Take the down round

The greatest trick that VCs ever pulled is convincing startups that down rounds are a disaster. Plus, pricing, again.

Hello from Keyser Söze Venture Partners!

A brief programming note: I started this blog a couple years ago mostly for my own entertainment. Though my day job at Mode gave me plenty of opportunities to think about data and technology, the work was necessarily pragmatic, about immediate customer needs and practical product choices. Inside of a startup, there's not much time for idle curiosities. 

This blog was a side project for exactly that—a hobby, for fun. There wasn’t any motive beyond that, and there still isn't. For better or for worse, I’m here for the fun of it.

It’s still a good time, and I appreciate all of you who've stopped by. But over the last several months, the structure—one long rant, about data pipelines, or AI, or the same half-dozen companies, or whatever—has started to feel tired. So, I might try out some new things. There might be less on tooling, and more on data in the real world. There might be the occasional longer essay, because the problem with 2,000 word posts about esoteric Silicon Valley minutiae is that they’re not 5,000 words. I might become a cliché, and start a podcast. I might realize that I only have one move, and keep regurgitating the same three points over and over again with more and more strained analogies. I might spend a few days in upstate New York, fall in love with the mountains and hills, become a farmer, a man of the land, with only his animals and till, and never write another word. I might finally write that piece about Pitbull.

Still, inevitably, every comedian tries to be a dramatic actor. Every TikToker tries to be a pop star. And every niche blogger convinces themselves they can be Matt Levine. So this week, in a blatant stylistic and substantive ripoff of Money Stuff, there’s not one big topic; there’s a couple sections about data and money. Years from now, let’s see if I regret writing it.¹ 

---

Up with the down round

For all that Silicon Valley tries to celebrate failing, it sure punishes struggling. 

For a startup, the ultimate sign of struggling is the dreaded down round. A down round, for those who are fortunate enough to have never been acquainted with one, is when a startup raises money that values the company less than it had been valued in a previous round.² 

Suppose, for example, that a company raises $4 million by selling 20 percent of itself to venture capitalists, valuing the entire company at $20 million. It then spends the money on employees’ salaries and marketing programs and Nvidia GPUs.³ If it isn’t profitable by the time that money is gone, the startup will need to raise more money. So it goes to venture capitalists and asks for another round of funding. 

The expectation is to raise that money at a higher valuation in what’s called an up round—by, say, selling another 20 percent of the company for $8 million, implying that the company is now worth $40 million. But for a variety of reasons—the business is struggling, a change in market conditions, the founder will be in jail for the next eleven years—the company may not find any investor willing to buy shares at that price, or at any price above $40 million. In that case, the startup, which needs money to continue operating, may have to sell 50 percent of itself to raise $8 million, which would value the entire company at $16 million. 

That’s a down round. And in the zeitgeist of Silicon Valley, they’re a disaster. They’re humbling comedowns; big falls from grace; catastrophic events that are “the worst possible thing that can happen outside of a tragic accident of some sort.” 

For early stage companies, this pessimism makes sense. If two founders and an idea are worth $20 million to some angel investors, it’s a pretty bad sign if those same two founders and that same idea, plus eighteen months of work and four million dollars spent, is only worth $16 million. 

As startups get bigger and more mature, however, their valuations are more and more affected by public markets—markets that, quite reasonably, move up and down over cycles that last months and years. For companies caught raising across these cycles, the stigma around down rounds look a lot less like a reflection of an actual disaster, and a lot more like a VC-created boogeyman. 

—

There are two ways for a startup to raise a funding round that values it at one billion dollars:

• It raises a round that values the company at less than a billion dollars—say, $500 million. It then spends that money to continue building its products and businesses. After a couple years of disciplined growth and fiscally restrained management, it grows its annual revenues from $50 million to $100 million, and raises another round that values the company at one billion dollars. 

• It raises a round that values the company at more than a billion dollars—say, $1.5 billion. It then spends that money on drugs and Run-D.M.C. concerts—and some useful business initiatives that grow its revenue from $50 million to $100 million, albeit very inefficiently. After a couple years, it makes somber promises to be more responsible, offers to sell new shares at a discount from the previous price, and raises a down round that values the company at one billion dollars. 

The first path is “normal” and “correct.” It’s the path that venture-backed startups are both supposed to take, and supposed to want to take. The second path is everything that’s wrong with Silicon Valley—a company gets caught up in some frothy hype cycle, raises money it doesn’t deserve at a price it can’t justify, and crashes back to earth when the price comes down. 

But in both cases, the company is in the same spot—worth $1 billion, with $100 million in revenue and some new money in the back. So why all the catastrophizing over down rounds? I think there are three reasons—one that’s legitimate but should be minor; one that’s misleading; and one that, cynically, explains what’s really going on. 

The legitimate reason is that down rounds harm employees. When an employee joins a startup, they’re typically granted stock options that they can purchase at roughly the same price that the last investor paid for them. Employees that joined the company between between two fundraising rounds would have their options priced based on the first round, and then valued—on paper, at least—based on the second round. A lot of people join startups because of the potential value of their equity, and that potential is a lot harder to get excited about when you have to pay a dollar to buy a share that’s currently worth seventy cents. 

That said, this can get at least partially corrected. Companies can reprice or exchange options so that employees can purchase theirs at the new price. Though these aren’t complete solutions, they can mitigate a lot of the impact. Moreover, employees are a lot more affected by layoffs than down rounds. Layoffs are often seen as healthy corrections, whereas down rounds are “the worst possible thing,” which betrays the argument that the cultural shame in down rounds is primarily about their damaging effect on employees.

The second reason that we’re warned about down rounds is, I think, a kind of logical sleight of hand. When people talk about down rounds being punitive and dilutive, they tend to talk about them relative to the prior round or a hypothetical up round. If a company raised a round at a $1.5 billion valuation, raising the next round a $1 billion valuation probably dilutes existing shareholders more than the first round, and way more than an up round would.

But that strikes me as the wrong comparison. Excluding the occasional criminally delusional startup, most companies are going to do the best they can to be successful, regardless of how much money they raise. A startup’s growth trajectory, from its founding to the point of standing on a precipice of taking a down round, likely wouldn’t have been higher had it taken less money.⁴ One way or another, it was going to reach this point, and be about to raise at roughly this valuation. The question to ask, then, is not if a down round is dilutive relative to some hypothetical $2 billion valuation that was never attainable; the question to ask is if how much more dilutive is it to have passed through a $1.5 billion valuation on the way to a $1 billion valuation than it is to have passed through a $500 million valuation?

The exact answer to that depends on a lot of complex and variable terms in how that intermediate round was raised, but the general answer is, it doesn’t really matter that much. Across a number of different permutations—if the first round was small or large; if it valued the company at under a billion dollars, at just over a billion dollars, or at a lot more than a billion dollars; if the round had structure⁵ or was clean—founders are roughly financially indifferent. What matters far more than raising a down round is what happens after the down round: The only scenarios in which down rounds are particularly punitive are those in which the startup’s valuation continues to fall.⁶ And the risk of taking a bit more dilution might be a perfectly reasonable price to pay for the opportunity that a larger war chest of cash can provide. 

So what’s the real reason for the outsized fearmongering about down rounds? Cynically, I’d argue it’s because down rounds are bad for VCs, and companies being scared of them is good for VCs.

—

On the surface, that’s just math. Raising rounds at high valuations is expensive for VCs; moreover, at a given valuation, ownership of a startup is zero sum. The more founders and employees own, the less investors own. A startup ecosystem that emphasizes discipline and encourages founders to be more conservative in how they raise helps VCs fund companies at lower prices than one that aggressively chases big valuations and rapid growth.

But I think that’s actually backwards. VCs don’t make money by buying shares of slow growing startups at a discount; they make money by investing in really fast growing companies, at almost any price. An ecosystem full of cautious companies is no good; they want one full of companies that are just gonna send it. 

Stigmatizing down rounds protects that. It’s the side-scrolling left wall of Silicon Valley, always pushing startups forward to chase something bigger. 

The problem is that there are times when it’s not obvious if a company should chase something bigger. Its product may be hitting its plateau; its immediate market may be tapped out, without a good one to turn to next. Unfortunately, in these moments, the incentives of VCs and startups aren’t aligned. Because, for VCs, their portfolios are exactly that—portfolios. Famously, the “biggest secret in venture capital is that the best investment in a successful fund equals or outperforms the entire rest of the fund combined.” Better to have every company swing for the fences, have most strike out and have one hit a grand slam⁷ than to have them all chasing singles.⁸ 

But founders and employees don’t make money off of an index. Their prospects are tied to one at bat. For them, a single could change their lives, and a strikeout could ruin it. When they’re down in the count, they may be better off choking up and looking to make contact.

My point here is not that companies should carelessly chase hype, or that big bad VCs are conspiring against meek and righteous startups. Companies have a lot of say in these decisions, and the collective pressure to chase big outcomes is both a faceless cultural force, and exactly what makes Silicon Valley as a whole work. Instead, my point is that what’s good for the ecosystem may not be good for individual companies within it. The biggest problem with a down round is the optics of a down round. Those optics push companies to accept punitive terms to avoid them, and to keep chasing goals that are drifting out of reach. The sooner we recognize that, the sooner we can soften the stigma—and the sooner companies can do what’s best for them, and take the down round.

---

How much for that DAG compute engine in the window?

The drama of the week is pricing, again. This time, it’s about Snowflake, and how much Instacart spends on it. 

To quickly recap, Instacart filed to go public last week; in their S-1,⁹ which is essentially a giant disclosure document about the goings-on of the business, they said they used to spend a bunch of money on Snowflake and now they spend a lot less; people started asking what happened and what it meant; it was clearly bad for Snowflake, and Instacart, and Databricks; no, it was clearly good for Snowflake, and Instacart, and Databricks; Snowflake tried to explain their side of the story; and now we all believe whatever we believed before this happened, but even more so. Gergely Orosz has the story.

I have no idea what to make of it, but it doesn’t strike me as a terribly useful signal either way. Instacart is a business that is uniquely data-dependent: They have to optimize a three-sided marketplace, solve 263 million traveling salesmen problems a year, and manage a $740 million ads business. I don’t know how much that business should spend on Snowflake, but I know that their data needs are more unique than most. They’re in data’s one percent; to me, the more interesting question is about what’s happening inside of the 99 percent.

That said, all of the recent fuss about pricing—and in particular, usage-based pricing—does make me wonder if we’ll ever see a shift back towards perpetual licensing. Surely, someone will try to differentiate by coming out with a cloud product that promises full price transparency, with a single, lifetime fee of $5,000 per user. Surely, if there’s a product pendulum that constantly swings back and forth between bundling and unbundling, there must be a pricing and packaging pendulum too; surely, someone will go back to shipping software in a box.

If nothing else, I do think there’d be one advantage of buying over leasing: It encourages us to stop shopping, and invest in what we have. 

In keeping with the tradition of strained analogies, tools are like relationships. A lot could be good, but none are perfect. To make them really work, we have to both figure them out, and we have to adapt ourselves around them. If we work through that process, they can be what we dream of them to be. But if we never need to commit, or are afraid to—if we’re scared of heights, scared of love; if we run away; if we put our past problems on the new thing¹⁰—it’ll always be insufficient, and we’ll never find something that’s as good as what we could have if we just made a decision and stuck to it.

1. There are 243 words in this song’s chorus. Three of them—night, night, and night—are more than four letters long.

2. I have plenty of personal grudges about various things in Silicon Valley, but this isn’t one of them. Mode’s never took a down round, nor did we ever take some huge up round that made a down round hard to avoid. This was one of the rare benefits to raising money in the middle of a pandemic-induced economic collapse.

3. Payroll $470k
AWS $130k
Rent $40k
Nvidia GPUs $3.3m
Software $60k
someone who is good at the economy please help me budget this. my startup is dying

4. No doubt, some people will say that less money makes you smarter, that constraints breed creativity, and so on. Which, yes, sure, but that’s primarily about efficiency, not absolute scale—and startups valuations are based more on the latter than the former. In other words, it’s one thing to say that less is more efficient; it’s another to say that less is outright more. 

5. “Structure” refers to terms in fundraising agreements that aren’t just about the amount of money raised and at what valuation. Typically, these are provisions for investors that grant them additional rights, like the ability to participate in future fundraising rounds, veto power over major company decisions, and guarantees that they’ll get paid back for their investments before anyone else makes any money. Some fundraising agreements include terms that explicitly protect investors from down rounds; these are called anti-dilution provisions.

6. The table shows how much of a startup’s founders and employees would in different fundraising scenarios. In each case, the company raised two rounds of funding, and the second round was always the same—$100 million at a $1 billion valuation. I also assumed that the first round had a 1x liquidation preference, which is fairly typical, and broad-based weighted-average anti-dilution protection, which is less common.

I then looked at how exit payouts for founders and employees would be affected by the size of the first round. The first row shows a scenario for which the second round is an up round; the second and third rows show scenarios for which it would be a down round. As the table shows, the only case in which the down round is particularly costly is the one in which the company raised a lot of money and its exit valuation is less than $1 billion. In other cases, the down round doesn’t have much effect—and can actually be beneficial.

(It’s also worth noting that this is a severe oversimplification of how all this actually works, but the point generally holds for more complex cases.)

7. Live video of a venture capitalist watching said grand slam. 

8. Also of note: For founders and employees, it doesn’t matter if equity prices fluctuate between when they join a company and when it exits. VCs, however, have to regularly report on the value of their portfolios, and have to report losses after a down round. That creates an incentive for them to keep intermediate valuations high that founders and employees don’t have.

9. Ok, but why are we talking about Instacart’s Snowflake spend and not about its legal name being Maplebear, Inc.?

10. It’s out and it’s a banger.