The Future of Startup Funding, Paul Graham

August 2010

The Future of Startup Funding

The opportunity is a lot less unexploited now. Investors have poured into this territory from both directions. VCs are much more likely to make angel-sized investments than they were a year ago. And meanwhile the past year has seen a dramatic increase in a new type of investor: the super-angel, who operates like an angel, but using other people's money, like a VC.

Though a lot of investors are entering this territory, there is still room for more. The distribution of investors should mirror the distribution of startups, which has the usual power law dropoff. So there should be a lot more people investing tens or hundreds of thousands than millions. [2]

In fact, it may be good for angels that there are more people doing angel-sized deals, because if angel rounds become more legitimate, then startups may start to opt for angel rounds even when they could, if they wanted, raise series A rounds from VCs. One reason startups prefer series A rounds is that they're more prestigious. But if angel investors become more active and better known, they'll increasingly be able to compete with VCs in brand.

Of course, prestige isn't the main reason to prefer a series A round. A startup will probably get more attention from investors in a series A round than an angel round. So if a startup is choosing between an angel round and an A round from a good VC fund, I usually advise them to take the A round. [3]

But while series A rounds aren't going away, I think VCs should be more worried about super-angels than vice versa. Despite their name, the super-angels are really mini VC funds, and they clearly have existing VCs in their sights.

They would seem to have history on their side. The pattern here seems the same one we see when startups and established companies enter a new market. Online video becomes possible, and YouTube plunges right in, while existing media companies embrace it only half-willingly, driven more by fear than hope, and aiming more to protect their turf than to do great things for users. Ditto for PayPal. This pattern is repeated over and over, and it's usually the invaders who win. In this case the super-angels are the invaders. Angel rounds are their whole business, as online video was for YouTube. Whereas VCs who make angel investments mostly do it as a way to generate deal flow for series A rounds. [4]

On the other hand, startup investing is a very strange business. Nearly all the returns are concentrated in a few big winners. If the super-angels merely fail to invest in (and to some extent produce) the big winners, they'll be out of business, even if they invest in all the others.

VCs #

Why don't VCs start doing smaller series A rounds? The sticking point is board seats. In a traditional series A round, the partner whose deal it is takes a seat on the startup's board. If we assume the average startup runs for 6 years and a partner can bear to be on 12 boards at once, then a VC fund can do 2 series A deals per partner per year.

It has always seemed to me the solution is to take fewer board seats. You don't have to be on the board to help a startup. Maybe VCs feel they need the power that comes with board membership to ensure their money isn't wasted. But have they tested that theory? Unless they've tried not taking board seats and found their returns are lower, they're not bracketing the problem.

I'm not saying VCs don't help startups. The good ones help them a lot. What I'm saying is that the kind of help that matters, you may not have to be a board member to give. [5]

How will this all play out? Some VCs will probably adapt, by doing more, smaller deals. I wouldn't be surprised if by streamlining their selection process and taking fewer board seats, VC funds could do 2 to 3 times as many series A rounds with no loss of quality.

But other VCs will make no more than superficial changes. VCs are conservative, and the threat to them isn't mortal. The VC funds that don't adapt won't be violently displaced. They'll edge gradually into a different business without realizing it. They'll still do what they will call series A rounds, but these will increasingly be de facto series B rounds. [6]

In such rounds they won't get the 25 to 40% of the company they do now. You don't give up as much of the company in later rounds unless something is seriously wrong. Since the VCs who don't adapt will be investing later, their returns from winners may be smaller. But investing later should also mean they have fewer losers. So their ratio of risk to return may be the same or even better. They'll just have become a different, more conservative, type of investment.

Angels #

In the big angel rounds that increasingly compete with series A rounds, the investors won't take as much equity as VCs do now. And VCs who try to compete with angels by doing more, smaller deals will probably find they have to take less equity to do it. Which is good news for founders: they'll get to keep more of the company.

The deal terms of angel rounds will become less restrictive too — not just less restrictive than series A terms, but less restrictive than angel terms have traditionally been.

In the future, angel rounds will less often be for specific amounts or have a lead investor. In the old days, the standard m.o. for startups was to find one angel to act as the lead investor. You'd negotiate a round size and valuation with the lead, who'd supply some but not all of the money. Then the startup and the lead would cooperate to find the rest.

The future of angel rounds looks more like this: instead of a fixed round size, startups will do a rolling close, where they take money from investors one at a time till they feel they have enough. [7] And though there's going to be one investor who gives them the first check, and his or her help in recruiting other investors will certainly be welcome, this initial investor will no longer be the lead in the old sense of managing the round. The startup will now do that themselves.

There will continue to be lead investors in the sense of investors who take the lead in advising a startup. They may also make the biggest investment. But they won't always have to be the one terms are negotiated with, or be the first money in, as they have in the past. Standardized paperwork will do away with the need to negotiate anything except the valuation, and that will get easier too.

If multiple investors have to share a valuation, it will be whatever the startup can get from the first one to write a check, limited by their guess at whether this will make later investors balk. But there may not have to be just one valuation. Startups are increasingly raising money on convertible notes, and convertible notes have not valuations but at most valuation caps: caps on what the effective valuation will be when the debt converts to equity (in a later round, or upon acquisition if that happens first). That's an important difference because it means a startup could do multiple notes at once with different caps. This is now starting to happen, and I predict it will become more common.

Sheep #

The reason things are moving this way is that the old way sucked for startups. Leads could (and did) use a fixed size round as a legitimate-seeming way of saying what all founders hate to hear: I'll invest if other people will. Most investors, unable to judge startups for themselves, rely instead on the opinions of other investors. If everyone wants in, they want in too; if not, not. Founders hate this because it's a recipe for deadlock, and delay is the thing a startup can least afford. Most investors know this m.o. is lame, and few say openly that they're doing it. But the craftier ones achieve the same result by offering to lead rounds of fixed size and supplying only part of the money. If the startup can't raise the rest, the lead is out too. How could they go ahead with the deal? The startup would be underfunded!

In the future, investors will increasingly be unable to offer investment subject to contingencies like other people investing. Or rather, investors who do that will get last place in line. Startups will go to them only to fill up rounds that are mostly subscribed. And since hot startups tend to have rounds that are oversubscribed, being last in line means they'll probably miss the hot deals. Hot deals and successful startups are not identical, but there is a significant correlation. [8] So investors who won't invest unilaterally will have lower returns.

Investors will probably find they do better when deprived of this crutch anyway. Chasing hot deals doesn't make investors choose better; it just makes them feel better about their choices. I've seen feeding frenzies both form and fall apart many times, and as far as I can tell they're mostly random. [9] If investors can no longer rely on their herd instincts, they'll have to think more about each startup before investing. They may be surprised how well this works.

Deadlock wasn't the only disadvantage of letting a lead investor manage an angel round. The investors would not infrequently collude to push down the valuation. And rounds took too long to close, because however motivated the lead was to get the round closed, he was not a tenth as motivated as the startup.

Increasingly, startups are taking charge of their own angel rounds. Only a few do so far, but I think we can already declare the old way dead, because those few are the best startups. They're the ones in a position to tell investors how the round is going to work. And if the startups you want to invest in do things a certain way, what difference does it make what the others do?

Traction #

In fact, it may be slightly misleading to say that angel rounds will increasingly take the place of series A rounds. What's really happening is that startup-controlled rounds are taking the place of investor-controlled rounds.

This is an instance of a very important meta-trend, one that Y Combinator itself has been based on from the beginning: founders are becoming increasingly powerful relative to investors. So if you want to predict what the future of venture funding will be like, just ask: how would founders like it to be? One by one, all the things founders dislike about raising money are going to get eliminated. [10]

Using that heuristic, I'll predict a couple more things. One is that investors will increasingly be unable to wait for startups to have "traction" before they put in significant money. It's hard to predict in advance which startups will succeed. So most investors prefer, if they can, to wait till the startup is already succeeding, then jump in quickly with an offer. Startups hate this as well, partly because it tends to create deadlock, and partly because it seems kind of slimy. If you're a promising startup but don't yet have significant growth, all the investors are your friends in words, but few are in actions. They all say they love you, but they all wait to invest. Then when you start to see growth, they claim they were your friend all along, and are aghast at the thought you'd be so disloyal as to leave them out of your round. If founders become more powerful, they'll be able to make investors give them more money upfront.

(The worst variant of this behavior is the tranched deal, where the investor makes a small initial investment, with more to follow if the startup does well. In effect, this structure gives the investor a free option on the next round, which they'll only take if it's worse for the startup than they could get in the open market. Tranched deals are an abuse. They're increasingly rare, and they're going to get rarer.) [11]

Investors don't like trying to predict which startups will succeed, but increasingly they'll have to. Though the way that happens won't necessarily be that the behavior of existing investors will change; it may instead be that they'll be replaced by other investors with different behavior — that investors who understand startups well enough to take on the hard problem of predicting their trajectory will tend to displace suits whose skills lie more in raising money from LPs.

Speed #

The other thing founders hate most about fundraising is how long it takes. So as founders become more powerful, rounds should start to close faster.

Fundraising is still terribly distracting for startups. If you're a founder in the middle of raising a round, the round is the top idea in your mind, which means working on the company isn't. If a round takes 2 months to close, which is reasonably fast by present standards, that means 2 months during which the company is basically treading water. That's the worst thing a startup could do.

So if investors want to get the best deals, the way to do it will be to close faster. Investors don't need weeks to make up their minds anyway. We decide based on about 10 minutes of reading an application plus 10 minutes of in person interview, and we only regret about 10% of our decisions. If we can decide in 20 minutes, surely the next round of investors can decide in a couple days. [12]

There are a lot of institutionalized delays in startup funding: the multi-week mating dance with investors; the distinction between termsheets and deals; the fact that each series A has enormously elaborate, custom paperwork. Both founders and investors tend to take these for granted. It's the way things have always been. But ultimately the reason these delays exist is that they're to the advantage of investors. More time gives investors more information about a startup's trajectory, and it also tends to make startups more pliable in negotiations, since they're usually short of money.

These conventions weren't designed to drag out the funding process, but that's why they're allowed to persist. Slowness is to the advantage of investors, who have in the past been the ones with the most power. But there is no need for rounds to take months or even weeks to close, and once founders realize that, it's going to stop. Not just in angel rounds, but in series A rounds too. The future is simple deals with standard terms, done quickly.

One minor abuse that will get corrected in the process is option pools. In a traditional series A round, before the VCs invest they make the company set aside a block of stock for future hires — usually between 10 and 30% of the company. The point is to ensure this dilution is borne by the existing shareholders. The practice isn't dishonest; founders know what's going on. But it makes deals unnecessarily complicated. In effect the valuation is 2 numbers. There's no need to keep doing this. [13]

The final thing founders want is to be able to sell some of their own stock in later rounds. This won't be a change, because the practice is now quite common. A lot of investors hated the idea, but the world hasn't exploded as a result, so it will happen more, and more openly.

Surprise #

I've talked here about a bunch of changes that will be forced on investors as founders become more powerful. Now the good news: investors may actually make more money as a result.

A couple days ago an interviewer asked me if founders having more power would be better or worse for the world. I was surprised, because I'd never considered that question. Better or worse, it's happening. But after a second's reflection, the answer seemed obvious. Founders understand their companies better than investors, and it has to be better if the people with more knowledge have more power.

One of the mistakes novice pilots make is overcontrolling the aircraft: applying corrections too vigorously, so the aircraft oscillates about the desired configuration instead of approaching it asymptotically. It seems probable that investors have till now on average been overcontrolling their portfolio companies. In a lot of startups, the biggest source of stress for the founders is not competitors but investors. Certainly it was for us at Viaweb. And this is not a new phenomenon: investors were James Watt's biggest problem too. If having less power prevents investors from overcontrolling startups, it should be better not just for founders but for investors too.

Investors may end up with less stock per startup, but startups will probably do better with founders more in control, and there will almost certainly be more of them. Investors all compete with one another for deals, but they aren't one another's main competitor. Our main competitor is employers. And so far that competitor is crushing us. Only a tiny fraction of people who could start a startup do. Nearly all customers choose the competing product, a job. Why? Well, let's look at the product we're offering. An unbiased review would go something like this:

Starting a startup gives you more freedom and the opportunity to make a lot more money than a job, but it's also hard work and at times very stressful.

Much of the stress comes from dealing with investors. If reforming the investment process removed that stress, we'd make our product much more attractive. The kind of people who make good startup founders don't mind dealing with technical problems — they enjoy technical problems — but they hate the type of problems investors cause.

Investors have no idea that when they maltreat one startup, they're preventing 10 others from happening, but they are. Indirectly, but they are. So when investors stop trying to squeeze a little more out of their existing deals, they'll find they're net ahead, because so many more new deals appear.

One of our axioms at Y Combinator is not to think of deal flow as a zero-sum game. Our main focus is to encourage more startups to happen, not to win a larger share of the existing stream. We've found this principle very useful, and we think as it spreads outward it will help later stage investors as well.

"Make something people want" applies to us too.

Notes #

In this essay I'm talking mainly about software startups. These points don't apply to types of startups that are still expensive to start, e.g. in energy or biotech. Even the cheap kinds of startups will generally raise large amounts at some point, when they want to hire a lot of people. What has changed is how much they can get done before that.

It's not the distribution of good startups that has a power law dropoff, but the distribution of potentially good startups, which is to say, good deals. There are lots of potential winners, from which a few actual winners emerge with superlinear certainty.

As I was writing this, I asked some founders who'd taken series A rounds from top VC funds whether it was worth it, and they unanimously said yes. The quality of investor is more important than the type of round, though. I'd take an angel round from good angels over a series A from a mediocre VC.

Founders also worry that taking an angel investment from a VC means they'll look bad if the VC declines to participate in the next round. The trend of VC angel investing is so new that it's hard to say how justified this worry is. Another danger, pointed out by Mitch Kapor, is that if VCs are only doing angel deals to generate series A deal flow, then their incentives aren't aligned with the founders'. The founders want the valuation of the next round to be high, and the VCs want it to be low. Again, hard to say yet how much of a problem this will be.

Josh Kopelman pointed out that another way to be on fewer boards at once is to take board seats for shorter periods.

Google was in this respect as so many others the pattern for the future. It would be great for VCs if the similarity extended to returns. That's probably too much to hope for, but the returns may be somewhat higher, as I explain later.

Doing a rolling close doesn't mean the company is always raising money. That would be a distraction. The point of a rolling close is to make fundraising take less time, not more. With a classic fixed sized round, you don't get any money till all the investors agree, and that often creates a situation where they all sit waiting for the others to act. A rolling close usually prevents this.

There are two (non-exclusive) causes of hot deals: the quality of the company, and domino effects among investors. The former is obviously a better predictor of success.

Some of the randomness is concealed by the fact that investment is a self fulfilling prophecy.

The shift in power to founders is exaggerated now because it's a seller's market. On the next downtick it will seem like I overstated the case. But on the next uptick after that, founders will seem more powerful than ever.

More generally, it will become less common for the same investor to invest in successive rounds, except when exercising an option to maintain their percentage. When the same investor invests in successive rounds, it often means the startup isn't getting market price. They may not care; they may prefer to work with an investor they already know; but as the investment market becomes more efficient, it will become increasingly easy to get market price if they want it. Which in turn means the investment community will tend to become more stratified.

The two 10 minuteses have 3 weeks between them so founders can get cheap plane tickets, but except for that they could be adjacent.

I'm not saying option pools themselves will go away. They're an administrative convenience. What will go away is investors requiring them.

Thanks to Sam Altman, John Bautista, Trevor Blackwell, Paul Buchheit, Jeff Clavier, Patrick Collison, Ron Conway, Matt Cohler, Chris Dixon, Mitch Kapor, Josh Kopelman, Pete Koomen, Carolynn Levy, Jessica Livingston, Ariel Poler, Geoff Ralston, Naval Ravikant, Dan Siroker, Harj Taggar, and Fred Wilson for reading drafts of this.