Serverless myths debunked in 2025 – we take a hard look at serverless myths that persist in 2025. It’s a candid discussion between experienced engineers and leaders who have worked at the coalface of cloud adoption for over a decade. We unpack what has evolved, what hasn’t, and why mindset and fundamentals still matter more than ever.
The Serverless First Age Revisited
When The Value Flywheel Effect was written in 2021-2022, serverless was entering what the authors called the “next best action” phase. Since then, the cloud landscape has evolved rapidly, but many of the myths that plagued serverless adoption remain remarkably resilient.
We revisited these myths not to reminisce, but to examine their current validity and relevance. Here’s what stood out:
1. Cold Starts Are a Problem
This myth still lingers, but in reality, cold starts have largely become a solved problem. While unfamiliar teams may still encounter them, the ecosystem now offers well-established mitigation patterns, and platforms have made huge strides in predictive warm-ups and latency reduction.
Cold starts now serve more as an indicator of architectural maturity than a true blocker.
2. You Can’t Test Serverless
This was once a serious friction point. But with tooling like LocalStack, ephemeral environments, and advanced IDE integrations, testing serverless apps has become significantly more robust.
Testing in the cloud is not only possible—it’s often the best approach.
3. Observability Is Impossible
Distributed architectures do add complexity, but the rise of OpenTelemetry and cloud-native observability standards has made it much easier to gain end-to-end visibility. The shift away from vendor lock-in towards open standards is helping immensely.
Observability is no longer a limitation, it’s a discipline.
4. Serverless Is Not the Future (AI Is)
There’s no denying the buzz around GenAI and agentic patterns, but this isn’t a zero-sum game. Fundamentals still matter. System design, business value, and delivery mindset trump shiny new tech.
If your systems thinking is weak, AI won’t save you.
5. We’ll Get Locked Into a Cloud Provider
This myth has faded somewhat. Modern architecture patterns, composability, and smarter cloud strategies make switching more feasible—when it’s actually necessary. But most teams are better off leveraging the innovation of a single provider than maintaining brittle abstractions.
Lock-in is less about tech, more about mindset and governance.
6. We’re Different. We Need Our Own Standards.
This one’s still hanging around. Many teams still attempt to reinvent the wheel. But opinionated platforms like serverless thrive on enabling constraints and proven best practices.
Unless your use case is truly novel, adopt the standards.
7. Serverless Is Expensive
Serverless can be expensive—if you’re wildly successful. But cost control has improved, and understanding your workload is key. Tools and patterns for scaling cost-effectively are maturing fast.
Total cost of ownership (TCO) matters more than raw compute cost.
8. Engineers Won’t Do What I Tell Them
This was always more of a leadership issue. Good engineers want to work in environments where they have agency, clarity, and purpose. Mandating tools or approaches from the top-down rarely works.
Empowered teams ship better software.
9. Our Engineers Are Disconnected From the Business
This gap is closing. Concepts like Team Topologies and fast feedback loops are becoming mainstream. Still, not every org has bridged the gap between tech teams and business value.
Modern orgs know that alignment is a competitive advantage.
10. Security Won’t Let Us Go Serverless
This has improved dramatically. Isolation, patching, and runtime hardening in serverless platforms now often exceed traditional architectures. Most objections are no longer technical, but cultural.
Security is now a reason to go serverless, not a blocker.
11. We Can’t Support OPEX in Our Finance Model
This myth is slowly dying as CFOs and finance teams adapt to cloud-native models. Cloud cost predictability tools have helped bridge the gap.
Finance is finally catching up to cloud.
12. A Consultancy Will Sort Us Out
Expertise matters, but strategy and ownership can’t be outsourced. Successful teams own their decisions and leverage consultants for enablement, not direction.
You can’t buy transformation. You have to lead it.
Final Thoughts
What struck us most is how many of these myths have less to do with technology and more to do with organisational mindset and leadership maturity. As serverless matures, it’s becoming clearer that success isn’t about tools or trends—it’s about focus, clarity of purpose, and a culture of learning.
Serverless Craic from The Serverless Edge
Check out our book, The Value Flywheel Effect
Follow us on X @ServerlessEdge
Follow us on LinkedIn