State of SDLC Fragmentation: The Follow-Up

Last October, we surveyed 1,200 software teams. This October, we did it again. The results surprised us. Fragmentation did not improve. It got worse. But the kinds of fragmentation changed. Here is what we found.

What we did

In October 2024, WalnutAI surveyed 1,200 engineering leaders across North America and Europe. We asked about their SDLC tooling, team structure, pain points, and success metrics.

In October 2025, we surveyed 1,200 teams again. Some overlapping teams, some new. Same questions. One year later.

The goal: understand how SDLC fragmentation evolved in 12 months.

The headline finding

87%

of teams use 6+ point solutions (up from 79% in 2024)

64%

report integration overhead as top pain (up from 51%)

41 avg

tools per enterprise (up from 34)

3.2x

longer incident response due to tool silos (vs unified platforms)

Fragmentation got worse, not better

This might seem counterintuitive. Over the past 12 months, the entire industry has talked about consolidation. Zoom acquired Five9. Figma acquired UMO. Datadog acquired Cloudsmith. Everyone is talking about single platforms.

But in our survey, the opposite happened. Teams added more tools.

Why?
• New problems emerged (AI monitoring, vector databases, prompt engineering)

• New point solutions launched (specialized, focused, cheap upfront)

• Platform strategy failed to deliver (promised integrations delayed)

• No tool does everything well (specialization outweighs consolidation)

What changed year-over-year

1. The AI toolchain explosion In 2024, 23% of teams had dedicated AI tooling for SDLC tasks. In 2025, that jumped to 67%. New tools entered the market: • LLM providers (Claude, GPT-4o, Llama) • LLM ops platforms (Lantern, Prompt engineering, RAG pipelines) • AI-native test automation (Walnut, Testim, Applitools) • AI code analysis (SonarQube, Checkmarx, Snyk) Teams did not remove old tools. They added AI tools on top. Fragmentation expanded vertically.
   

   1. Testing infrastructure fragmentation In 2024, the top 3 testing tools were Selenium, Pytest, Postman. In 2025, the list expanded to 7: Selenium, Pytest, Postman, Playwright, Cypress, Jest, Vitest. Why the split? • Frontend teams chose Playwright/Cypress (faster, better DX) • API teams stayed with Postman (existing investment) • Mobile teams adopted new frameworks (React Native, Flutter specific) • Unit testing fragmented (Jest vs Vitest vs Deno) No single testing tool dominates anymore. Teams pick and mix.

   2. Observability infrastructure got more complex In 2024, the top observability stacks were: Prometheus + Grafana, Datadog, New Relic. In 2025, teams added: • Custom dashboards (because no single tool fits all needs) • Open-source tools (Tempo for traces, Loki for logs, Prometheus for metrics) • Specialized vendors (Honeycomb for debugging, Axiom for log storage) Result: a typical enterprise now runs 4-6 observability tools instead of 2-3.

1. The Git + CI/CD fragmentation deepens In 2024, 78% of teams used GitHub or GitLab as their primary Git provider. In 2025, that stayed the same. But Git plus CI/CD fragmentation deepened: • GitHub: GitHub Actions, GitHub Advanced Security • GitLab: GitLab CI/CD, GitLab Security • Bitbucket: Bitbucket Pipelines + separate security tools • AWS teams: CodePipeline + CodeBuild + AWS security No single CI/CD platform satisfied everyone. Teams stitched together Git provider + specialized CI + specialized security. The cost of fragmentation More tools means more friction:
   3.2 days/month

   spent context-switching between tools (up from 2.1 days)

   47 integrations

   to maintain per team (up from 31)

   $82K/year

   average cost per team in tool licenses + integration work (up from $67K)

   18% of capacity

   spent on tool maintenance, not feature work (up from 12%)

   Where consolidation actually happened

   Not all trends moved toward fragmentation. A few areas consolidated:

   • Container orchestration: Kubernetes won. Swarm, Nomad, ECS adoption dropped.

   • Cloud infrastructure: AWS dominance stayed at 64%. No change.

   • Authentication: Okta/Auth0 consolidated to 71% of enterprises (up from 59%).

   • Incident management: PagerDuty + Opsgenie picked up steam.

   Consolidation happened where one tool was demonstrably better. Kubernetes was simpler than Swarm. Auth0 was easier than rolling your own. PagerDuty was faster than in-house escalation.

   Fragmentation happened where different teams had different needs and no single tool fit all of them.
   What teams want (according to the survey)

   We asked: if you could wave a magic wand, what would you fix in your SDLC?

   Top answers:

   • Single source of truth for test results (67% of teams)

   • Unified view of code quality across repos (71% of teams)

   • Integration that actually works without custom code (58% of teams)

   • Better observability in production (62% of teams)

   • Less time spent context-switching (81% of teams)

   Notice: they did not ask for more features. They asked for simplicity. They asked for tools to work together.
   The business implication

   Software tools have hit peak fragmentation. Teams have reached saturation.

   The next cycle will not be about adding more point solutions. It will be about stitching them together.

   Platforms that solve this win. Platforms that add more features lose.

   For engineering leaders
   

   What should you do with this data?

   1. Audit your toolchain. Count the number of tools. Calculate the integration cost.

   2. Identify the pain. Where do you lose the most time? Where do integrations break?

   3. Prioritize ruthlessly. Not every tool deserves your team's attention. Some should be sunset.

   4. Evaluate new tools on integration, not just features. A new tool that requires custom integration is not a win.

   5. Push back on fragmentation. Every new tool is a debt. Every integration is a future maintenance burden.

See how your team compares. Book a walkthrough https://www.walnutai.ai/ with our research team. We will help you interpret the data for your org.