What's next for Platform Engineering?

What's next for Platform Engineering?
Photo by Fer Troulik / Unsplash

I was chatting with my coworker PePe over lunch last week when he flat-out declared, "Platform Engineering shouldn't exist." I raised an eyebrow—typical PePe, but he went on: teams don't want to figure out pipelines or develop infrastructure as code. Product teams barely notice the platform and once it's built, that's it... finished! (Which is usually far from reality). That shrug in his voice made me realize how deep the disillusionment runs.

I've been in Software Engineering for 28 years, with the past 5 of those years in Platform Engineering. Let's unpack why Platform Engineering lost its shine, where the industry signals are pointing, and how AI is reshaping the entire landscape.

The Boom-and-Bust of Platform Engineering

DevOps was always a confused notion—ask five people what it was and you'd get five different definitions. ("It's not a role, it's a culture!")

In theory, self-service portals and reusable templates would eliminate ad-hoc infrastructure provisioning and empower development teams. In practice, tooling was fragile, self-service only half-baked, and platform teams ended up simply building pipelines and provisioning cloud infrastructure for others.

Some companies did see gains—developers deployed faster and ops gained governance—but the reality proved more complex. Internal development portals like Spotify Backstage became overwhelming to set up and maintain. Many organizations discovered they'd spent 12-18 months installing Backstage with little to no adoption, mistaking the portal for the entire platform. As industry leaders now recognize, Backstage is not your platform—it's just the front end, a UI that allows developers to discover and access the platform's underlying capabilities.

Building and maintaining a good platform is hard, and the ROI story grew murky:

  • Platforms that lack visible value or clear developer adoption metrics are costly to build
  • Many efforts never reached critical adoption
  • Teams wanted flexibility, not locked-down workflows
  • Organizations focused on the wrong layer—the portal instead of the backend

Platform Engineering emerged as a profession from DevOps, and also means different things to different people; Cloud Architecture, DevSecOps, Developer Experience (DevEx or DX), Service Reliability Engineering (SRE), Automation Engineering and Self-Servicing.

Gartner now places Platform Engineering in the trough of disillusionment—expectations ran ahead of what the technology could deliver. Yet the same analyst firm that predicted 80% of enterprises would have platform engineering initiatives by 2026 has also created a dedicated hype cycle for the space, signalling its fundamental importance even amid the growing pains.

The Industry's Course Correction

The Platform Engineering community has exploded—PlatformCon received over 800 talk submissions in 2024, and the Slack community now boasts 24,000 members. But growth has brought clarity about what doesn't work.

The industry is experiencing three major shifts that address the core problems:

Backend-First Architecture

Instead of starting with portals, successful platform teams are building solid backends first—focusing on APIs and orchestration, then adding user interfaces later. This prevents the common mistake of shoehorning business logic into the front end.

Controlled Self-Service

The era of developers having unlimited access to cloud consoles is ending. While developers enjoyed almost unfettered console access, this created unoptimized resources, increased costs, and operational nightmares. Platform engineering is becoming a vending machine layer where developers select infrastructure from pre-defined templates and guardrails.

Pareto Efficiency

Platform initiatives must benefit all stakeholders without negatively impacting others. A platform that provides a 10x improvement for developers but forces operations teams to work extra hours isn't truly successful. Platform Engineering is a multiplayer game involving developers, infrastructure, security, architects, and executives—everyone needs to win.

AI Adoption

While platforms wrestled with adoption woes and organizational challenges, AI rapidly transformed infrastructure management. In IDEs and CI pipelines, tools like Claude Code, Pulumi AI, and GitHub Copilot scaffold code, detect drift, and flag compliance issues. AI Agents are already writing decent infrastructure as code, securing configurations, and automating routine tasks.

This doesn't replace engineers—it reframes their role. Instead of hand-crafting every line of code, engineers review, validate, and optimize AI-generated output, focusing on governance, security, and quality. That AI presence is moving into every corner, from IDE prompts to policy-as-code enforcement in pipelines.

The next wave will see AI orchestrating entire workflows—generating Jira (how long do Jira or other collaboration tools exist in the post-AI apocalyptic landscape?) requirements from architecture diagrams (or vice versa), scaffolding compliant pipelines, and enforcing guardrails on the fly. This technological shift addresses many of the adoption and maintenance challenges that plagued early platform engineering efforts.

Where Does That Leave Platform Teams?

Platform Engineering isn't dead, yet. The novelty has faded, but the core work remains vital: stable CI/CD pipelines, landing zones, and coherent cloud architectures that tie microservices and shared services together.

High-performing teams are shifting from product builders to ecosystem stewards:

  • Governance-as-a-Service: Implementing adaptive guardrails that enable developer velocity while maintaining compliance
  • AI Orchestration: Managing and optimizing AI agents that handle routine infrastructure tasks
  • Ecosystem Curation: Maintaining the templates, modules, and policies that enable consistent, secure deployments
  • Multi-Stakeholder Optimization: Ensuring platform improvements benefit developers, operations, security, and executives without creating new burdens for any team

The Ops Side of the Equation

Traditional operations teams aren't just evolving—they're standing at the precipice of the most fundamental shift in IT operations since the advent of cloud computing. While many assume ops roles will gradually adapt, the reality is more radical: we're witnessing the emergence of an entirely new operational paradigm.

The immediate shift shows ops professionals moving from manual intervention to AI collaboration—not building autonomous systems from scratch, but integrating AI capabilities into existing operational workflows. They're becoming the bridge between AI tools and real-world infrastructure complexity, teaching AI systems the nuanced decision-making that comes from years of production experience.

But this is just the beginning. As AI agents become more sophisticated, the concept of "operations" itself transforms. Why have separate monitoring, incident response, and capacity planning when AI can maintain a continuous understanding of system health and autonomously optimize across all dimensions? The future ops professional becomes more like an AI systems architect—designing intelligent infrastructure that self-heals, self-scales, and self-optimizes.

Visionary organizations are already moving beyond platform engineering to autonomous infrastructure—systems that don't just provide developer self-service but actively participate in the development lifecycle. These AI-driven platforms understand application behavior, anticipate resource needs, and even suggest architectural improvements.

The highest form of operational excellence is when operations becomes so intelligent and autonomous that developers never have to think about it, and ops teams transform from reactive maintainers to proactive architects of intelligent systems.

The New Shape of Software Engineering

Over the past few decades, we've seen tooling evolve rather than disappear. But AI represents a fundamental shift. Unlike previous technologies that improved specific workflows, AI has the potential to eliminate entire categories of tools by absorbing their core functions. Why maintain separate systems for project management, documentation, and communication when AI agents can coordinate work directly, retain institutional knowledge, and facilitate collaboration through natural language?

Traditional tools like Jira and Confluence exist because humans need structured interfaces to manage complexity - but AI agents don't. They can track dependencies, maintain context across conversations, and orchestrate workflows without requiring the rigid frameworks that constrain human users. For the first time, we're not just seeing tools evolve - we're witnessing the possibility of tools becoming obsolete entirely.

AI Augments Infrastructure Management

Engineers now review and refine AI output instead of laboring over every line. The next phase will see model-driven automation where architecture diagrams spawn requirements, trigger AI agents that scaffold code, and enforce rules from clean architecture principles and well-architected frameworks.

Rethinking Teams

As AI handles repetitive tasks, roles will polarize:

  • Generalists orchestrate AI tools, maintain architecture, and align cross-team workflows
  • Specialists exist in SRE, multi-cloud resilience, disaster recovery, ML systems, or distributed diagnostics
  • Custodian platform teams maintain the ecosystem—tools, modules, and policy-as-code—that everyone builds upon

Final Thoughts

Platform engineering evolved from DevOps, but it may be solving the wrong problem. While we've been building better platforms for developers to use, AI is poised to eliminate the need for platforms altogether.

The real question isn't how to make platform engineering indispensable—it's whether platform engineering should exist at all. Why create specialized teams to build internal platforms when AI can empower developers to handle infrastructure directly? The future points toward AI-native tooling that makes infrastructure as intuitive as writing application code.

Enabling reliable deployments, secure environments, and coherent architecture is vital work. But instead of centralizing this through platform teams, we're moving toward AI-augmented developers who can generate infrastructure-as-code, configure CI/CD pipelines, and manage operational concerns through intelligent automation.

The writing is on the wall: every layer of abstraction AI eliminates makes Platform Engineering less necessary. Organizations are already questioning the ROI of large platform teams when AI tools can provide similar outcomes at a fraction of the cost and complexity.

The future isn't about improving Platform Engineering tooling and processes—it's about making it obsolete through AI-powered developer self-service. In my opinion, the question isn't whether this will happen, but how quickly we'll get there.