The IT Nerd

CloudBees today released the State of Code Abundance 2026, finding that AI-generated code is straining the enterprise systems built to deliver it, revealing a widening gap between confidence in AI-readiness and operational reality. 

The survey of more than 200 enterprise technology leaders reveals rising infrastructure costs, weak governance frameworks, and mounting operational risk, with 81% reporting production failures tied to AI-generated code. Meanwhile, “token anxiety” is emerging as finance teams struggle to forecast AI spend quarter to quarter. The pattern mirrors the early days of cloud adoption, when limited visibility and control left enterprises exposed to runaway costs.

More Code Isn’t Translating Into Business Value

AI is now deeply embedded in enterprise software development, with 64% of leaders saying it is widely adopted or fully integrated into engineering workflows. But increased code output has not translated into clear business impact, leaving organizations struggling to connect AI-driven development to measurable ROI. 

CloudBees’ findings reflect broader industry trends: external research shows that despite 75% of developers using AI coding tools, most organizations report no measurable improvement in business results.

Key findings include:

• Code volume surges: 67% of enterprise technology leaders report a significant increase in code volume over the past 12 months, while 52% cite higher development output in features and pull requests. 
• Value realization lags: Despite this surge, 36% of organizations track AI spend without measuring ROI or don’t measure ROI at all.
• Attribution gap persists: Organizations rate themselves highly on ROI measurement confidence (51% very confident), yet only 31% of AI spend can be attributed to specific business outcomes.

Token Anxiety Emerges as the New Cloud Anxiety

AI-related costs are escalating across multiple layers, not just in token consumption, but in the downstream expenses building across testing, infrastructure, and security.

Additional key findings include:

• Infrastructure costs are climbing: 54% report a significant increase in CI/CD infrastructure spend over the past 12 months, while 53% say testing, security scanning, and deployment costs have risen alongside growing code volume. 
• Cost management remains reactive: Only 27% of organizations have set hard limits or quotas on token usage, and just 18% have implemented automated controls.
• Budget forecasting remains unresolved: Only 45% describe their AI spend as very predictable quarter-to-quarter.

AI Velocity Is Outpacing Enterprise Governance

AI is compressing the time between code creation and deployment, but governance, validation, and accountability frameworks are not keeping pace. When no human fully engages in the cognitive process of building, ownership of failures becomes harder to assign. 

For example: 

• Production issues rise as governance lags: 81% have experienced an increase in production issues attributable to AI-generated code.
• Validation can’t keep up with volume: 70% now view test suite maintenance as a bigger burden than writing code itself, as AI generates more code than teams can effectively validate.
• Accountability defaults upwards: 46% say the CTO or VP of Engineering is ultimately accountable for AI-related failures, while only 12% report having a dedicated governance function in place.

CloudBees Introduces Proprietary CARE Index to Measure Enterprise AI Readiness

As part of this research, CloudBees introduces the Code Abundance Readiness Evaluation (CARE) Index, a proprietary composite score designed to assess how effectively enterprises can track, attribute, and forecast AI-driven costs against productivity outcomes. Based on six dimensions of operational readiness, the CARE Index establishes an industry baseline for AI governance maturity and will serve as a recurring benchmark for measuring enterprise progress year-over-year.

The 2026 industry baseline: 83.6/100 — reflecting strong self-reported confidence in AI readiness across enterprises. However, when measured against operational data, a significant gap emerges between perceived preparedness and actual capability.

The index reveals:

• High confidence, low attribution: Organizations score highest on ROI measurement confidence (51% very confident), yet only 31% of AI spend can be attributed to specific business outcomes.
• Visibility without predictability: Cost visibility ranks among leaders’ highest self-reported CARE scores (54% report very clear visibility), yet only 45% describe their AI spend as highly predictable quarter-to-quarter.

The State of Code Abundance 2026 was unveiled at Agentic DevOps World 2026, a virtual summit hosted by CloudBees bringing together CIOs, CTOs and VPs of Engineering to address the growing challenges of governance, cost visibility and delivery confidence at scale. To learn more, visit here. 

Methodology: The study was conducted by independent research agency TrendCandy on behalf of CloudBees and included 213 enterprise technology leaders. The margin of error is +/-8% at the 95% confidence level.

CloudBees, one of the world’s leading software development solution providers, today announced that CloudBees Smart Tests, its award-winning AI-driven test intelligence solution for continuous integration and continuous delivery (CI/CD), is now generally available for all customers.

As AI tools dramatically increase code output, the bottleneck in modern software delivery is shifting from writing code to validating it. With roughly 41% of all code now AI‑generated and more than 80% of developers are using AI tools daily, CI pipelines are under growing pressure from a surge of pull requests that expand regression suites and slow feedback loops.

CloudBees Smart Tests set a new standard for controlling AI-generated code. By ensuring the right tests run for each code change, developers are empowered to maintain velocity without sacrificing reliability and control.

Early enterprise deployments demonstrate measurable impact: 

• 30% faster test execution: Reduced from 54 minutes (69 test cases) to 4 minutes for 18 parallelized tests, with further gains expected when applied across the full pipeline. 
• Automated test failure analysis: Failures are now automatically segmented, replacing manual triage and speeding identification of unstable tests.
• 40% better infrastructure utilization: Reduced infrastructure from 10 executors across 2 VMs to 4 executors on 1 VM for the same workload. 

To address these challenges, the team turned to CloudBees Smart Tests to streamline release testing, accelerate feedback loops, and reduce cloud spend. 

Built for Modern Software Delivery 

CloudBees Smart Tests apply machine learning–based Predictive Test Selection and failure pattern analysis. Key advantages include: 

• Accelerated testing: Runs only the tests most relevant to each code change, finding failures 40-80% faster.
• Controlled CI costs: Reduces unnecessary execution that drives CI cost and delays.
• Reduced cognitive load: Groups failures by root cause to identify what failures to prioritize.
• Streamlined dev-test process: Identifies flaky, reliable, and long-running tests to speed up dev-test interaction, giving both engineers and leaders a shared view of test performance.

Longstanding developer roadblocks like large test suites, flaky failures, reruns, manual triage, and a CI bill that grows with every wasted test minute are amplified with the proliferation of vibe coding and AI-generated code,” said Shawn Ahmed, Chief Product Officer at CloudBees. “Beyond providing time and cost savings, CloudBees Smart Tests restores developer confidence. We’re giving teams the ability to ship AI-generated code knowing it’s been properly validated.

CI-Agnostic to Modernize Without Disruption 

Enterprises rarely operate in a single-CI environment. Multi-team, multi-repository estates often span Jenkins, GitHub Actions, GitLab CI,  and other frameworks. 

Without requiring costly migrations, Smart Tests integrates seamlessly into existing pipelines and works across heterogeneous environments. This flexibility allows organizations to pilot Smart Tests in a single repository, validate impact end-to-end, and expand based on measurable results.

Availability

CloudBees Smart Tests is available today. Enterprises can request a CI Waste Assessment to evaluate optimization opportunities within their existing CI environment.

To learn more, visit www.cloudbees.com.