Let’s talk about Safes.

If you’re not familiar, a Safe is an option contract that lets investors pre-pay for the right to receive equity in the future. Safe stands for Simple Agreement for Future Equity. Typically the Safe converts into equity in a later financing round and because the investor pre-paid she’s given better terms than the new investors.

Safes are easy.

The standard template—created by the startup accelerator, Y Combinator—is six pages. By contrast, the NVCA templates used in most preferred equity financings are more like 150 dense pages. It’s possible to cut out 144 pages of terms because Safes—like convertible notes—simply inherit the terms of the next “priced equity” financing. In other words, next time the company raises money and negotiates that full set of NVCA docs, the Safe investors get those terms too.

(Except the Safe investor gets a better price. Otherwise everything is the same.)

This isn’t perfect. But it works pretty well.

It works pretty well, that is, when the Safe is used as intended. In other situations it’s not so great.

There are two reasons why.

First, if you’re going to invest your money and let someone else negotiate your terms, you’d better have a pretty good idea of what they’re going to ask for. YC created and uses the Safe because YC invests in one type of company: high-growth startups structured as C corporations and destined to raise venture capital. (Or destined to try. More on that in a second.)

In this context, YC feels safe (hah!) handing off negotiations because they know what the venture capitalists who fund priced-equity rounds are going to ask for, at least generally.

Note that part of an investor’s risk in accepting a Safe is that the next investors will negotiate stupid terms. In the YC ecosystem this risk is low because VCs that push stupid terms get the painted rope. (Metaphorically speaking. Of course.)

This doesn’t work when today’s investors don’t know what tomorrow’s investors will do.

And outside of startups, that’s the situation.

Service business investments aren’t predictable from round to round. They’re idiosyncratic. Manufacturers. Content creators. Import/export firms. Film studios. Cattle ranches. Deep sea exploration ventures. (Too soon.) These are all fine businesses and they may raise round after round of capital, but unless they’re in an industry as niche and homogenous as venture-backed startups, the terms of the next round are not predictable enough for investors to bet on.

The second reason Safes only work in Startupland is that they were designed for all or nothing bets.

Every investor hopes for a great return. But if the bet goes bad, most investors want to recoup what they can.

Not YC.

Not seed-stage startup investors.

Why? Because when a startup fails—particularly when it fails at the seed stage, where Safes are prevalent—there’s nothing left. Maybe a few sticker-caked laptops. Maybe some intellectual property—the code base, branding, and trade secrets—which might sound valuable. But it’s probably not worth much. The company just failed after all.

Recovering some value from a failed startup is way more trouble than it’s worth.

Take the Y Combinator example. YC works with five to six hundred startups every year. Something like 20% of them fail—probably way more. If YC were a bank it would need a team dedicated to recovering and reselling worthless IP and depreciated laptops. Or it could cut its losses and focus on its winners, including many of the most successful companies around—Dropbox, DoorDash, AirBNB, Stripe—which bring unfathomable returns to YC.

And that’s just what a Safe does.

A Safe is only a right to future equity, without even the creditor rights you’d get on a convertible note. (Which, tellingly, YC stopped using when they created the Safe.)

The Safe investor has no right to company assets, hence no right to assets after a failure—and no obligation to pursue them or help wrap things up for that matter.

In almost every other situation when a business fails there is something meaningful to recoup by selling off assets or through a fire-sale acquisition. In those situations, the investors would prefer to recover something. And it’s worth the effort to do it. But if the investors used a Safe they’re out of luck.

One final note. The Safe is a template. YC (and industry practice) discourage deviation from the template terms. But sometimes a Safe will be changed to address these issues. But that negates the value of a Safe.

Safes are great where future terms are predicable and failed bets go to zero. They’re fast. They’re efficient. They’re aligning. In every other situation it is better to use some other instrument—equity, debt, or options—and fully negotiate the terms today.