60% of Global Enterprises Deploy Untested Code to Production Due to AI-Accelerated Software Development

Tricentis Announces Results of Global Survey of Enterprise IT Leaders

One in five companies suffers losses of up to $5 million annually due to declining software quality

In critical industries such as financial services, retail, and energy/utilities, the pressure to prioritize speed over quality persists, expanding potential vulnerabilities

Management perception gaps threaten software quality at global enterprises, making it an urgent management issue

Tricentis Japan LLC – June 5, 2026 – Tricentis, the global leader in agent-based quality engineering, today announced the results of its second 'Quality Transformation Report'. The survey, focused on trust issues in software quality, revealed that while software development teams have expanded AI adoption over the past year and significantly increased release velocity, the growing scale and complexity have introduced new risks to the software development lifecycle (SDLC), leaving many companies struggling to maintain confidence in software quality.

Kevin Thompson, CEO of Tricentis, stated:

"Accelerating business transformation is a top priority for executives, and AI has the potential to increase the speed of software development teams like never before. However, increased speed comes with increased risk. When software quality processes fail to keep pace with development velocity, many companies resort to shortcuts, which can significantly compromise quality and reliability. This survey highlights the intense pressure teams face in balancing speed, quality, and control as software development accelerates. As the impact on financial performance and customer trust becomes more visible and quantifiable, software quality is no longer just an engineering challenge but a critical management priority."

The Quality Transformation Report 2026 is based on a survey of over 2,500 respondents worldwide, including CEOs, CIOs, CTOs, Engineering VPs, DevOps/QA leaders, and software developers across a wide range of industries such as manufacturing, energy/utilities, retail, financial services, and the public sector. Key findings include:

Companies Still Prioritize Development Speed, Leading to Widespread Deployment of Untested Code: Despite the evolution of AI technology and increased adoption of AI tools, globally, 6 out of 10 companies (65.6% in Japan) reported deploying untested code to production. This rate is nearly unchanged from 63% (62% in Japan) in the 2025 survey. However, while the 2025 survey primarily cited 'accidental inclusion of untested code' (40% globally / 32% in Japan), this year's findings reveal a shift towards intentional deployment of untested code, driven by strong executive pressure to prioritize speed over quality (32% globally / 25% in Japan) and the overwhelming volume of AI-generated code that teams cannot adequately test (30% globally / 31% in Japan).

Pressure to Increase Development Speed is Widespread Across Industries: In all major industries surveyed, a majority of companies reported deploying untested code to production. This trend was particularly strong in the financial services (64% globally / 69% in Japan), retail (63% globally / 64% in Japan), and energy/utilities (58% globally / 66% in Japan) sectors, which are under significant pressure.

Quality Management and Governance Lag Behind AI Adoption Speed: While 48% (47% in Japan) of companies reported enterprise-wide AI adoption, over half of them stated that their AI tools and processes change frequently. Additionally, one-third (33% globally / 27% in Japan) of teams cited tool complexity and proliferation as a major barrier to achieving consistent software quality at scale. Other challenges include skill shortages (33% globally / 26% in Japan), code volume exceeding manageable limits (28% globally / 30% in Japan), and a lack of clear metrics for quality and reliability (26% globally / 28% in Japan).

Gap Between Executive Optimism and On-the-Ground Reality: What executives perceive as progress in AI adoption is often seen as an operational burden by software development teams. While approximately four out of five CEOs (81% globally / 80% in Japan) expressed high confidence in AI-driven systems and tools, only 56% (43% in Japan) of QA/DevOps personnel shared that confidence. Furthermore, 44% (26% in Japan) of board members reported being 'well-prepared to operate, manage, and scale AI agents across the SDLC', compared to only 23% (8% in Japan) of QA/DevOps personnel.

High Expectations for Agentic AI, but On-the-Ground Challenges Remain Severe: While 83% (69% in Japan) of companies said they 'trust agentic AI for release decisions' and 82% (64% in Japan) said they are 'ready to operate and control AI agents at scale', many still face significant challenges including untested code (60% globally / 66% in Japan), tool proliferation (33% globally / 27% in Japan), security concerns (27% globally / 33% in Japan), skill shortages (24% globally / 21% in Japan), and data quality issues (24% globally / 18% in Japan).

Financial and Operational Risks from Poor Software Quality are Growing: One in five companies (20% globally / 23% in Japan) reported incurring losses exceeding $1 million annually due to poor software quality. The primary causes were security and compliance issues (30% globally / 34% in Japan) and technical debt/rework costs (28% globally / 25% in Japan). Additionally, 45% (23% in Japan) of companies estimated annual losses between $500,000 and $1 million.

Thompson further added:

"Many companies still rely on quality processes that are not adapted to AI-era software development. As development speed accelerates, executives need clearer visibility into software quality risks and must strengthen collaboration across engineering, QA, and business units. The companies that will succeed in the future are those that can scale by balancing speed and control."

Tricentis' 'Quality Transformation Report 2026' shows that the challenge has shifted from whether AI can be adopted to whether trust, control, and confidence in released software can be maintained at scale.