17: My Highlights on Snowflake & Unity IPO Filings, TikTok's Numbers from Court Filings, and How Fast Are You Spinning?

VPS: "Virtual Private Snowflake"

The major difference between a thing that might go wrong and a thing that cannot possibly go wrong is that when a thing that cannot possibly go wrong goes wrong, it usually turns out to be impossible to get at and repair. —Douglas Adams


Investing & Business

Timing is Everything: Unity Files S-1 for IPO

I’m sure that the people at Unity didn’t waste any time putting the finishing touches on their IPO papers when they saw the battle of titans between Apple and Epic (makers of the main competing 3D game engine, Unreal).

Their S-1 was filed earlier this week:

approximately 1.5 million monthly active creators [...] The applications developed by these creators were downloaded over three billion times per month in 2019 on over 1.5 billion unique devices. [...]

Improvements in computational power, greater connectivity and the proliferation of devices like smartphones, PCs and consoles have enabled an explosion of immersive and interactive content. The gaming industry has benefited enormously from these factors, with over 2.5 billion gamers driving the fastest growing sector in media today. [...]

The size of gaming is always kind of mind-blowing when you take the time to really think about it.

We estimate that in 2019, on a global basis, 53% of the top 1,000 mobile games on the Apple App Store and Google Play and over 50% of such mobile games, PC games and console games combined were made with Unity. [...]

We believe today we address a total market opportunity of approximately $29 billion across both gaming and other industries. Looking to the future, we believe there are large opportunities within and beyond the industries and use cases we currently serve that represent a market potential multiple times larger than our opportunity today. [...]

In 2019, global box office revenue for movies was estimated at $42.5 billions. In gaming, just the slice of the market where Unity plays (rendering engines and the various tools that surround them) is estimated at $29 billions.

Our revenue grew from $380.8 million to $541.8 million for the years ended December 31, 2018 and 2019, respectively, representing year-over-year growth of 42%, and from $252.8 million to $351.3 million for the six months ended June 30, 2019 and 2020, respectively, representing period-over-period growth of 39%. [...]

Of course, they had to talk about the value of the data they get from users:

We continuously capture and analyze valuable end-user behavior and application performance data from over 50 billion in-app events per day across over 20 different platforms as end-users interact with games and applications made with Unity. This data and analytics capability allows us to optimize content performance, end-user acquisition and engagement and monetization based on predicted lifetime values of our customers’ end-users, driving value for both our customers as well as their end-users.

They’re showing 79% gross margins for the first six months of 2020.

Spending 47.5% of revenue on R&D.

Ownership shows that Sequoia has 24.1% (they’re everywhere), Silver Lake Partners 18.2%. The CEO owns 4.4%.

This part of the CEO’s letter is interesting, about uses for their engine outside of gaming:

We’re a company born from gaming. But we see so much more. Five years ago, we believed that the real-time 3D tools we build would have applications outside of gaming. We saw that many industries were still wed to create tools that were conceived and built in the 1990s and early 2000s, before the explosive growth of compute power in the personal devices we all now carry with us most everywhere, and in the networks of computers that live in the cloud adding even more processing capability. We first started with an unproven thesis that we could serve customers outside of gaming. We signed up test customers working in augmented and virtual reality, architecture, construction, training, media and entertainment retail and in the auto/transportation industry. Small tests with Unity grew to larger customer relationships, and we started to see innovation from our customers we did not really anticipate. What started as an unproven thesis, transformed to being an opportunity where we needed to hone product market fit, and now, today, it seems inevitable. We’re thrilled to be helping lead the way to the future where much of the world’s content will be real-time and fully 3D.

In Practice has more on this:

Unity has 80-90% market share in AR/VR

Unreal gets a lot of press because it's sexy, but Unity seems to have a huge advantage in industrial applications [...]

But the big opportunity IMO is Unity in industrial applications

Unity's tech is lighter and far easier to adopt than [Unreal]

More details here (subscription required).

These are some highlights that jumped at me while skimming it. I didn’t read it with a fine-tooth comb or anything, so I may have missed interesting nuggets.

❄️ Snowflake S-1 ❄️

This is one that I didn’t know that much about going in, but it seems like interesting tech, so I want to learn.

If this is a company that you’re very familiar with and you want to share your opinion of it, don’t hesitate to reply to this email or DM me on Twitter, I’m curious to hear from people with direct experience of the products.

So on the origin story: Two of the three founders were at Oracle, and their experience there informed them to reboot the data warehousing/database architecture from the ground up for the cloud era. This isn’t a legacy on-prem tech stack that is being ported to the cloud, this is a fresh start, which has many advantages over much of the competition.

It’s also worth noting that the current CEO, Frank Slootman, was previously CEO of ServiceNow during much of its ascension, and also ran Data Domain to its IPO (acquired in 2009 by EMC, which became part of DELL in 2016).

Also, in the pointless trivia category, Slootman’s previous company’s ticker was NOW and this one is SNOW.


It’s interesting that Snowflake both partners and competes with the hyperscalers (AWS, Azure, GCP — though a “substantial majority” of is runs on AWS).

One innovation that helped make the company stand out was the separation of Storage and Compute (“pay only for what you use”) and the much better user experience (UX includes a lot of things, like UI, documentation, service/support, reliability, etc) than the competition. Also, unlike the hyperscalers’ offerings, it is fully aligned with being multi-cloud.

Anyway, enough with the preamble, let’s have a look at the S-1 (this is non-exhaustive, I’m just highlighting interesting stuff I find, but not going over everything in detail):

Our platform enables customers to consolidate data into a single source of truth to drive meaningful business insights, build data-driven applications, and share data. We deliver our platform through a customer-centric, consumption-based business model, only charging customers for the resources they use. [...]

From July 1, 2020 to July 31, 2020, we processed an average of 507 million daily queries across all of our customer accounts, up from an average of 254 million daily queries during the corresponding month of the prior fiscal year.

They point out that the number of queries doesn’t directly correlate with revenue, since some queries are more data and/or compute intensive than others.

We generate the substantial majority of our revenue from fees charged to our customers based on the storage, compute, and data transfer resources consumed on our platform as a single, integrated offering.

Basically how many terabytes stored, how much computing used on the data, and how many terabytes transferred. Intuitive stuff for customers.

Because it’s usage-based rather than flat rates, the more data customers move over and/or generate in their business, the more fees Snowflake should be able to charge. But like other cloud players, they’ll probably also face deflationary pressure on prices as storage and compute and bandwidth gets cheaper over time and the hyperscalers keep cutting prices per unit of storage/compute/etc.

As of July 31, 2020, we had 3,117 total customers, increasing from 948 and 2,392 as of January 31, 2019 and 2020, respectively. [...]

↗️ Growth growth growth:

For the fiscal years ended January 31, 2019 and 2020, our revenue was $96.7 million and $264.7 million, respectively, representing year-over-year growth of 174%. For the six months ended July 31, 2019 and 2020, our revenue was $104.0 million and $242.0 million, respectively, representing year-over-year growth of 133%

Net revenue retention rate was 158% as of July 31, 2020.

Gross margin were 62% during the last quarter, up from 53% a year ago.

They cite some IDC forecasts:

According to IDC, 49% of data will be stored in public cloud environments by 2025, an increase from approximately 30% today. Additionally, according to a 2019 IDC report, 90% of Global 1000 Organizations will have a multi-cloud management strategy by 2024. [...]

According to IDC, there will be 175 zettabytes of data by 2025, representing a CAGR of 27% from 33 zettabytes of data in 2018

With the general idea that everybody is becoming a “data consumer” and that lots of business value and insights are stored in all this data that needs to be managed and analyzed… but legacy data architectures don’t scale, things are siloed and hard to get to and analyze, it’s expensive and not optimized for the big data tasks you may need, etc.

Basically, they claim to have a better mouse trap to warehouse all your important and not-so-important-but-you-never-know 1s and 0s.

Like any good tech company, they claim network effects:

Our business benefits from powerful network effects […] The more customers adopt our platform, the more they can share data with, or receive data from, other Snowflake customers, partners, and data providers, enhancing the value of the insights delivered by our platform for all users.

The claim seems a bit too strong, but ease of sharing is a useful advantage of not having your data stuck on-prem.


We believe the addressable market opportunity for our Cloud Data Platform is approximately $81 billion as of January 31, 2020.

They have 2,037 employees operating across 19 countries.

CEO Frank Slootman owns 5.9% of the shares, previous CEO Robert Muglia (formerly of Microsoft) owns 3.3%.

The three co-founders are in the low-single digits, and Sequoia, Altimeter, Sutter Hill, Redpoint, ICONIQ all own pretty big slugs (Sutter hill is the biggest at 20.3%, followed by Altimeter at 14.8% — great investments at much smaller valuations).

There’s some interesting context and questions about Snowflake and its competitors here.

Oh, and I thought this was cute…

You can pick various tiers when you sign up, from “Standard” to “Enterprise” to the top one.. VPS (Virtual Private Snowflake).

(VPS is a common acronym in tech, it usually means “Virtual Private Server”)

If you want to dig a bit deeper in the product, here’s comparisons of Snowflake to its main competitors:

What TikTok Tells Us About Facebook & Court Filing Numbers

Some may read this and go “Yeah but Instagram did great anyway during that time and Facebook too”, but the point is more nuanced than that. Not that long ago it seemed like it was almost impossible to get real traction against the Facebook properties (and against Youtube too), as Snapchat was decelerating and Twitter was having trouble getting out of its own way.

But TikTok showed that it was possible to go from nothing to ginormous in a short period of time, and that there were still different approaches that could work well to grab a huge userbase (TikTok isn’t social in the way that Instagram is, and isn’t video-centric in the way that Youtube is).

The TikTok court filing for their lawsuit against the US federal government has some interesting numbers:

TikTok’s user base has grown at a rapid pace in the United States. By January 2018, TikTok had 11,262,970 U.S. monthly active users.

By February 2019, TikTok’s base more than doubled, with a total of 26,739,143 U.S. monthly active users.

By October of that same year, TikTok’s total number of U.S. monthly active users had climbed to 39,897,768.

And by June 2020, TikTok’s total number of U.S. monthly active users had soared to 91,937,040.

Today, based on quarterly usage, 100 million Americans use the TikTok application. [...]

it currently has approximately 50 million daily active users in the United States

I think that qualifies as hypergrowth… From 11 million users in the US to 100 million in a couple years…

But that was just in the United States. The Worldwide numbers are impressive too:

TikTok’s growth in the United States paralleled its expansion worldwide. By January 2018, TikTok had 54,793,729 global monthly active users.

By December of that year, TikTok had 271,188,301 global monthly active users. And one year later, in December 2019, TikTok had 507,552,660 global monthly active users.

As of July 2020, TikTok had 689,174,209 global monthly active users,
and by August 2020, TikTok surpassed two billion global downloads

Morbid curiosity almost makes me want to see Oracle try to steer that ship…

Heico Q2*

You know things are a little rough when they focus on "first nine month of Fiscal 2020" rather than actual calendar Q2….

"The Flight Support Group's net sales were $178.2 million in the third quarter of fiscal 2020, as compared to $320.0 million in the third quarter of fiscal 2019"

That’s -44.3% for FSG.

Electronic Technologies Group (ETG) is only down 2% in the quarter (-6% organic), though… Nice to have a large segment of the company (lower revenue, but higher margins than FSG) that is less correlated to Revenue Passenger Miles (RPM). Release here.

*I refuse to think in fiscal quarters… Ok, it’s fiscal Q3, but it’s still calendar Q2. Let’s get rid of daylight saving time first, and then offset fiscal years. Who’s with me?

How Startup Founders Get Their Family In

Interesting short thread by Patrick McKenzie (highlights mine):

Realized some people don’t know this:

It is fairly common for founders to give money as a gift to parents or other family members and have them immediately invest the money in a new company, at a very low valuation. This is not to fund the company but rather to assist family.

Example: in the US, you have something like a $10k annual limit for gifts after which you start eating into a $1M lifetime cap for gifts before gift/inheritance tax comes into play.

So you give your dad a check for $10k. He then immediately invests it at $1M valuation.

Numbers are pretty arbitrary here; the main thing the valuation is doing is holding up to scrutiny as “This could plausibly have been an arms length transaction” when literally nobody who looked at it would consider it an arms length transaction.

The effect of this is, if your business does well, your parents’ retirement is taken care of and the tax implications of that are a lot cheaper than you doing it post liquidity.

A more sophisticated strategy, which generally involves professional advisors heavily, is a trust.

Many entrepreneurs actual do a friends and family round to raise money, but many entrepreneurs are effectively the retirement plan administrator and funding source for at least some stakeholders. (Sometimes both at once!)

Science & Technology

How Fast Are You Spinning?

Source. Via Massimo.

The Electric Chair for Weeds (kind of…)

This is pretty cool, and greener than chemical alternatives.

In the 1890s, U.S. railroad companies struggled with what remains a problem for railroads across the world: weeds. The solution that 19th-century railroad engineers devised made use of a then-new technology—high-voltage electricity, which they discovered could zap troublesome vegetation overgrowing their tracks. [...]

One company trying to replace herbicides with electricity is RootWave, based in the U.K. [...] Unlike the electric weed-killing gear that’s long been sold in the United States, RootWave’s equipment runs at tens of kilohertz—a much higher frequency than the power mains.

This seems a bit like how electric cars have been around forever, but needed to be iterated on a bunch to compete in the modern word.

This brings two advantages. For one, it makes the equipment lighter, because the transformers required to raise the voltage to weed-zapping levels (thousands of volts) can be much smaller. It also makes the equipment safer, because higher frequencies pose less of a threat of electrocution. Should you accidentally touch a live electrode “you will get a burn,” says Diprose, but there is much less of a threat of causing cardiac arrest than there would be with a system that operated at 50 or 60 hertz.

RootWave has two systems, a hand-carried one operating at 5 kilowatts and a 20-kilowatt version carried by a tractor. [...]

But most importantly, compared to the chemical herbicides that are an arms-race with evolution:

Of the many advantages CNH touts for its weed-electrocution system … is “No specific resistance expectable.

The Rootwave website is here, if you want to know more.

Source: IEEE Spectrum, via NZS Capital.

The Arts

Map of Canada in the Style of Super Mario World

Not saying it’s accurate, especially for my neck of the woods, but it is cool. Source.

It may not mean anything for those who have never played Super Mario World on the Super Nintendo, though. If you’re curious, here’s the reference they’re making.


My Neighbor Totoro (1988)

After 'Ponyo' (2008) was a hit a few days ago, I decided to watch 'My Neighbor Totoro' (1988, Miyazaki) for the first time with my boys (2 & 6yo).

Another big hit. They were captivated despite the slower pacing. Superb artistry, very subtle understanding of the psychology of kids. Recommended for all ages.