastest when the note is outstanding longest, which is exactly when the company is taking longer than planned, which is exactly the period in which the maturity clock is also running against you. The two effects are not independent. A note is calibrated so that the slower things go, the more it costs, and it charges you in the currency you are shortest of.

Three: the seniority nobody reads until the bad day

This is the difference I would put first if I were writing for investors rather than founders, and it is the one founders almost never model. A convertible note is a creditor claim. If the company winds down, noteholders are paid out of whatever is left before a single dollar reaches equity of any kind.

You do not have to take my word for the ranking, because the SAFE says it about itself. Section 1(d) of Y Combinator's post-money safe sets out that on a dissolution event the investor's cash-out amount is junior to payment of outstanding indebtedness and creditor claims, and the document names convertible promissory notes explicitly in that category. It is on par with other safes and with preferred stock, and it is senior to common. The most widely used SAFE in the world states in writing that a convertible note outranks it.

Put a number on it. A company with a $1,000,000 note outstanding sells its assets in a soft landing for $3,000,000. The noteholder is a creditor and gets made whole first. Now run the same landing with a $1,000,000 SAFE instead: the SAFE holder queues with preferred, behind whatever debt exists, and takes a pro-rata share of what is there. On a strong exit the distinction is invisible, because everything converts and everybody is paid. On a mediocre one it decides who is made whole.

Now go back to Carta's sector finding. Energy, biotech and pharma, and medical devices are capital-intensive, run on long clinical and engineering timelines, and have a meaningfully higher probability of ending in a wind-down where the residual assets, the equipment, the data, the patents, are still worth real money. That is precisely the scenario in which creditor seniority pays. The 7 percent is not nostalgia. It is people pricing a specific risk correctly.

Four: the balance sheet, the auditor, and the passport

A note is a liability, plainly, on every set of financials you hand to a bank, a grant program, an acquirer, or a landlord doing a credit check. Stack enough of them and a company can be technically balance-sheet insolvent, which is not merely cosmetic: it changes how lenders underwrite you and it is the kind of fact directors are expected to notice.

Founders assume the SAFE cleanly avoids this. It is messier than that. Grant Thornton's 2024 viewpoint on issuers' accounting for SAFEs observes that in practice it is common for an issuer to classify a SAFE as a liability and subsequently measure it at fair value, and that there is genuine diversity in practice over whether the instrument falls under ASC 480 or is classified as a liability under ASC 815-40. In other words, "SAFEs are equity" is a founder belief, not an accounting conclusion. The first time it bites is usually the first audited financial statement, which is also usually the worst time to discover it.

Then there is jurisdiction, which is the difference most likely to be relevant to a reader outside the United States and least likely to be mentioned. The SAFE is an instrument tuned to a Delaware C corporation and to US securities practice. Elsewhere the fit varies from awkward to unworkable, which is why Cooley GO publishes a dedicated piece on the risks of convertible notes in Southeast Asia and India, where local corporate law, exchange-control rules, and pricing regulations shape what a convertible instrument can even be. If your company is not a US C corp, the right question is not which instrument is more founder-friendly. It is which instrument your jurisdiction reliably recognizes. Ask counsel that before you download a template.

Five: the clock neither instrument starts

Here is a difference that is really a similarity, and it may be the most valuable paragraph in this essay. Neither a SAFE nor a convertible note is stock. For qualified small business stock treatment under Section 1202, the holding period begins when shares are actually issued at conversion, not on the day the instrument was signed. Money that sits in an unconverted instrument for two years is money whose clock has not started.

That timing got more consequential in 2025. Under the One Big Beautiful Bill Act, for stock issued after July 4, 2025, the exclusion became tiered rather than all-or-nothing: 50 percent of gain at three years held, 75 percent at four, and 100 percent at five. The same legislation raised the gross asset ceiling from $50 million to $75 million and the per-taxpayer exclusion cap from $10 million to $15 million. A partial exclusion now arrives two years earlier than it used to.

The practical consequence is that sophisticated angels have a new and unsentimental reason to want your round priced. Every month you spend stacking one more instrument is a month nobody's holding period is running. That does not make it wrong to raise on a SAFE, and it is not a reason to price a round before the company is ready. It is a reason to stop treating "we will just do another SAFE" as free. It is not free, and now it is not free in a way your investors can quantify. This is general information rather than tax advice, and it is fact-specific enough that anyone with real money at stake should be talking to qualified counsel.

So which one, and when

My default is the post-money SAFE, and I do not think that is a close call for a US company raising pre-seed or seed money. No maturity clock, no accruing interest, no creditor overhang, a standard document nobody bills you to negotiate, and ownership sold that is knowable on the day you sign. If nothing in your situation argues otherwise, use the market standard and spend your negotiating energy on the cap.

Four situations argue otherwise. First, when an investor requires genuine downside protection and will not do a SAFE, which is common with family offices, corporate investors, and the capital-intensive sectors Carta's data flags. Arguing them out of it is usually a worse trade than accepting a clean note. Second, when the company is not a US entity and local law does not cleanly recognize a SAFE. Third, when the money is a real bridge between priced rounds and the hard date is the point rather than the problem. Fourth, when everyone involved understands the money as a loan, in which case call it what it is and paper it properly.

What to negotiate on each

On a note, in this order: the amendment provision, majority in interest rather than unanimous consent, because it decides who owns you at maturity. Then a maturity honest to your actual plan with a buffer on top, since a date you will obviously miss is a date you are scheduling a bad negotiation for. Then whether interest is simple or compounding, and what happens on a change of control, which is a separately negotiated term and not a standard one. Then whether conversion at a qualifying financing is automatic or elective at the holder's option, and what dollar threshold makes a financing qualifying. The interest rate itself is close to the least important term on the page.

On a SAFE, the cap, and confirming it is the post-money version, which it almost certainly is. Whether a discount stacks on top of the cap, and if so, modeling both. What the most-favored-nation clause will drag in from instruments you sign later. How many pro-rata side letters you have quietly issued and what they will do to the room you have for a lead. And above all the total stack, because the failure mode I see most often is not one bad instrument but five reasonable ones that were never added up. I wrote about that specific failure at length in raising on SAFEs, and about sizing the round that ends the stacking in how much should you actually raise.

The takeaway from someone who prices both

The SAFE won on merit and I would use one tomorrow. But the way the win gets reported, as though the convertible note were an obsolete technology, hides the actual lesson. The two instruments diverge only in the scenarios founders do not model, and founders do not model those scenarios precisely because they are the unpleasant ones. Maturity matters when you have not raised. Interest matters when you are late. Seniority matters when the company is worth less than the money that went in. Balance-sheet treatment matters when somebody underwrites you. And the holding-period clock matters when the exit finally comes and somebody counts the years.

Choose the instrument for the world where things go wrong, because that is the only world in which the choice does anything. Then negotiate the terms that govern that world, which are not the terms anybody spends their time on. And whichever paper you sign, keep a live model of the whole stack on an as-converted basis, because the most expensive mistake in this entire category is not picking the wrong instrument. It is losing count of how many of the right one you signed.

Common questions

What is the actual difference between a SAFE and a convertible note?

A convertible note is debt and a SAFE is not. That single fact generates every real difference. Because a note is a loan, it carries a maturity date at which principal and accrued interest become payable, it accrues interest that converts into additional shares, it sits in the liabilities section of the balance sheet, and it ranks as a creditor claim ahead of all equity if the company winds down. A SAFE has no maturity, no interest, and no repayment right, and it converts only when a triggering event happens. Everything else people list in comparison tables, the valuation cap, the discount, the most-favored-nation clause, pro-rata side letters, conversion into a shadow series at the priced round, is available on both instruments and is not a real point of difference.

Are convertible notes still used, or have SAFEs replaced them?

Largely replaced, but not entirely. On Carta's pre-seed data, SAFEs made up 93 percent of pre-priced deals in the fourth quarter of 2025 against 7 percent for convertible notes, and 89 percent of the dollars against 11 percent. By the first quarter of 2026 convertible notes hit a record low of 7 percent of pre-seed rounds and 8 percent of pre-seed dollars. Carta also notes the residual usage is not evenly spread: energy, biotech and pharma, and medical devices carry the highest representation of convertible notes. Those are capital-intensive sectors with long timelines and a meaningful chance of a wind-down in which residual assets are worth something, which is exactly where creditor seniority pays.

What happens when a convertible note reaches maturity and the startup has not raised?

Formally, principal plus accrued interest becomes due and payable. In practice almost no early-stage investor calls a note, because calling it usually bankrupts the company and pays nobody, so the standard outcome is an amendment extending maturity, commonly by a year. The extension is not free, and the negotiation happens at the exact moment your leverage is lowest. What decides how expensive it gets is the amendment clause. If the note can be amended by holders of a majority in interest, you negotiate with the group. If it requires unanimous consent, any single small holder has a veto over the extension and can price that veto. Read the amendment mechanics before you read the interest rate.

Which is better for a founder, a SAFE or a convertible note?

For a US Delaware C corporation raising pre-seed or seed money, the post-money SAFE is the sensible default: no maturity clock, no accruing interest, no creditor overhang, a standard document, and essentially no negotiation cost. A convertible note earns its place in four situations. When an investor requires genuine downside protection and will not do a SAFE, which is common with family offices, corporates, and investors in capital-intensive sectors. When the company is not a US entity and local law does not cleanly recognize a SAFE. When the money is a real bridge between priced rounds and a hard deadline is the point rather than the problem. And when everyone involved understands the money as a loan. Neither instrument is cheap. Both sell an unpriced claim on your next round.

Does a SAFE or convertible note start the QSBS holding period?

No. Neither instrument is stock, so the qualified small business stock holding period under Section 1202 begins only when shares are actually issued at conversion, not on the date the SAFE or note was signed. That timing became more consequential after the One Big Beautiful Bill Act of 2025. For stock issued after July 4, 2025, the exclusion is tiered rather than all-or-nothing: 50 percent of gain at three years, 75 percent at four, and 100 percent at five, with the gross asset ceiling raised from $50 million to $75 million and the per-taxpayer exclusion cap raised from $10 million to $15 million. Every additional month a SAFE or note sits unconverted is a month the clock is not running, which is a quiet argument for pricing a round instead of stacking one more instrument. This is general information rather than tax advice, and the analysis is fact-specific enough to warrant qualified counsel.

Related reading

• Raising on SAFEs: the dilution you signed and forgot.
• How to read a cap table before it reads you.
• How much should you actually raise?
• The five term sheet clauses that move money.

Go deeper

If you are stacking instruments and you have lost track of what converts into what, let's model the whole stack on an as-converted basis before your next round prices it for you, or read more of how I think about it.

Tomasz Felpel is an investor, founder, and advisor in private markets and healthcare, based in New York. He is a three-time founder of Value Alpha, an AI-powered private-markets valuation platform, Sonnerie VC, an early-stage healthcare venture firm, and Pond. Previously he led corporate development and M&A at Fortune 500 scale, pricing more than 1,000 private companies. Columbia Business School EMBA. Read the full story. This essay is educational and is not legal, tax, or investment advice; work with qualified counsel on any document you sign.