As AI supercharges software development velocity, enterprises face mounting pressure to prove ROI and maintain production stability, governance, and cost control
SAN FRANCISCO, Calif. – May 19, 2026 – CloudBees, the leading software delivery solutions provider for enterprises, today released the 2026 State of Code Abundance, 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.
"Enterprises are living through the same movie they watched with cloud. Adopt fast, figure out the economics and security implications later, and panic when the bill arrives," said Anuj Kapur, CEO of CloudBees. "This time, the spend is less visible, the benchmarks less clear, and the technology changes week to week. Developers are responding to unclear governance— like teenagers with a credit card and no spending limit. At CloudBees, we believe you can't replace one black box with another and call it progress. Winning isn't just about moving fast. It's about building something you can actually control”
"We've never been able to build and write code like this before; it's exhilarating. But we need to remember that code is just an artifact; it's not the actual outcome organizations are pursuing. The real outcome is a quality product. You don't get that until the code is verified, tested, audited, and deployed into the hands of users." – Phil Nash, Developer Relations Engineer - AI, Agents & MCP, Langflow Project, IBM
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.
About CloudBees
CloudBees is the trusted software delivery solution for the agentic era, bringing control, governance, and intelligence to modern enterprise software delivery.
As organizations adopt AI-driven development and manage increasingly fragmented toolchains, they face growing challenges around visibility, compliance, and operational consistency. CloudBees addresses these challenges by unifying DevSecOps across teams, tools, and environments, providing a consistent layer of control, policy, and insight across the entire software delivery lifecycle.
At its core is CloudBees Unify, the AI-powered context and control plane for enterprise DevSecOps. Unify connects disparate toolchains into a single view, delivering trusted context, consistent policy, and auditable automation across every stage of delivery without requiring rip-and-replace migration.
Founded in 2010, CloudBees is backed by Goldman Sachs, Morgan Stanley, Bridgepoint Capital, HSBC, Golub Capital, Delta-v Capital, Matrix Partners, and Lightspeed Venture Partners.
Visit us at www.cloudbees.com.