
No Business for Four Years, the Atlas Pivot

Mitchell Hashimoto’s new way of writing code
Topic 7 of 17
Published Feb 25, 2026 · Clip 32:42–45:07 (12:24)
For four years HashiCorp had no real business — just open source and a build-it-in-five-years mentality, with the comfort that if the company failed, the community would carry the ideas. Mitchell explains the seed/A/B-round framework: seed builds product, A proves hints of product-market fit, B proves it, later rounds build repeatable revenue. Their first commercial product, Atlas, failed because it required adopting all products and no single budget owned it. After a dispiriting board meeting, he and Armon ran a sunk-cost-free thought experiment, whiteboarded per-product enterprise open-core with Vault first, threw everything away over a weekend, and pivoted. Nobody quit; the team was energized by clear direction, and Vault Enterprise shipped by year's end with noticeably better sales conversations. Key lesson: customers were asking about secrets replication all along — they just hadn't listened.