Heroku

Overview

Heroku is an acquired platform success, not a dead startup. James Lindenbaum, Adam Wiggins, and Orion Henry founded it in July 2007 and entered Y Combinator's Winter 2008 batch. The company turned application deployment from infrastructure work into a developer workflow, reached more than 105,000 hosted applications, and signed a merger agreement with Salesforce in December 2010. The acquisition closed January 3, 2011.^[1][2]

The announced transaction estimate was approximately $212 million in cash net of acquired cash. Salesforce later reported actual total cash consideration of $216.7 million. Employee restricted equity, cash for unvested shares, and accounting allocations were separate measures and should not be casually added into a single purchase-price claim.^[3][4]

Heroku survived for more than fifteen years after closing, evolving from Ruby hosting into a polyglot platform with dynos, add-ons, managed Postgres, buildpacks, and multiple platform generations. In February 2026 Salesforce moved it to sustaining engineering, stopped offering new Enterprise Account contracts, and said existing service remained supported and production-ready.^[5] This is investment narrowing after unusually long product survival, not a shutdown.

Founding Story

The founding mission was to make deploying, managing, and scaling web applications as easy as writing them. Heroku commercially launched in April 2009 around Ruby. It abstracted virtual machines, software installation, and hardware management while designing deployment around developers' existing habits.

That workflow orientation was the product. Instead of asking developers to become infrastructure operators, Heroku presented an application abstraction and handled the runtime beneath it. Early deployment became closely associated with pushing code through a Git-like flow, while the platform supplied process execution, routing, logs, data services, and scaling.

By May 2010, Heroku reported more than 60,000 applications. It raised a $10 million Series B led by Ignition Partners, with Redpoint, Baseline, and Harrison Metal participating, bringing reported outside funding above $13 million.^[6] TechCrunch named Shopify, Comcast, and Best Buy as customers. Heroku launched its add-on marketplace in September 2010 and a PostgreSQL add-on in November, expanding from runtime convenience into an ecosystem of attached services.

Timeline

• July 2007: Lindenbaum, Wiggins, and Henry founded Heroku.
• Winter 2008: The company joined Y Combinator.
• April 2009: Heroku launched commercially with Ruby support.
• May 2010: The platform reported more than 60,000 applications and raised a $10 million Series B, taking outside funding above $13 million.
• September to November 2010: Heroku introduced its add-on marketplace and PostgreSQL add-on.
• December 7, 2010: Salesforce and Heroku signed a merger agreement through Hi'iaka Acquisition Corporation.^[7]
• December 8, 2010: Salesforce announced the transaction when Heroku had about 30 employees and powered more than 105,000 applications.
• January 3, 2011: The acquisition closed with more than 110,000 applications on the platform.^[8]
• May 2011: Cedar introduced Procfile-defined process types, independently scalable dynos, consolidated logs, one-off processes, and dyno-hour billing.^[9]
• 2012: Cedar became the default stack; buildpacks made the runtime polyglot; Wiggins published The Twelve-Factor App.
• 2013: Heroku acknowledged that stewardship and routing documentation for the older Bamboo stack had slipped while attention moved to Cedar.^[10]
• November 28, 2022: Heroku began shutting down free dynos, Postgres, and Redis plans, citing fraud and abuse-management costs.^[11]
• June 2025: An unintended operating-system update contributed to an outage lasting up to 24 hours for many customers; Heroku reported no security breach or data loss.^[12]
• February 2026: The platform moved to sustaining engineering while remaining supported and production-ready.

What They Built

Heroku's core abstraction was the application rather than the machine. Current documentation still describes dynos as isolated virtualized Unix containers with ephemeral filesystems. Git, GitHub, or API input becomes a build artifact; a release combines that artifact with configuration and attached services; dynos execute declared processes.^[13]