The acquisition of HashiCorp ($HCP) by IBM is being finalized this week, with a liquidation price of $35 per share.
-
One of the most significant changes in developer productivity over the past decade has been the Cloud.
-
As of last quarter, the Big 3 cloud providers generated $234.4B in annual revenue (on a run-rate basis, and this isn’t purely cloud infrastructure costs).
-
Considering Amazon only started reporting AWS revenue about 10 years ago, barely hitting $1B in Q1 2014, the growth over the past decade is astonishing.
-
Companies that previously operated their own servers could physically see the machines their programs ran on, but the cloud “virtualized” these, freeing them from physical constraints.
-
When owning and using machines, each one was like a pet. You knew each one by name and could even physically inspect it if a problem arose.
-
In the cloud, machines are more like cattle. Each has a unique identifier, but if one has a problem, it’s replaced with another machine. Here, <Pets vs. Cattle> is a commonly used analogy in the industry.
-
There’s no need to always operate the same number of servers. Because it’s a pay-per-usage rental model, it’s also possible to save costs by reducing the number of servers when traffic is low.
-
When servers become ephemeral and elastic like this, it opens up the possibility of increasing productivity in software operations in a completely different way.
-
It’s not just an improvement in the productivity of existing methods; by effectively utilizing paradigm-shifting technologies, productivity can be significantly increased. Docker, k8s (Kubernetes), and Terraform are major examples.
-
To leverage these new environments and technologies, organizations within companies named DevOps, SRE (Site Reliability Engineering), and Platform Engineering have emerged. These teams are more deeply involved in the entire product lifecycle than traditional IT departments and help improve corporate productivity.
-
Nowadays, with the increasing need to run AI/ML on top of that, names like MLOps and LLMOps have also emerged, but I believe the essence is holistic software lifecycle management - from development and testing through deployment and operations — combined with effective cloud utilization and complexity management. The last ten years have brought substantial progress. As a result, cloud-inspired approaches are now significantly boosting productivity even in on-premise environments that don’t rely on cloud services.
-
HashiCorp, founded around 2013 by two University of Washington undergraduate alumni, was a significant company leading such a paradigm. Their published methodology, <The Tao of HashiCorp>, is particularly worth reading for those in the field.
-
One of the most significant directions among these is Infrastructure-as-Code (IaC). This involves declaratively defining and operating the intended state (intention) of what a server should look like. While the traditional procedural approach of operating physical servers focused on how servers should run, the new approach allows for a focus on what.
-
They weren’t the first to create such innovations. The advantages of declarative SQL have been well-known since the 1970s, and similar tools like Ansible or Puppet also share the same foundation.
-
HashiCorp’s differentiating factor was its cloud-native and vendor-neutral strategy. Instead of trying to solve all problems, they focused on the cloud, but in a way that wasn’t dependent on a single vendor like AWS.
-
As mentioned above, as the cloud grew tremendously and competition among the Big 3 intensified, this position became very powerful. HashiCorp further capitalized on this position, continuously developing products necessary for overall platform engineering, such as Vault and Consul.
-
Aided by the tailwinds of the cloud and rapid growth during the COVID period, HashiCorp completed its IPO in December 2021 at $80 per share, hitting a peak of $102.95 by year-end. In its S-1 filing, the Total Addressable Market (TAM) for 2026 was estimated at $73B. At that time, HashiCorp’s annual revenue was $259M on a run-rate basis, resulting in a P/S ratio of over 50, but its high annual growth rate of 75% justified this high valuation.
-
Anyone interested in finance can marvel at their fantastic timing. Three months later, the Fed began raising interest rates, bringing an end to the zero interest rate policy (ZIRP) era, and tech IPOs disappeared for a while.
-
As companies also began to cut costs, FinOps became important, and maintaining high revenue growth rates became difficult. HashiCorp’s annual growth rate dropped to 15%, and its stock price also hit a low of $19.29 in November 2023. This was significantly lower not only than the IPO price but also than the Series E price of $28.92 in April 2020.
-
With annual revenue still at only $0.5B and no profit, the growth slowdown was a difficult problem to endure. Since there’s hardly any cost of goods sold for their products, the Non-GAAP gross margin reached 85%. However, continued investment in R&D and S&M was still necessary, operating expenses (G&A) couldn’t be ignored, and non-cash stock-based compensation also constituted a large portion, meaning the path to profitability was still long.
-
Ultimately, on April 24, 2024, they agreed to a merger with IBM at $35 per share, and the merger was completed last week. HashiCorp had a dual-class share structure, common in the industry, where a few founders and investors held over half the voting rights, but naturally, minority shareholders were also bought out at the same price.
-
I think this is a good case study for thinking about what constitutes the value of a software company. Their main products are open source and accessible to anyone. However, customers still pay them for services. This suggests that code is not the entirety of a company’s value.
-
Nevertheless, their struggle to thrive independently may have stemmed from targeting too broad a problem domain. Instead of using HashiCorp, it became a rational choice for individual companies to invest in their own platform engineering capabilities and operate them internally.
-
A desperate HashiCorp tried changing its code license from the Mozilla Public License (MPL) to the Business Source License (BUSL) and attempted to restrict other companies from using its code to conduct business, but the Linux Foundation immediately forked it into OpenTofu.
-
In contrast, companies like MongoDB ($MDB), GitLab ($GTLB), and Elastic ($ESTC) were able to switch to open-source licenses that were more favorable for business. When you think about it, when using a product like Slack, you don’t really question whether its code is open. Being open source doesn’t necessarily make it difficult to do business; depending on the problem, it can be an effective strategy or a constraint.
-
HashiCorp was still a valuable company. One reason the stock price didn’t fall further was that besides IBM, there were quite a few other potential buyers. In the end, IBM, which owns RedHat and focuses on cloud operations, seems like an appropriate acquirer.
As I wrote in my previous post <Macro Dev Wave and Career>, big waves in the software industry open and close various opportunities. This is true not only for those directly running businesses but also for employees and investors.