Paul Graham recently made a very interesting point about venture funding: it's how companies choose their growth rate. Plenty of companies don't need outside money, because they're closer to a regular job: lawyers, dentists, and realtors all fit into this model, but so do founders of small-scale software companies that can quickly and profitably saturate their market but can't really expand it. Taking outside money is a bet that the current opportunity set, or a near future opportunity set, will include expansion opportunities that outstrip a company's ability to fund its growth internally, and that if it doesn't have the capability to burn money for a while in response, it will cede the market to someone who can.¹ For example, a business model like "CRM software for independent gym operators" may or may not be a huge business, but it's a business with a fairly fixed market and limited long-term growth. "CRM for physical stores," on the other hand, is a huge business, and one that, if it doesn't fail, will be able to profitably deploy more capital than it can earn from operations.

The first step a company takes towards this growth model is raising money from angel investors. These deals used to be mysterious and ad hoc, but now there's a pretty established template: do a Simple Agreement for Future Equity (SAFE) using the Y Combinator documents.² There are a few permutations of this, but the basic idea of the SAFE is that it's money in now with exact terms, based on whatever the company raises when it does a real round. The SAFE investors typically get paid for the risk they take relative to the next investors through either a valuation cap or a discount on whatever the priced round ends up being.

If you think of investments on a spectrum where risk and reward are roughly correlated, angel investing is an extreme case of high-risk/high-reward. But that obfuscates a critical difference: they include more Knightian uncertainty; you can make reasonable assumptions about what the stock market will do, or what individual stocks will do; a late-stage investment is typically underwritten with plenty of data about a company's economic model, so even if you can't be sure that the business will succeed, you can at least come up with a reasonable enumeration of what might go right or wrong. But the earlier things get, the more it's fundamentally unknowable. The business might pivot. It might pivot multiple times! OpenAI might casually launch a feature that kills it. The founders might have a fight and break up—if a three-person company raises a round, and a few weeks later one of the founders quits, there's a very real sense in which one third of the company's assets have walked out the door.³

And supposing everything goes right, there's a whole new set of risks. Early-stage investors tend to use standardized documents and terms. But later-stage investors don't. Instead, they get relatively less alpha from choosing the right company to invest in, but relatively more by choosing the right terms. And the "right terms," for them, means dividing the economic pie more in their favor. There are other groups to consider, of course: they don't want to do this at the expense of founders or employees, if they can help it, because for both groups equity is a big incentive to maximize the value of the investor's stake. But they have fewer compunctions about structuring deals that weaken the economics of the earliest-stage investors. Those investors won't be doing deals alongside the late-stage VCs, on average, and don't have a lot of pull with the company. The founder might feel a sentimental tug at the idea of diluting someone who believed in them and trusted them at a critical time, but that founder also needs to do what's right for the company and the employees. Often, everybody gets roughly what they wanted when a company succeeds, but if someone isn't going to get what they wanted, it's likely to be angel investors, ex-employees, and other groups whose equity ownership exceeds their ability to help or harm the company.

And even if everything goes perfectly well, the angel investor has another problem, albeit a nice one: they have a comically un-diversified portfolio, and their ability to monitor their biggest asset will tend to decline as that asset gets bigger, at least until the IPO. It is, of course, much better to be wealthy but undiversified than to not be wealthy in the first place, but it’s also nerve-wracking. Almost no one exits right at the peak, so the options are either a) sell some early and always know how much more you could have made, or b) sell late, and know how much more you could have had if you’d held. This is true for every investor in every asset class, of course, but when the numbers are bigger the psychology gets harder.

So why do people angel invest? There are a few good reasons:

1. Even if you get diluted, even if future rounds continuously squeeze you out, even if you get less and less information about how the company is doing over time—backing the right company can produce a life-changing amount of money. Jonah Peretti cofounded the Huffington Post, and when that company sold to AOL, he invested $10,000 of the proceeds into Uber, and then made twice as much from backing Uber as he did from Huffpo. (This fun detail comes from Ben Smith's Traffic.)

2. Angel investing is both a good forcing function for keeping an eye on what risk-tolerant people are building and a good way to amortize the cost of doing so. If you've spent the last six or twelve months obsessing about AI (or the last few years—good for you!), then you'll be better positioned to understand which AI-related pitches are weak and which ones aren't. Realistically this means going from, say, 1% odds of making your money back to 3% (of course, at an appropriate discount rate). But it also offers the chance to have a truly big win. It’s worth noting that this is an expensive way to keep track of what’s going on in the world; even the semi-active angel investor will see more cash outflows than they’d have from subscribing to a Bloomberg Terminal, and the value of the information ought to be considered accordingly. But for some people, it’s a perfectly reasonable decision.

3. Alex Danco proposes another model: people do angel investing, despite low risk-adjusted returns, specifically for bragging rights. Tech is an industry where, at least in recent years, following a standard track and working at whichever company is the high bidder for your time can produce results that the rest of the country would characterize as "wealthy." So what's a poor Palo Alto millionaire to do when everyone else they know is also a millionaire? If they can say they were one of the first checks into Databricks, it actually sets them apart even if it hasn't affected their liquid net worth just yet.

Angel investing is, and should be, difficult. Investing in mature companies is hard enough, and investing in companies that basically don't exist yet is that much harder. Deal flow is a particular challenge for investors: they need to get the pitches, and not just pitches from people who have been turned down by everyone else first. As much as it pays to be a contrarian, the best early-stage investors are in fact very good at what they do, and getting into deals they've rejected is a very hard way to make money. The bragging rights, however, are unbeatable.

Read More in The Diff

The Diff spends a lot of time on the venture world, with angel investing itself getting a bit less attention but still playing an important role. These include:

• VC investment is partly driven by individuals’ desire to get credit for owning a stake in a hot company, even if that means paying up for a later round.

• VC is a systematic search for outliers ($), but outliers are by definition hard to spot systematically.

• Early-stage valuations are surprisingly stable, because higher volatility offsets lower expected values ($).

1. This isn't necessarily required, and some companies can scale with little or no external capital. But to do this sustainably, they need to be in a business where the opportunities always look worse from the outside than they really are. It's hard to maintain that information gap indefinitely; if nothing else, someone will eventually notice that the company is growing its headcount fast enough that it must be doing something right.

2. This, incidentally, is a fun example of commoditizing the complement. YC is providing free documents that make it easier to raise money from non-YC investors, but that also makes the signaling value of YC stronger, both because it’s a branding exercise and because it means that YC's funding goes to a smaller set of the fundable companies.

3. Rough number, of course; if we assume that the most disagreeable founder is also the most likely to quit, perhaps the company's biggest liability just left, too. But events like this also force investors to update their thesis: if one of the people who is best-informed about a company's prospects decides it's no longer worth their time, that's an important signal.

Join the discussion!