Heroku Isn’t Dead, But It’s Dying in Slow Motion

Scroll down

Blog

Heroku Isn’t Dead, But It’s Dying in Slow Motion

By Janakiram MSV

The Quiet Fade That Everyone Saw Coming

Salesforce has officially placed Heroku on what it calls a “sustaining engineering model,” a corporate euphemism that translates to one thing: no new features, no strategic investment, just keeping the lights on. The announcement, buried in a blog post by Heroku CPO Nitin Bhat on February 6, 2026, also confirmed that Enterprise Account contracts will no longer be offered to new customers. For a platform that once defined how an entire generation of developers thought about deployment, this is not a sudden death. It is a slow, deliberate winding down.

What makes this particularly striking is the timing. Just six months ago, Salesforce was celebrating Heroku’s recognition as a Leader in the 2025 Gartner Magic Quadrant for Cloud-Native Application Platforms. The company had invested in a Fir-generation infrastructure overhaul, added .NET support, launched Managed Inference and Agents for AI workloads, and built AppLink to integrate Heroku with Agentforce. Now, all of that momentum has been placed on indefinite pause.

What “Sustaining Engineering” Actually Means

Heroku’s blog post is a masterclass in corporate ambiguity. As developer Simon Willison noted in his reaction, the announcement promises clarity while delivering the opposite. “We want to be clear about what this means for customers,” the post reads, then proceeds to bury the real message under reassurances that existing customers can keep paying their bills.

Let’s decode what sustaining engineering actually entails. Heroku will continue to receive security patches and reliability fixes. The platform will keep running. But there will be no new features, no competitive responses to market shifts, and no engineering investment in keeping Heroku relevant against a rapidly evolving PaaS landscape. The one exception noted is a previously announced revamp of the managed PostgreSQL service, which will still ship. Everything else is frozen.

For existing credit-card customers, nothing changes immediately. For enterprise prospects evaluating Heroku for new projects, the message is unambiguous: look elsewhere. And for the developer community that once treated Heroku as a verb (“just Heroku it”), this confirms what many suspected after the 2022 free-tier removal and the OAuth token breach. Salesforce has been gradually divesting from Heroku for years.

Why Salesforce Is Walking Away

The official explanation points to Salesforce “focusing product and engineering investments on areas where we can deliver the greatest long-term customer value, including helping organizations build and deploy enterprise-grade AI.” In practice, that means Agentforce. Salesforce’s agentic AI platform generated over half a billion dollars in annual recurring revenue by Q3 FY2026, with ARR growing 330% year over year and nearly 10,000 paid deals closed since its September 2024 launch.

The math is straightforward. Agentforce is a growth engine generating billions in pipeline. Heroku is a platform-as-a-service that Salesforce acquired in 2010 for $212 million and has struggled to monetize at enterprise scale ever since. When a company with $41 billion in annual revenue has to choose where to allocate engineering headcount, the decision practically makes itself.

There is also the uncomfortable overlap problem. SiliconANGLE’s reporting highlights that Heroku’s Managed Inference and Agents feature directly overlaps with Salesforce’s Agent Builder, which launched in 2024. Rather than maintain two competing AI agent platforms, Salesforce is consolidating around Agentforce and its broader ecosystem of Data 360, MuleSoft, and the recently acquired Informatica.

A Decade of Neglect Culminates in a Controlled Demolition

Heroku’s decline did not begin this week. It began almost as soon as Salesforce completed the acquisition. While competitors like AWS, Google Cloud, and a new generation of PaaS upstarts invested heavily in containers, serverless, and edge computing, Heroku’s core product remained largely static for years. The platform that pioneered simple deployment with “git push heroku main” failed to evolve alongside the developers it once inspired.

The timeline of decline is well documented. Investment dried up through the mid-2010s. An April 2022 security breach involving stolen OAuth tokens damaged trust with the developer community. Later that year, Heroku eliminated its free tier, cutting off the pipeline of hobbyists and students who had made the platform synonymous with learning web development. Significant outages in 2025, including a 15-hour disruption in June, further eroded confidence.

The irony is that Heroku did attempt a reinvention in 2024 and 2025. The Fir generation brought Kubernetes-based infrastructure, new language support, AI capabilities, and tighter Salesforce integration through AppLink. CCS Insight noted in April 2025 that Heroku was being “reorientated to serve as the developer on-ramp for Salesforce’s generative AI ecosystem.” That strategic thesis lasted barely nine months.

The PaaS Market Moved On Without Heroku

While Heroku stagnated, a vibrant ecosystem of alternatives emerged to fill the gap. Railway now offers usage-based pricing, automatic scaling, and native multi-region support. Render provides Heroku-like simplicity with integrated DDoS protection and more competitive pricing. Fly.io runs applications on lightweight VMs distributed globally for low-latency performance. Each of these platforms was explicitly built to capture developers fleeing Heroku’s rising costs and declining innovation.

The shift is not just about price. These newer platforms natively support containers, offer persistent storage, provide built-in environment management, and deliver the kind of developer experience that Heroku pioneered but stopped improving. For organizations evaluating cloud-native application platforms today, Heroku’s Gartner recognition rings hollow when the platform’s parent company has publicly signaled it is no longer investing in new capabilities.

What This Means for Teams Running on Heroku

For the thousands of organizations still running production workloads on Heroku, this announcement triggers a difficult but necessary conversation. Heroku will not shut down tomorrow. Salesforce has every incentive to keep collecting subscription revenue from existing customers for as long as they stay. But a platform in sustaining engineering mode is a depreciating asset.

Security patches will arrive, but they will be reactive rather than proactive. Performance improvements will not come. New language versions and framework updates will arrive slower, if at all. The add-on ecosystem, already thinning, will lose further vendor interest as the market signal becomes impossible to ignore. Engineering teams that delay migration planning are not being cautious; they are accumulating technical debt with a deadline they cannot predict.

The practical advice is uncomfortable but clear. Begin evaluating alternatives now. Railway, Render, and Fly.io each offer migration paths designed for teams leaving Heroku. For organizations deeply embedded in the Salesforce ecosystem, the transition to Agentforce-native development may be the intended path. Either way, building new features on a platform that its owner has explicitly deprioritized is a bet against the house.

The Bigger Signal for Enterprise Platform Decisions

Heroku’s transition to sustaining engineering carries a lesson that extends beyond one platform. When a large enterprise acquirer stops investing in an acquired product, the decline follows a predictable pattern. First, the free tier goes. Then enterprise sales slow. Then engineering investment shifts to “maintenance.” Then the sunset announcement arrives, always preceded by years of assurances that nothing is changing.

Salesforce is not unique in this pattern. Google has done it with dozens of products. Oracle did it with Sun’s hardware portfolio. IBM did it with Lotus Notes. The specifics differ, but the trajectory is consistent: acquisition, integration attempts, strategic deprioritization, and eventual end-of-life. Heroku is somewhere between the third and fourth stages.

For CXOs evaluating platform dependencies, the Heroku story reinforces a principle that should already be table stakes. Understand your vendor’s strategic priorities, not just their current feature set. A platform with excellent capabilities today is a liability tomorrow if its owner is publicly redirecting investment elsewhere. Heroku’s 2025 Gartner recognition did not protect it from Salesforce’s 2026 resource allocation decisions. Your vendor’s analyst rankings will not protect you either.

Heroku is not dead. But for any organization making forward-looking technology decisions, the distinction between “not dead” and “not dying” matters enormously. Salesforce has told the market, in the clearest terms corporate communication allows, that Heroku’s future is its past. The only question left is how long the slow motion continues before the final credits roll.