Wednesday, December 28, 2016

GitHub financials discussion

One challenge with a company like Uber or GitHub is that they are private, so they don't have to release their financials, which can make it a challenge to understand how well (or poorly) they are doing.

Every so often there are clues, though.

A few weeks ago, Bloomberg published this article by Eric Newcomer: GitHub Is Building a Coder’s Paradise. It’s Not Coming Cheap.

Newcomer's lede is surprisingly harsh; the first paragraph reads:

Though the name GitHub is practically unknown outside technology circles, coders around the world have embraced the software. The startup operates a sort of Google Docs for programmers, giving them a place to store, share and collaborate on their work. But GitHub Inc. is losing money through profligate spending and has stood by as new entrants emerged in a software category it essentially gave birth to, according to people familiar with the business and financial paperwork reviewed by Bloomberg.

"Profligate spending?" "Stood by as new entrants emerged?" These are strong words, indeed.

Newcomer also takes the expected swing at the management scandals at GitHub, and even drops a rumor of a recent layoff.

As I said, it's hard to know what to make of such an article, given that the company is private and we just have to take people's word for things. Still, this article definitely smacked of being a bit of a hit piece, particularly since it spoke so glowingly of GitHub's upstart competitor, GitLab:

The issue took on a new sense of urgency in 2014 with the formation of a rival startup with a similar name. GitLab Inc. went after large businesses from the start, offering them a cheaper alternative to GitHub. “The big differentiator for GitLab is that it was designed for the enterprise, and GitHub was not,” says GitLab CEO Sid Sijbrandij. “One of the values is frugality, and this is something very close to our heart. We want to treat our team members really well, but we don’t want to waste any money where it’s not needed. So we don’t have a big fancy office because we can be effective without it.”

Well, anyway, a few weeks passed, and, perhaps not surprisingly, there is now a counter-part article. Medium published this article by Moritz Plassnig: GitHub is Doing Much Better Than Bloomberg Thinks - Here is Why.

Plassnig criticizes Newcomer, justifiably I think, for focusing too much on 2014, and not enough on 2016:

GitHub was the darling of the developer community but 2014 (harassment scandal, Tom Preston-Werner resigning) and 2015 (slower progress, increased competition) were challenging and it suddenly wasn’t set in stone anymore that GitHub will dominate and own their vertical in the same way as for example Amazon Webservices owns theirs.

The September 2015 ARR number of $90M seems to reflect that. But, if they were struggling in 2015, they blew it out of the water in 2016 and went from $90M in September 2015 to $140M in August 2016.

And, perhaps more importantly, Plassnig works to try to redirect attention away from GitHub's free hosted services and toward their booming enterprise business:

GitHub offers three different products as of December 2016:
  • personal plan
  • organizational plan
  • GitHub Enterprise (on-premise/VPC product)
All of the growth in the last two years came from GitHub’s organization plans or GitHub Enterprise. The revenue from their organization plans roughly doubled and the revenue of GitHub Enterprise tripled over the course of 23 months (Sep’14 — Aug’16) while the revenue of their personal plans stagnated according to the numbers Bloomberg published.

50% of GitHub’s ARR came from GitHub Enterprise as of August 2016 compared to 35% back in September 2014. Their efforts to get into larger organizations and become more of a traditional enterprise software vendor also explains the higher burn rate. Historically, most of GitHub’s revenue came from their offering which was completely self-serve while their GitHub Enterprise customers require more handholding and a more traditional enterprise sales process.

Those are, indeed, remarkable numbers.

Newcomer, too, is aware of the dramatic size and spectacular growth of the GitHub Enterprise numbers, although he chooses to frame it a different way, as a core tension in the company rather than an engine of growth:

GitHub says it has 18 million users, and its Enterprise service is used by half of the world’s 10 highest-grossing companies, including Wal-Mart Stores Inc. and Ford Motor Co.

Some longtime GitHub fans weren’t happy with the new direction, though. More than 1,800 developers signed an online petition, saying: “Those of us who run some of the most popular projects on GitHub feel completely ignored by you.”

The backlash was a wake-up call, Wanstrath says. GitHub is now more focused on its original mission of catering to coders, he says. “I want us to be judged on, ‘Are we making developers more productive?’” he says.

It's not obvious to me that being "focused on its original mission of catering to coders" is at all counter to the strategy of making money by selling GitHub Enterprise; as far as I know, the features that are made available to hobbyists and open source teams on hosted GitHub are also present in GitHub Enterprise, so improvements to, e.g., the code review process benefit all.

To me, the more telling and fascinating finding, which neither Newcomer nor Plassnig seem to discuss much at all, is the counter-industry-trends success of GitHub Enterprise.

In this era of the cloud, when company after company is moving everything possible onto Amazon Web Services or Microsoft Azure, the fact that so many organizations are still choosing to make massive deployments of SCM services inside their own corporate data center is astonishing to me.

I'm not sure what is driving this particular aspect of the business, although I have a few theories.

One theory is that the IP stored in an SCM system is still viewed as the most precious, private, and critical IP within an organization, and many organizations still aren't willing to trust the cloud with these materials.

Another, perhaps more likely, theory, is that deployments like GitHub Enterprise are displacing older SCM systems within organizations, and those older SCM systems are almost always deployed in the corporation's main internal data center, so it is natural to replace like with like.

Finally, inside large corporations, SCM systems are rarely used "out of the box" as-is; rather, they are foundational pieces of infrastructure, with considerable flexibility and extensibility, and they are typically deployed as a base atop which the organization builds their own proprietary workflow and development tools, integrates with their own proprietary security system, etc.

In such a configuration, you may well need to have near-total control over the SCM deployment inside your own data center in order to port your existing development toolchain to it and extend that tooling over time.

This last argument, while it might well be the basis of the current choice to deploy tools like GitHub in their on-premise "enterprise" editions, seems like it has nothing fundamental to it in the long run, so I still think that, eventually, most organizations, even giant ones, will find themselves running their SCM services in the cloud.

It's just that, as these current GitHub financials appear to show, we aren't there just yet.

No comments:

Post a Comment