Salesforce Halts Heroku Development: What It Means for Cloud Platforms [2025]

When Salesforce announced in early February 2025 that it would stop developing new features for Heroku, the news reverberated through the developer community like a dropped anvil. Nobody was shocked exactly, but everyone felt it.

For years, Heroku was the platform-as-a-service gold standard. You could deploy a Ruby on Rails app in minutes. No Dev Ops degree required. No infrastructure nightmares. Just push to Heroku and watch it scale. It was magic for the 2010s.

Then came Agentforce. Salesforce's new agentic AI platform. And suddenly, Heroku looked like yesterday's technology to the executives making decisions.

Here's what actually happened, why it matters, and what you should do if you're running apps on Heroku right now.

TL; DR

• Heroku enters sustaining engineering mode: No new features, but stability and security updates continue indefinitely
• Existing customers unaffected short-term: Current subscriptions work as-is with no price increases or service disruptions
• New enterprise contracts halted: You can't buy new business contracts, but individual developers and credit-card customers can still sign up
• AI-first strategy shift: Salesforce is betting hard on Agentforce, leaving Heroku behind as a legacy product
• Long-term risk unclear: "Sustaining engineering" historically precedes sunsets, but no official discontinuation date exists yet

Heroku's pricing is estimated to increase by 45% over the next 3 years due to infrastructure optimization costs. Estimated data.

What Actually Happened: The Official Announcement

On February 3, 2025, Nitin T Bhat, Senior Vice President and General Manager at Salesforce, posted a brief statement confirming what many in the industry had suspected: Heroku's days of active development were over. Salesforce has stopped selling enterprise Heroku subscriptions and is scaling back upgrades.

"Sustaining engineering" is the term Bhat used. In tech-speak, this means: we're not building new stuff anymore, but we'll keep the lights on and patch security holes. It's not quite a sunset, but it's also not a future. Think of it like a restaurant that stops taking new orders but keeps serving the existing customers until closing time.

The official messaging was careful and measured. Salesforce emphasized that nothing changes immediately. Existing customers can keep using Heroku. Renewals are honored. Pricing stays the same. Credit-card customers (small teams, individual developers) can continue to sign up through the regular dashboard.

What changed: No new enterprise contracts. No new roadmap items. No product innovation.

The timing wasn't random. Just weeks earlier, Salesforce reported cautiously optimistic earnings, with investors and analysts pressing the company on its AI strategy. Salesforce's AI ambitions had crystallized around Agentforce, not around maintaining a 15-year-old infrastructure platform that doesn't fit the agentic AI narrative.

This is corporate strategy in motion. When a company shifts its entire identity toward AI, legacy products that don't align with that vision get deprioritized. It's not malice. It's focus.

Estimated data shows AWS, Google Cloud, and Azure dominating the market, with newer platforms like Vercel and Netlify gaining traction post-Heroku's decline.

Understanding Sustaining Engineering: What It Really Means

"Sustaining engineering" sounds bureaucratic and boring. That's intentional. It's the term companies use when they want to manage a product into irrelevance without admitting it.

Let's break down what sustaining engineering actually includes and what it explicitly doesn't.

What Sustaining Engineering Covers

Security patches are guaranteed. When a vulnerability is discovered, Heroku will fix it. This isn't optional. Running software with known security holes exposes Salesforce to massive liability.

Stability updates are also included. If an underlying system component breaks, the Heroku team will fix it to keep the platform operational. Database bugs, networking issues, load balancer problems, server failures. These get addressed because they directly impact revenue from existing customers.

Billing and support continue. Your account doesn't get canceled. Support tickets still get responses (though response times might slow over time as the team shrinks). You can renew your subscription.

Backups and disaster recovery systems remain operational. Your data doesn't evaporate.

What Sustaining Engineering Explicitly Doesn't Cover

No new features. Zero. You won't see new runtime versions until years after they're released elsewhere. You won't get access to new services or integrations. The product roadmap is frozen.

No performance improvements. If Heroku gets slower, it gets slower. The engineering effort required to optimize might not happen.

No UI improvements. The interface stays as-is. This might seem minor, but when competitors are shipping slick new dashboards, Heroku's stays unchanged.

No new language or framework support. If you want to deploy a Rust app? You're waiting indefinitely, or you're choosing another platform.

No pricing improvements. If competitors drop prices, Heroku's stay where they are.

The precedent here is instructive. Adobe Animate faced a similar announcement in 2024. The company said it was shifting to "maintenance mode." Within weeks, Adobe clarified that the tool wasn't dying, but it also wasn't getting new features. Existing users could continue. New users should consider alternatives.

For Heroku, the same pattern is likely to play out. What's unclear is the timeline.

Why Did Salesforce Abandon Heroku?

Heroku was acquired by Salesforce in 2010 for $212 million. At the time, it was a brilliant move. Salesforce owned the CRM space and wanted to own the "build custom apps on our platform" space too. Heroku gave them that.

For over a decade, Heroku was the default choice for startups and small-to-medium teams. It had a reputation. It had momentum. It was beloved.

But the cloud landscape shifted dramatically. AWS, Google Cloud, and Azure became increasingly accessible. Containers and Kubernetes changed how people think about deployment. Vercel, Netlify, and Railway emerged as modern alternatives with better DX. Docker commoditized containerization.

Meanwhile, Heroku's costs remained high relative to competitors. Many teams found they could run the same workloads on AWS or Digital Ocean for 60-70% less money. Heroku's convenience premium wasn't worth it anymore.

Most critically: Heroku doesn't align with the AI narrative that now dominates tech.

Salesforce wants to be the company that runs enterprise AI. It's building Agentforce as its flagship product. Agentforce is an AI platform for building autonomous agents. It's the future, in Salesforce's vision.

Heroku is a container deployment platform. It's infrastructure. It's commoditized. It doesn't have AI built in. It doesn't make Salesforce look forward-thinking to potential customers.

When you're a public company and your stock price depends on being seen as an AI leader, you allocate engineering resources to AI products, not to maintaining legacy infrastructure platforms.

The Shift Away from Platform-as-a-Service

Heroku's decline also reflects a broader market shift. The entire Paa S category is contracting relative to other segments. Companies that once relied on Heroku have moved to:

• Managed Kubernetes services (like AWS EKS, Google GKE)
• Serverless platforms (like AWS Lambda, Google Cloud Functions)
• Container registries and CI/CD pipelines (managing deployment themselves)
• Modern alternatives (like Railway, Fly.io)

Heroku was always positioned as "no Dev Ops required." But Dev Ops became mandatory. The market moved on. Salesforce recognized this belatedly and decided that Heroku wasn't worth the engineering investment to keep it competitive.

Railway offers a more cost-effective solution compared to Heroku, especially for simple and medium apps, with savings of up to 70% for similar workloads. Estimated data based on typical usage.

What This Means for Existing Heroku Customers

If you're running apps on Heroku right now, the immediate impact is zero. Your apps keep running. Your dynos spin up and down. Your database backups happen. Logs are collected. Everything works.

But there are ripple effects worth understanding.

The Support Question

With a frozen product and no new features, the engineering team supporting Heroku will shrink. This is inevitable. You don't need a 50-person team to maintain a product. You need maybe 10 people to fix critical bugs and security issues.

This means response times on support tickets will slow. Complex issues might take weeks instead of days. The team won't have bandwidth for workarounds or custom solutions. You'll get the basics: "Here's the official answer. Good luck."

The Feature Desert

You need Heroku to support Python 3.13? Unlikely. You want better monitoring tools? Not happening. You'd like a native Redis integration? They don't care.

This sounds harsh, but it's the reality of sustaining engineering. The team is focused on not breaking things, not on making things better.

The Performance Plateau

Heroku's infrastructure will age. New hardware gets deployed at cloud providers constantly. Heroku won't upgrade proactively. Performance improvements that competitors get (faster cold starts, better CPU efficiency, lower latency) won't come to Heroku.

Your apps won't suddenly get worse, but they'll stay where they are while the world moves forward.

The Pricing Question

This is critical: Salesforce isn't obligated to maintain Heroku's pricing forever. Once the new feature development stopped, the path to increasing prices became clearer.

Expect price increases in the next 24-36 months. These will be justified as "infrastructure improvements" or "supporting continued operations." But the reality is simpler: Heroku will charge more because there's less competitive pressure on existing customers. The switching costs are high. Migrating apps is painful. Salesforce knows this.

The Sundown Scenario: What Comes Next

Heroku isn't shutting down tomorrow. But the trajectory is clear if you understand how tech companies manage legacy products.

Historically, sustaining engineering phases last 2-3 years before companies announce official discontinuation dates. During that period:

Year 1-2: Slow Decline

New customers trickle away. Word spreads that Heroku is no longer being developed. Sales slow to a trickle. The team becomes smaller and more focused on keeping the lights on.

During this period, Salesforce will probably increase prices 1-2 times. The stated reason will be "infrastructure modernization" or "cost optimization," but it's really about extracting value from customers who can't easily leave.

Performance issues might crop up. When they do, fixes are slower. Heroku's reputation continues to erode.

Year 3-4: The Sunset Announcement

Salesforce announces a formal discontinuation date, probably 12-18 months out. Existing customers get generous notice. Migration tools are released (probably basic ones). The team gets slightly more focused on helping people leave than on keeping people from leaving.

This sounds counterintuitive, but it's actually common. Once a sunset date is set, the optics flip. Salesforce needs to avoid being the villain who destroyed people's apps. Migration assistance becomes part of the narrative.

Year 4-5: The Actual Shutdown

The platform goes dark. Databases are deleted. Apps stop running. It's over.

For customers, this is when pain actually becomes real. If you've delayed migration, you're now on a tight deadline. You're paying premium prices for a platform that's being shut down. You're competing with thousands of other companies trying to migrate at the same time.

This is the scenario you want to avoid by starting your migration plan now.

Estimated data suggests that Heroku's customer experience will decline over the next few years due to slower support, fewer features, and stagnant performance, while pricing is expected to increase.

Evaluating Alternative Platforms: What's Actually Better Than Heroku

If you're going to migrate off Heroku, you need to know where to go. Let's talk about real alternatives that actually make sense.

Railway.app: The Modern Heroku Replacement

Railway is what Heroku would look like if it were built in 2024 instead of 2007. It's a platform-as-a-service that handles deployment, scaling, and infrastructure for you. It supports multiple languages, has a simple pricing model, and the developer experience is genuinely excellent.

What Railway does right: You connect your Git repository. Push to deploy. Railway detects the app type (Node, Python, Ruby, Go, whatever). It builds and deploys automatically. Scaling happens automatically. Databases are one-click deployments. Environment variables are managed easily.

Pricing is usage-based, starting at

Migration from Heroku to Railway typically takes 1-2 hours for simple apps, maybe a day for complex ones. No code changes required. Just connect your repo and set environment variables.

The catch: Railway is smaller than Heroku. The team is smaller. There's slightly more risk that Railway itself could face financial pressure (though they've raised Series A funding and seem stable). If you need enterprise SLAs and 24/7 phone support, Railway isn't there yet.

Fly.io: The Developer-Friendly Container Option

Fly.io is container-based but abstracts away the Kubernetes complexity. You provide a Dockerfile (or Fly generates one for you). Fly handles the deployment, networking, and global distribution.

What makes Fly special: Global deployment is built in. Your app automatically runs in multiple regions. Users connect to the nearest instance. This is something Heroku never offered well, and it matters for latency-sensitive applications.

Pricing is also usage-based. Simple apps are free on the hobby tier. Paid tiers start at roughly the same price as Railway but include more generous compute allocation.

The learning curve is slightly higher than Railway because you need to understand Docker. But if you're comfortable with containers, Fly is incredibly powerful. The deployment process is fast. The infrastructure is modern. The team is responsive.

Migration from Heroku: You need to create a Dockerfile. For some apps, this is trivial. For others, it's a few hours of work. Once the Dockerfile exists, deployment is straightforward.

Digital Ocean App Platform: The Middle Ground

Digital Ocean App Platform is a managed platform that sits between Heroku's simplicity and Kubernetes's complexity. You push code to Git, and Digital Ocean handles deployment.

What's good: Straightforward pricing (no per-dyno costs, per-tier costs instead). Great documentation. Excellent community. One-click database deployments. Good performance for the price.

What's less good: The UI isn't as polished as Heroku's. The feature set is smaller. Global deployment isn't built-in (though you can distribute manually).

If you need simplicity and affordability, Digital Ocean is solid. If you need cutting-edge features or global distribution, other options are stronger.

AWS ECS/EKS: The Enterprise Option

If you have the engineering resources, managed Kubernetes on AWS EKS or container orchestration on ECS is powerful and scalable.

But this is for companies with Dev Ops expertise. The learning curve is steep. The operational overhead is substantial. You're responsible for more infrastructure decisions.

Pricing can be cheaper than Heroku at scale, but the engineering time required to manage it might not be worth the savings for small teams.

The Strategic Implications: What Heroku's Decline Says About Salesforce

Heroku's decline isn't really about Heroku. It's about Salesforce's strategic priorities.

Salesforce is betting everything on enterprise AI. Einstein AI, Agentforce, and AI-powered CRM features are where the company sees growth. They're where innovation dollars are flowing. They're where the executive attention is focused.

Infrastructure platforms don't fit this narrative. Developers care about them, but they don't move the needle on Salesforce's stock price. Wall Street doesn't get excited about container orchestration.

This is actually a pattern across the industry. Companies are consolidating around AI and moving away from infrastructure products. Google shut down some cloud services. Amazon is winding down certain infrastructure products. Everyone is chasing the AI narrative.

For developers, this is a problem. We need boring, reliable infrastructure. We need platforms that are maintained and improved. But we don't always get them because investors and executives think AI is more exciting.

The lesson: Platform lock-in is real. Don't build a business on another company's infrastructure assuming it'll be there forever. Build with portability in mind. Use Docker, standard APIs, and platforms that are financially healthy and strategically aligned with your needs.

What This Says About Platform Consolidation

Heroku's fate accelerates a broader consolidation. The Paa S category is dying as a separate market. Instead, we're seeing:

• Managed cloud services (AWS, Google Cloud, Azure) taking over the enterprise
• Developer-friendly alternatives (Railway, Fly.io, Vercel) taking over the startup space
• Everything else getting consolidated into existing ecosystems or discontinued

Heroku was profitable. It was an attractive product. But it wasn't growing fast enough to justify the investment in Salesforce's eyes. When a profitable product doesn't fit your strategic narrative, you deprioritize it.

This might sound cold, but it's how public companies work. You have to serve multiple masters: customers, employees, investors, regulators. Sometimes those interests conflict.

Sustaining engineering ensures security patches, stability updates, billing, and backups, but does not cover new features, performance, UI, language support, or pricing improvements.

Migration Strategies: How to Actually Move Off Heroku

If you're ready to migrate from Heroku, here's how to do it systematically without breaking things.

Step 1: Audit Your Current Setup

Before you migrate, understand what you're running. Export your:

• Application code (already in Git, probably)
• Environment variables and secrets (export from Heroku config)
• Database dump (backup your Postgres, My SQL, or Mongo DB)
• Add-on configurations (what third-party services are you using?)
• Buildpack information (what language/runtime are you using?)

Create a simple spreadsheet documenting all of this. It takes an hour and saves you 10 hours later when you can't remember why you have that weird environment variable.

Step 2: Choose Your New Platform

Based on our earlier analysis:

• If you want the easiest migration: Railway
• If you want advanced features: Fly.io
• If you want the cheapest option at scale: Digital Ocean or AWS
• If you want global distribution: Fly.io or AWS Cloud Front + origin

Spend a day evaluating. Create a test account. Deploy a simple app. Feel the experience. You'll spend months on this platform. Choose one you actually like.

Step 3: Migrate One App at a Time

Don't migrate everything at once. Pick your simplest, least-critical app first. Get the process down. Learn the quirks. Iron out the issues.

Then migrate your next app, then the next. By the time you get to your most critical app, you'll be an expert.

For each migration:

1. Deploy the app on the new platform
2. Run both versions in parallel for 1-2 weeks
3. Monitor both for issues
4. Switch traffic to the new version
5. Keep the Heroku version running for 2 weeks as a rollback option
6. Delete the Heroku version once you're confident

Step 4: Database Migration

This is usually the hardest part. If you're using Heroku Postgres, you need to:

1. Create a database backup (Heroku provides this easily)
2. Download the backup to your computer
3. Restore it to your new platform's database
4. Test that everything works
5. Update your application connection strings

If you're using managed databases on your new platform (Railway Postgres, Fly.io Postgres, Digital Ocean Managed Database), the process is straightforward.

If you're using multiple databases, plan extra time. Complex schema migrations can uncover unexpected issues.

Step 5: Add-on Migration

Many Heroku apps use add-ons: Redis, logging services, monitoring tools, etc.

For each add-on:

• Document the current configuration
• Research the equivalent on your new platform
• Set it up on the new platform
• Update your app configuration
• Test it

Sometimes the new platform's built-in options work better than the Heroku add-ons. Sometimes you'll need to keep using third-party services (and just pay them directly instead of through Heroku).

Step 6: Testing and Monitoring

Once your app is running on the new platform:

• Run your automated tests (you have automated tests, right?)
• Manually test critical user flows
• Monitor error rates, response times, and resource usage
• Check database query performance
• Verify that your scheduled jobs still run
• Test your backup and disaster recovery process

If something breaks, you want to catch it before your users do.

Timeline Estimates

For a simple CRUD app: 2-4 hours For a medium app with databases and background jobs: 1-2 days For a complex app with multiple services, real-time features, and custom infrastructure: 3-5 days For a large team with many apps: 2-4 weeks to migrate everything

Building Resilience: The Real Lesson Here

Heroku's decline teaches us something broader about building software: Platform independence matters.

The companies that are least vulnerable to Heroku's shutdown are the ones that built with portability in mind from day one. They used Docker. They managed their own configuration. They didn't rely on proprietary Heroku features.

The companies that'll suffer most are the ones that got too comfortable. They used Heroku add-ons instead of managed services. They optimized everything for Heroku's specific quirks. They built a bunch of technical debt that only Heroku can run.

For future projects, consider:

Use containers from the start. Docker adds complexity, but it also gives you portability. Once your app runs in a container, it can run anywhere.

Minimize vendor lock-in. Use standard tools and libraries. Avoid proprietary services unless they solve a real problem that generics don't.

Plan for migration. Assume your current platform won't exist in 5 years. Build with switching costs in mind. If migrating takes a week, you're good. If it takes 3 months, something's wrong.

Monitor industry trends. When platforms stop getting invested in, the signs appear before the announcement. Watch for hiring freezes, delayed feature releases, and analyst reports saying "platform is mature."

These principles matter more than any specific platform choice.

Estimated data suggests DigitalOcean scores highest in features, while Railway offers the best developer experience. Pricing varies slightly among platforms.

The Agentforce Question: Is This Salesforce's Future?

Salesforce is betting hard on Agentforce as its core product for the next decade. The company is positioning itself as the platform for enterprise AI agents.

Agentforce lets businesses build autonomous agents that handle customer service, sales, support, and operations. Agents that can reason, plan, and execute tasks. Agents powered by AI models (probably large language models like GPT-4 or proprietary Salesforce models).

It's an interesting bet. The market for AI agent platforms is huge. Every enterprise wants to automate repetitive tasks. If Salesforce can become the standard platform for building and deploying AI agents, the revenue potential is enormous.

But here's the risk: The AI agent market is crowded. Open AI is moving into this space. Anthropic is building Claude with agent capabilities. Every cloud provider is working on this. Generic frameworks are emerging.

Salesforce's bet is that enterprise customers will prefer a CRM-integrated AI agent platform to a generic alternative. That's reasonable. But it's not guaranteed.

Heroku's deprioritization makes sense if Agentforce works out. But it also means Salesforce is putting all its chips on one bet. If Agentforce doesn't achieve the explosive growth the company is expecting, Salesforce could end up with a smaller platform portfolio and fewer diversified revenue streams.

Salesforce is trying to be the enterprise AI platform. Whether that works out remains to be seen. But it's clear where the company's priorities are, and Heroku isn't in that list.

Timeline: What Happens When

Here's what we expect to happen over the next few years, based on historical patterns and current signals.

February 2025 (Now)

• Heroku enters sustaining engineering mode
• New enterprise contracts are halted
• Existing customers and credit-card users can continue indefinitely
• No price changes announced yet

H2 2025

• Heroku team shrinks as projects get reallocated
• First price increase possibly announced (framed as "infrastructure optimization")
• More developer coverage of Heroku's decline in tech media
• Companies begin actively planning migrations

2026-2027

• Feature gaps widen relative to competitors
• Performance issues might crop up as infrastructure ages
• Second price increase possible
• Most informed teams have migrated

2027-2028

• Salesforce announces formal discontinuation date (probably 12-18 months out)
• Migration tools released by Salesforce
• Final rush of migrations

2028-2029

• Heroku platform is shut down
• Last apps are migrated or discontinued

This is a rough timeline. It could compress or expand depending on Salesforce's financial situation and how quickly Agentforce gains traction. But the direction is clear.

What About Cloud Platform Alternatives You Haven't Considered?

Beyond the obvious alternatives, there are some interesting newer platforms worth evaluating.

Deno Deploy: If You're Running Modern Java Script

Deno Deploy is built for Java Script/Type Script on the edge. If your app is Node-based, Deno Deploy offers incredible performance through global edge deployment. Cost is per-request, so small apps are cheap.

Catch: Only works for Java Script. And Deno's ecosystem is smaller than Node's.

Render: The Emerging Contender

Render is a newer platform-as-a-service from the creator of Strong Loop. It's designed specifically as a modern Heroku replacement. Supports multiple languages, Git integration, one-click databases, simple scaling.

Pricing is competitive. The product is polished. The only question is long-term viability (newer companies do fail), but the backing and momentum are strong.

AWS Lightsail: If You Want AWS Without the Complexity

AWS Lightsail is AWS's attempt at a simple platform. Simpler than EC2, more powerful than bare Lightsail instances. It's actually a solid option if you don't mind AWS's ecosystem.

Catch: It's still more complex than Heroku, but cheaper at scale.

Each of these has merit for different use cases. The key is evaluating them early, before you're forced to migrate on a deadline.

The Broader Context: What This Means for Developer Tools

Heroku's story is part of a larger narrative about how technology platforms evolve and die.

The best developer tools are the ones that solve a specific problem better than alternatives. Heroku solved "how do I deploy web apps without Dev Ops knowledge?" brilliantly. For a decade, it was the best answer.

But as the problem evolved, so did the landscape. Docker commoditized containerization. Kubernetes became standard. Cloud pricing became more transparent. The differentiation that made Heroku special got diluted.

Now the best answer depends on your needs:

• Need simplicity? Railway or Fly.io
• Need cost efficiency? AWS or Digital Ocean
• Need advanced features? AWS or Google Cloud
• Need specific integrations? Whatever integrates with your tools

There's no universal answer anymore. The market fragmented.

For Salesforce, this fragmentation is actually an opportunity. Agentforce can own a space (AI agents) that's new and growing. Heroku owns a space (general Paa S) that's mature and commoditized. The calculus for where to invest is clear.

For developers, the lesson is: build for portability, stay informed about platform trends, and don't get too comfortable on any single platform.

Practical: What You Should Do This Week

If you're running apps on Heroku, here's what to do:

Monday: Assessment

• List all your Heroku apps
• Document what each one does, what runtime it uses, what databases it needs
• Estimate the effort to migrate each one
• Identify which apps are critical, which are nice-to-have, which could be deprecated

Tuesday-Wednesday: Research

• Spend an hour on Railway, Fly.io, Digital Ocean, and Render
• Deploy a test app to your top 2 choices
• Compare pricing, features, and developer experience
• Make a preliminary decision on your platform

Thursday-Friday: Planning

• Create a migration timeline (if you have 10 apps, migrate 1-2 per quarter)
• Assign owners to each migration
• Document the process for your first migration
• Schedule the first migration for next month

Don't panic. You don't need to migrate everything tomorrow. But you do need a plan.

The Bigger Picture: Why Platform Deaths Matter

When companies shut down products or move them to sustaining engineering, it matters beyond the immediate users. It affects:

The industry narrative. When a product dies, it changes what developers think is possible. Heroku's decline makes developers more skeptical of platform-as-a-service promises generally. That ripples through the industry.

Open source and standards. Products like Heroku drive innovation in open source (they use Docker, Postgre SQL, etc.). When they get deprioritized, funding and attention for their dependencies might shift too.

Employment. The Heroku team is being downsized. Hundreds of people are affected. This compounds the broader pattern of tech layoffs.

User trust. Companies that see Heroku get abandoned become more cautious about building on any vendor platform. They invest more in portability. This creates a defensive stance that's sometimes counterproductive.

Platforms dying is normal and healthy in some ways (makes room for better alternatives). But the way companies manage them matters. Salesforce's approach is relatively measured. They're not pulling the plug immediately. They're giving customers time to plan.

But the message is clear: If you build your business on another company's platform, assume it won't be there forever.

Looking Forward: What's Next for Heroku and Salesforce

In the immediate term, nothing changes. Heroku keeps running. Customers can renew. The lights stay on.

Over the next 2-3 years, you'll probably see:

• Price increases to extract remaining value
• Slow deterioration of the product and service quality
• Customer departures accelerating once migrations become standard knowledge
• Team shrinkage as Salesforce reallocates resources

Over the next 5 years, you'll probably see:

• Formal sunset announcement with a discontinuation date
• Migration tools released by Salesforce (probably minimal)
• Final exodus of remaining customers
• Platform shutdown and data deletion

For Salesforce, the question is whether Agentforce works out. If it does, Heroku's decline is a minor footnote. If it doesn't, Salesforce might regret not keeping a diversified platform portfolio.

For developers, the message is: migrate thoughtfully, build with portability in mind, and don't let platform ecosystems lock you in completely.

Conclusion: The End of an Era

Heroku built something genuinely great. In the 2010s, it changed how people deployed web applications. It democratized infrastructure. It made it possible for solo developers to run production apps without becoming Dev Ops engineers.

But technology moves on. Containers became standard. Kubernetes became mature. The problem that Heroku solved elegantly got solved adequately by dozens of competitors. The differentiation eroded.

Salesforce's decision to stop developing Heroku is disappointing but not surprising. It's the logical conclusion of a product that became commoditized in a market where the company isn't competing as a leader.

For teams running on Heroku, this is a wake-up call. Start planning your migration. Don't wait until the sunset announcement. Don't rush and break things. But do move deliberately toward alternatives that better fit the current landscape.

The good news: The alternatives are genuinely good. Railway, Fly.io, and others are as easy to use as Heroku but cheaper and more modern. Migrating is work, but it's not catastrophic.

The bigger lesson: Build software that can move between platforms. Use containers, standard protocols, and portable architectures. Assume every platform you use will eventually get deprioritized. Plan accordingly.

Heroku's story is sad for the loyal users and the employees affected. But it's also instructive. It's a reminder that in technology, everything is temporary. The platforms we build on, the frameworks we use, the languages we code in. They all have lifespans.

The teams that thrive are the ones that understand this and design for resilience. The ones that can adapt when platforms shift. The ones that don't mistake any single technology choice for a permanent solution.

Heroku taught us how to deploy apps simply. Its decline is teaching us something just as important: how to prepare for when technology we depend on goes away.

FAQ

What does sustaining engineering mean for Heroku customers?

Sustaining engineering means Heroku will continue to fix critical security vulnerabilities and infrastructure issues, but will not develop new features. Existing customers can keep using Heroku indefinitely, and their subscriptions can be renewed at current pricing. However, the engineering team will be much smaller, so support response times will likely slow down and no new capabilities will be added to the platform.

Is my Heroku app going to break immediately?

No. Your app will continue to run normally in the near term. Heroku will maintain the platform's stability and security. The immediate impact is zero. However, over the coming years, the lack of platform improvements means you'll fall behind competitors in performance, features, and developer experience. Eventually—probably in 3-5 years—Heroku will likely announce a formal discontinuation date, at which point you'll need to have migrated to maintain your app.

How do I migrate off Heroku without breaking my app?

Start by choosing a new platform (Railway, Fly.io, or Digital Ocean are good options). Export your code, environment variables, and database from Heroku. Deploy your app on the new platform and run both in parallel for 1-2 weeks to ensure everything works. Then switch traffic to the new platform and keep the Heroku version as a rollback option for another week before deleting it. The process is straightforward for most apps and takes 1-2 days for medium complexity applications.

Will my costs increase if I stay on Heroku?

Eventually, yes. Heroku will likely increase prices over the next 24-36 months, justified as "infrastructure optimization costs." Existing customers currently have no pricing changes planned, but the historical pattern is that platforms in sustaining engineering eventually see price increases as they extract maximum value before sunsetting. Migrating to alternatives like Railway typically costs 30-60% less for equivalent performance.

Which alternative should I migrate to: Railway or Fly.io?

Railway is easier if you want to minimize complexity and have a quick migration (it works almost identically to Heroku). Fly.io is better if you want advanced features like automatic global distribution, better performance optimization, and don't mind learning Docker. For most teams migrating from Heroku, Railway is the path of least resistance.

Should I migrate now or wait and see what Salesforce does?

Start planning now but don't rush the migration. Use the next 6-12 months to evaluate alternatives, pilot migrations on non-critical apps, and develop your team's expertise. Waiting until a formal sunset announcement is risky because you'll be migrating on a deadline with thousands of other companies competing for infrastructure resources. But rushing blindly and breaking things is also bad. Deliberate, phased migration over 6-12 months is the optimal approach.

Are there alternatives to platform-as-a-service that I should consider?

Yes. If you're willing to manage more infrastructure yourself, Docker with Kubernetes on managed services like AWS EKS or Google GKE gives you more control and typically costs less at scale. However, this approach requires more Dev Ops expertise and operational overhead. For teams without strong Dev Ops capabilities, staying on a managed platform like Railway is smarter.

Can I keep using Heroku for legacy apps while migrating newer ones?

Absolutely. In fact, this is a best practice. Migrate your simplest, least critical apps first while keeping important legacy apps on Heroku. Once you're confident in your migration process, move the critical apps. This staged approach minimizes risk and gives you room to learn the new platform before it matters.

What's Salesforce's plan for Heroku long-term?

Salesforce hasn't announced a formal discontinuation date for Heroku, but the pattern is clear. Sustaining engineering typically lasts 2-3 years before sunset announcements. Based on historical patterns with similar products, expect a formal discontinuation announcement in 2027-2028, with the platform shutting down 12-18 months after that. This timeline is estimated, not official.

Does this affect Salesforce's other products?

Not directly. Salesforce's core CRM products and Einstein AI features are not affected by Heroku's deprioritization. This is a strategic choice to focus on Agentforce and AI products rather than maintain a legacy platform-as-a-service offering that doesn't fit the company's new direction.

Is platform-as-a-service dying as a category?

The Paa S category is consolidating rather than dying. Generic platforms like Heroku are losing market share to specialized platforms (Railway, Fly.io for simple deployment; AWS, Google Cloud for enterprise), serverless functions for event-driven workloads, and containers for full control. The Paa S market is fragmenting. What was once dominated by Heroku and a few competitors is now much more diverse, with different platforms optimized for different use cases.

Key Takeaways

• Heroku is moving to sustaining engineering mode meaning no new features but continued security and stability updates indefinitely
• Existing Heroku customers face no immediate changes but should plan migrations over 12-18 months before likely sunset (2027-2028)
• Modern alternatives like Railway and Fly.io offer similar ease-of-use with 30-60% lower costs and better feature sets
• Platform lock-in is real: building with Docker and standard tools from day one ensures portability when platforms decline
• Salesforce's shift prioritizes Agentforce AI platform over legacy infrastructure products, reflecting broader industry trends toward AI-first strategies

Related Articles

• Europe's Sovereign Cloud Revolution: Investment to Triple by 2027 [2025]
• Spotify's New API Restrictions: What Developers Need to Know [2025]
• A16z's $1.7B AI Infrastructure Bet: Where Tech's Future is Going [2025]
• GitHub's Claude & Codex AI Agents: A Complete Developer Guide [2025]
• Adobe Animate Lives On: Maintenance Mode Reversal Explained [2025]
• OpenClaw AI Agent: Complete Guide to the Trending Tool [2025]