SAFT (Simple Agreement for Future Tokens)

What the agreement actually does

A SAFT is a purchase agreement. An investor pays now. The issuer promises to deliver a fixed quantity of tokens, usually at a discount to a later public price, when a token generation event occurs. Nothing exists onchain when the money moves. What the investor holds is a contractual claim against the issuer, and it ranks like one if the issuer fails.

The structure spread through the 2017 and 2018 token sale wave for a practical reason. Founders wanted institutional capital before a network existed, and private placement machinery already knew how to sell a contract to accredited investors. The token was the novel part. The wrapper was ordinary.

That the SAFT is itself a security is not contested. An investor pays money, into a common enterprise, expecting profits that depend on the issuer's work to build and launch the network. All four Howey prongs are present on the face of the instrument.¹ The argument has only ever been about the token at the other end.

The two-instrument thesis, stated plainly

The theory ran like this. Sell a security to accredited investors under Regulation D while the network is being built. By the time the network launches and tokens are delivered, the network works, the token is bought to be used rather than held for appreciation, and the delivered token sits outside Howey. One security in, one non-security out.

It is not a foolish theory, and it maps onto real factors. The SEC's 2019 staff framework treats a network and asset that are already fully developed and operational at the time of distribution as cutting against reliance on the efforts of others, alongside the absence of a participant whose essential managerial work the network still depends on, and marketing that does not emphasize price appreciation.²

The problem sits in the assumption between the two halves. The thesis needs the agreement and the delivery to be analyzed as separate events. If a court reads them as one offering, the second half never gets its own analysis at all.

The Regulation D machinery a SAFT actually runs on

The exemption carries the legal load, and it has specific conditions. Rule 506(b) permits no general solicitation or advertising, allows unlimited accredited investors plus up to 35 non-accredited but sophisticated purchasers, and lets the issuer rely on a reasonable belief about accredited status. Rule 506(c) permits general solicitation, but every purchaser has to be accredited and the issuer must take reasonable steps to verify it, which means reviewing tax returns, W-2s, brokerage statements or third-party confirmations rather than accepting a checkbox.⁴

Accredited status for an individual runs on income above $200,000, or $300,000 jointly with a spouse, in each of the two most recent years with a reasonable expectation of the same in the current year, or net worth above $1,000,000 excluding the primary residence. Entity tests are separate and broader, covering banks, registered funds, and entities holding over $5 million in assets that were not formed to make the specific investment.³

Then the filing. An issuer relying on either safe harbor does not register the offering, but must file a Form D notice on EDGAR within 15 calendar days after the first sale.⁵ For the offshore leg, Regulation S provides a separate safe harbor at 17 CFR sections 230.901 to 230.905, conditioned on the transaction being offshore and on no directed selling efforts into the United States, with distribution compliance periods of 40 days to a year depending on category. Reg S does not free the securities from US resale restrictions.

None of this is exotic. It is close to the paperwork a software company runs for a priced round. Token rounds rarely go wrong on the Form D. They go wrong at delivery.

Where the thesis met a court

The case commentators point to is SEC v. Telegram Group Inc. and TON Issuer Inc., No. 1:19-cv-09439 (S.D.N.Y.). The SEC filed suit in October 2019 and the court issued a preliminary injunction on March 24, 2020. Telegram had sold rights to future Gram tokens to accredited investors through private purchase agreements across 2018 and 2019, a SAFT-shaped structure, with delivery due once the TON blockchain launched.

The court found the SEC likely to succeed in showing that the entire scheme, the private agreements together with the planned public distribution and resale of Grams, was a single unregistered securities offering. It declined to treat the purchase agreement as the only security while the delivered token became a non-security utility asset. Telegram did not proceed with a US distribution of Grams.

Read that at its real weight. It is a preliminary injunction from one district court, decided on that record, and this page states nothing about the case beyond its docket, its date and that holding. But it is the ruling that moved the two-instrument thesis out of the mainstream, and in our view it did so on reasoning any court would find available again. If the investors were always going to resell into a public market, the resale was part of the plan being financed.

What the delivered token still has to answer for

A SAFT changes nothing about the analysis at delivery. On the day tokens land, the questions are the ones the 2019 staff framework sets out. Is the network fully developed and operational, or is the token going out on the promise of development still to come? Is there an identifiable active participant whose essential managerial or entrepreneurial efforts the network's value still depends on? Are holdings and governance concentrated enough that a promoter can still move that value? Does the marketing stress price appreciation and trading?²

The framework also disposes of the drafting answer directly. Disclaimers alone are not dispositive, because the analysis follows economic reality and the actual conduct of issuer and purchasers rather than the language in the documents.² A clause saying the token is not an investment does not survive a founder posting about listings.

Which produces the test we put to founders. If you delivered tokens tomorrow, what specifically would a buyer be relying on your team to do next, and what would have to change for that answer to be nothing at all? A team that cannot answer the second half does not have a delivery-date problem. It has a design problem.

What practitioners use now

The standalone SAFT has largely been displaced in US venture rounds. Across the rounds we have seen since 2021, the common structure is equity plus a token warrant or a token side letter: the investor buys equity in the company on ordinary terms and receives a separate right to a pro rata share of tokens if and when a token is issued. That keeps the priced round on familiar ground and stops the token from carrying the entire fundraising story.

Where a token-first instrument is still used, three things have changed in how it is drafted. Delivery is conditioned on network state rather than on a calendar date. Lockups and transfer restrictions run well past the generation event instead of ending at it. And the offshore leg runs through Regulation S with real attention to directed selling efforts rather than as a formality.

The policy backdrop moved in March 2026. The SEC issued an interpretation, joined by the CFTC, that sets out a token taxonomy and addresses how a non-security crypto asset may become subject to, and may cease to be subject to, an investment contract, applying that reasoning to airdrops, protocol mining, protocol staking and wrapping.⁶ That is Commission-level interpretive guidance: not a statute, not a court holding, and revisable by a future Commission.

Congress has not finished the job. The Digital Asset Market Clarity Act of 2025, H.R. 3633 in the 119th Congress, passed the House on July 17, 2025 by a recorded vote of 294 to 134 (Roll No. 199), went to the Senate Committee on Banking, Housing, and Urban Affairs on September 18, 2025, and was reported out with an amendment on June 1, 2026. Through that date it had not passed the Senate and had not been enacted. Its predecessor concept, FIT21, passed the House in May 2024 and expired unenacted.

What we settle before a token round

Five answers, in writing, before a term sheet circulates. Which exemption each tranche runs under, and whether solicitation has already happened in a way that forecloses 506(b). What triggers delivery, stated as a network condition rather than a date. What the buyer is expected to do with the tokens on day one, and whether anything on the network supports that. What restrictions survive delivery. And what the team's public communications may say for the whole period, because the offering is not over when the wire clears.

The pattern we see most often is a clean Reg D file and an unexamined delivery. Counsel papers the round, the round closes, and eighteen months later the same tokens go out to the same investors into a market the team has been promoting the entire time, with nobody having revisited whether that distribution is its own offering. That is the expensive gap, and it is a design problem before it is a legal one.

Whether a specific SAFT, delivery structure, or token is a security is fact specific and jurisdiction specific, and that call belongs to your counsel and, ultimately, a court. This page is reference material for structuring work. It is not legal advice, and it is not a recommendation to buy, sell, or hold any asset.