Why they built it
The founders had increased deployment frequency by more than 300 times at their previous company and assumed other companies would pay for faster delivery. More than 100 interviews with enterprise software leaders showed the real problem was fear of releases, with teams scheduling Friday-night downtime and spending weekends repairing failures. They concluded buyers wanted safety over speed and set out to make Spinnaker usable, governable and supportable inside large organizations.
How it ended
Armory ended in an asset sale when Harness bought its assets for about $7 million in cash in January 2024, hired many employees and kept supporting existing installations. It could not sustain a large independent company as enterprise buyers consolidated tools and favored broader delivery platforms, with Harness noting it is hard for single-point solutions to compete. Against more than $82 million raised, the price was a severe loss of invested capital.
What went wrong
Most important first.
- 1
Squeezed by bundled delivery suites Competition
Harness said single-point solutions were hard to compete with as enterprise buyers consolidated tools, while it could sell continuous delivery as one module in a larger platform with shared engineering and sales.
- 2
Heavy self-hosted operating burden Technology
Spinnaker was powerful but expensive to operate, with enterprise installations varying by customer, running inside customer infrastructure and requiring upgrades, support and connections to sensitive production accounts.
- 3
Hosted replacement arrived late Timing
Armory discussed a SaaS version in 2020 but its Continuous Deployment-as-a-Service product only reached public early access in May 2022, leaving little time to establish it while still maintaining the Spinnaker estate.
- 4
Free core limited paid conversion Monetization
Spinnaker gave Armory a credible engine and community, but customers could operate the core without paying Armory or choose newer controllers with a smaller operational footprint.
- 5
Capital base required a large outcome Capital
After raising $82 million through its Series C, Armory needed a large, fast-growing independent business that a durable support operation or modest deployment product could not deliver.
What could be different today
- Argo CD runs in nearly 60% of Kubernetes clusters managed by respondents in a 2025 CNCF survey, while environment promotion still relies heavily on manual work and custom scripts.
- SLSA 1.1, approved in April 2025, clarified artifact attestations and verification summaries, making release evidence portable across tools.
- GitHub introduced a coding agent in May 2025 that can take an issue, work in GitHub Actions and submit a pull request, making the number and origin of changes harder to review from commit history alone.
