Why one Databricks investor thinks the company may be undervalued

The recent Databricks funding round, a $1 billion investment at a $28 billion valuation, was one of the year’s most notable private investments so far.

For Databricks signaled its IPO readiness by disclosing to TechCrunch last year that it had scaled its revenue run rate from $200 million to $350 million in a year, so the new capital looked like the capstone on its private fundraising before an eventual public debut.

The Exchange explores startups, markets and money. Read it every morning on Extra Crunch, or get The Exchange newsletter every Saturday.

But I did have a few questions, starting with the price of the round.

At a $28 billion valuation and ARR of $425 million, Databricks is valued at around 66x top line. That’s steep, if not the highest number we can dredge up on the public markets. Of course, for Databricks shareholders, seeing the value of their stock rise so quickly is hardly a bad thing. They are hardly going to complain about having more paper wealth.

But what about the investor perspective? Does the price really make sense? The Exchange caught up with Battery Ventures’ Dharmesh Thakker earlier this week to discuss a number of things, one of which was Databricks’ round and pricing. Thakker is named in the Databricks Series D funding announcement, which brought Battery into the company.

What was surprising about our conversation was not that Thakker was bullish on Databricks — a company that he and his firm have backed since its $140 million, 2017 round when the company was worth just under $1 billion. What surprised me was that he thinks its new $28 billion valuation might be a little low.

Intriguing, yeah? So this morning for both of us, I’ve pulled out quotes from our chat to help explain how Thakker views the market for Databricks, unicorns at scale more broadly through the lens of risk-adjusted investing, and the scale of the market some unicorns are playing in.

At the close, we’ll remind ourselves what Databricks CEO Ali Ghodsi told TechCrunch when we asked him the same question. Let’s go!

Databricks at $28 billion

Here’s how the valuation part of my chat with the Battery Ventures’ investor went down:

The Exchange: I want to talk about Databricks, because I spoke to [CEO] Ali [Ghodsi] yesterday about this round, and hot damn, it’s a lot of money at a valuation that is roughly 64x ARR, give or take. I don’t understand the price, and I know it’s a boring thing to talk about. [It’s a] great company, I get their market, I’ve talked to them a bunch, I know their revenue numbers. [But] I don’t understand the price, and I was hoping you could tell me why I’m being too conservative.

Dharmesh Thakker:  I, for what it’s worth, think [the price] fair. If anything, I think it is on the lower end — he could have done better, frankly. But I think it comes down to three major things, right?

One is the addressable market. Just think about the addressable market of data. If there’s a trillion dollars spent in software or technology, I think you and I would be both hard pressed to say, almost all of that [isn’t] influenced by some data-oriented decisioning. Whether it’s digital transformation, whether it’s analytics, data is everywhere. So the TAM is massive … I think you and I both agree on that, whether it is $20 billion or $80 billion — it’s massive.

And second, I say there’s a new era that Databricks, Snowflake and to some extent Mongo[DB] have created. This is the ascent of the billion-dollar B2B companies that grow very rapidly, and they are growing very efficiently, which will [lead to] operating margins at scale.

So, TAM, growth rate and efficiency all translate to a business that can grow to a billion dollars and beyond on revenue within four to five years, which was unheard of. It took a company five years to get to $100 million and close to 10 years to get to a billion. Now obviously these companies get to a billion within four to five years of selling their first dollar of software, which is incredible.

But secondly, these companies look a lot more like consumer companies, where you don’t have expensive sales motions to sell to CIOs. You sell to practitioners, practitioners adopt [the product] in a grassroots way, then they bring it up to the higher-ups. What that means is that you’re selling it very, very efficiently.

As you grow from $100 million to a billion, your sales and marketing as a percentage of revenue drops very quickly. It used to be like 70%, but now it’s at 40%, 50%. R&D is flat, you don’t have to spend more on R&D just because you have an extra 100 customers. So R&D drops to like 20% of revenue and GNA is pretty low. So, what you see is in a business operating in an $80 billion TAM, you have a billion-dollar business going to five billion and operating margins going from negative 20% to positive 30%, 40%, which we’ve never seen before.

You are either a business [that is] growing very fast, but they burn a ton of money because they have to hire salespeople, and [service] these customers. Now, you build the software, you throw it out there, people try it, and then you grow very efficiently.

So now imagine what is a billion-dollar business going to five billion with 30%, 40% operating margins — that’s like, $2 billion in free cash flow. What is that worth? [TechCrunch: A bajillion dollars] If you told me that is worth 13 times free cash flow? I’d put money in that thing.

At this juncture I joked that we’d swapped positions, as earlier in the chat Thakker had discussed the early-stage startup prices and why paying a fair amount matters, while your humble servant had argued that given how large success in software can be, overpaying for early-stage shares didn’t seem so dangerous. He responded:

Dharmesh Thakker: On a risk-adjusted basis, at the Databricks’ scale, there’s maybe five companies that have got to that scale that efficient[ly]: Snowflake, Databricks, Zoom, MongoDB’s Atlas product. There’s a handful of companies.

So the odds of them, it’s kinda like, you know, Apple took 20 years to go to the first trillion. [And] going from $1 trillion to $2 trillion was like in six weeks, right? So you get to a certain scale, you get so much gravity around you, that on a risk-adjusted basis, you can grow 50, 60-plus percent very easily for a while. So going from $425 million ARR, which you reported — I won’t confirm or deny it — to a billion, to two billion. It’s much easier on a risk-adjusted basis for Databricks versus a startup, which has a one and five-hundred shot of getting [that] successful.

So I don’t want to overpay for a startup. [Because] there’s a 2% chance it will be successful. And if I’m paying you 20x your forward revenue, I better have more odds than 2%. [Whereas] in Databricks, I’m like there’s a 50%+ chance you’ll be successful. I’m paying a fair price right now, but I’m [planning] on you growing and continuing to do what you do. On a risk-adjusted basis I think it’s a it’s a better place to be.

Vetting the logic

Honestly, I kinda get it. The only real issue with Thakker’s take is what might happen if the broader market repriced all software revenures negatively. If that happened, then Databricks could have a bear of a time getting its worth to new peaks.

But when I spoke to its CEO, he said that if that happened it would only a take a year’s growth to sort out. Which is mostly true by my own math, provided that the markets only fell so much. The good news for Databricks is that if it can just keep growing at a high clip this year, all things will sort out for it, and its latest valuation will appear cheap when it finally does go public.