Armory
Winter 2017AcquiredContinuous Deployment at any scale, for all developers.
Pressure-test this opportunity
Explore the risks and possibilities with a prompt for ChatGPT, Claude, or your agent.
Continuous Deployment at any scale, for all developers.
Explore the risks and possibilities with a prompt for ChatGPT, Claude, or your agent.
Armory commercialized Spinnaker, the open-source continuous-delivery system created at Netflix and Google. It gave large companies a supported way to coordinate releases across clouds, apply security and approval policies, expose new versions gradually, measure their health, and reverse a bad deployment. The company raised more than $82 million and served customers including JPMorgan Chase, Autodesk, Informatica, Patreon, LaunchDarkly, and HelloSign.[1]
The venture outcome was poor. Harness acquired Armory's assets for about $7 million in cash in January 2024. Harness hired many employees and kept supporting existing installations, but its chief executive described the transaction as an asset deal. Armory had raised almost twelve times the reported purchase price.[2]
Armory's core insight was sound: software deployment is a separate operational problem from compiling and testing code. Its challenge was turning one stage of the development lifecycle into a large independent company. Spinnaker was powerful but expensive to operate. Simpler open-source tools improved. Integrated DevOps platforms expanded into deployment. Armory introduced a new hosted product, but only reached public early access in 2022 after discussing a SaaS version in 2020. When venture funding tightened, a point solution with heavy enterprise demands had limited room to maneuver.
Daniel R. Odio, Isaac Mosquera, and Ben Mappen had worked together before founding Armory in a Belmont, California, garage in late 2016.[3] At their previous company, they had increased deployment frequency by more than 300 times. They assumed other companies would buy faster delivery.
Mappen pushed the founders to test that belief before building. The team conducted more than 100 customer interviews, often lasting 30 to 60 minutes, and assembled hundreds of pages of notes. Odio wrote, “We didn't write a single line of code until we'd completed 100 customer conversations.”[4]
Those interviews changed the product's emphasis. Leaders at large companies did not first ask to deploy more often. They wanted releases to stop being frightening. Some still scheduled downtime on Friday nights, gathered a team around the release, and spent the weekend repairing failures. “What they are really looking for isn't velocity, it's safety,” Odio concluded.[4]
Spinnaker supplied the technical base for that safety. Netflix had open-sourced it in 2015 after building it to manage thousands of microservices and thousands of daily deployments. Armory packaged it for enterprises and contributed upstream. Mosquera explained why deployment mattered as architectures split into smaller services: “If the overhead of deployments is high, breaking up a monolith into smaller components only increases overall complexity.”[5]
The founders entered Y Combinator's Winter 2017 batch and positioned Armory as the company that would make Spinnaker usable, governable, and supportable inside large organizations.[6]
Armory's first main product was an enterprise distribution of Spinnaker installed in a customer's Kubernetes cluster. Spinnaker represented deployment as pipelines and could target AWS, Google Cloud, Kubernetes, and other environments. It supported blue-green releases, staged traffic shifts, and canary analysis, in which a small new deployment was compared with a baseline before wider release.[11]
Armory added the pieces large organizations expected around the open-source core: supported releases, installation and upgrades, policy controls, security and compliance rules, service-level commitments, integrations, and proprietary plugins. The goal was a paved road from an approved build to production, without every business unit inventing its own scripts.
The company later separated two offers. Continuous Deployment Self-Hosted remained its managed enterprise version of Spinnaker. Continuous Deployment-as-a-Service was a hosted, declarative service for Kubernetes. It could connect to an existing CI system, deploy a new version alongside the live one, shift traffic in stages, run health checks, and roll back if those checks failed.[8]
That second product addressed a real objection to Spinnaker: customers wanted deployment controls without operating a collection of services themselves. It also created transition risk. Armory had to maintain a complex installed base while building a simpler hosted architecture that could compete with newer GitOps tools and broader platforms.
Armory occupied the deployment layer between continuous integration and production infrastructure. It argued that CI tools create tested artifacts, while continuous deployment governs how those artifacts move through environments. That distinction mattered most at companies with many teams, clusters, clouds, approval rules, and failure costs.
Its initial buyers were Global 2000 platform-engineering and DevOps organizations. Named customers included JPMorgan Chase, Autodesk, Xero, Informatica, Patreon, LaunchDarkly, First Republic Bank, and HelloSign. B Capital said JPMorgan Chase used Armory to handle more than 25,000 deployments a day across public and private clouds.[12]
The hosted product widened the pitch to cloud-first teams using Kubernetes, including those that did not already run Spinnaker. That buyer wanted progressive release controls without replacing its source, CI, or monitoring tools.
Armory addressed a substantial portion of enterprise software operations: the people and systems responsible for turning a tested build into a safe production release. The budget could appear as a software subscription, open-source maintenance, platform-engineering headcount, or avoided outage cost. The $82 million invested and the competing suite vendors show that investors expected a large category, but Armory never published enough customer or revenue data to establish its own share.
Armory competed with Harness most directly, plus commercial Spinnaker support from OpsMx, other delivery platforms such as CloudBees and Codefresh, integrated GitHub and GitLab workflows, and internal systems built on Argo CD, Flux, Jenkins, or cloud-provider services. Open source cut both ways: Spinnaker gave Armory a credible engine and community, but customers could use the core without paying Armory or choose a newer controller with a smaller operational footprint.
Armory sold enterprise subscriptions, support, managed operation, and later hosted deployment software. The self-hosted offer monetized the gap between an open-source project and a production system that a large company could trust. Premium features and service commitments supported contract values well above a developer-seat product.
The model also carried costs that pure SaaS avoids. Enterprise Spinnaker installations varied by customer, ran inside customer infrastructure, required upgrades and support, and connected to sensitive production accounts. Armory needed experienced engineers and customer teams to support that estate. Its hosted service promised better economics and easier adoption, but arrived while the company still had to serve the old product.
Venture financing amplified the required outcome. Raising $82 million meant Armory needed a large, fast-growing independent business. A durable support operation or a modest deployment product could be valuable without being valuable enough for that capital structure.
Armory raised a $10 million Series A, $28 million Series B, and $40 million Series C. By October 2020 it had 75 employees and intended to double over the next year.[1] Y Combinator later listed a team size of 90.[6]
The product reached demanding enterprise environments. Public references included a bank running more than 25,000 daily deployments and customers across software, finance, and retail. Armory held seats on Spinnaker's steering and technical committees and helped found the Continuous Delivery Foundation.[12]
Armory did not publicly disclose reliable recurring revenue or customer counts. The reported $7 million asset price is therefore the clearest financial signal. It indicates that the company's technology, contracts, and team did not support an acquisition value near the amount invested, regardless of product usage.
The immediate end was an asset sale. Harness bought key intellectual property and technology, hired many Armory employees, continued support for existing installations, and invited customers to migrate to its larger platform.[10] At roughly $7 million in cash against more than $82 million raised, the transaction was a severe loss of invested capital.
Harness CEO Jyoti Bansal gave the most direct market diagnosis: “For single-point solutions, it's hard to compete.”[2] Enterprise buyers were consolidating tools and expenses. Harness could sell continuous delivery as one module in a larger software-delivery platform, spread engineering across more products, and use an existing sales force to reach the same accounts.
Armory also had a product-transition problem. In 2020, it described a SaaS version as coming the next year.[12] Public early access arrived in May 2022. That delay is not proof of a single execution failure, but it narrowed the time available to establish the hosted product before the 2023 funding market and software-budget consolidation. Meanwhile, the company maintained a demanding Spinnaker estate.
The broader technical direction moved toward simpler GitOps controllers and integrated developer platforms. Armory understood the need for a hosted, toolchain-compatible product, but competing from between free projects and bundled suites required sharper distribution and lower operating costs than its capital base allowed.