2025年软件交付状况报告

02The developer crisis is intensifying 03Developer toil disrupts productivity 05AI success, setbacks & surprises 09Shadow AI is the new shadow IT 11Unlocking value across the SDLC 14AI is coming for your job...or not 16Your next steps forward The developer crisis isintensifying. Relentless market demands have fundamentally transformed the role of softwaredevelopers, oftentimes creating impossible delivery expectations. Where developers oncefocused primarily on writing and maintaining code, they now navigate an ever-expandingscope driven by their users' insatiable appetite for rapid innovation. In the last 24 months, AI-powered development tools have emerged as a potential solution. how well organizations areadopting these tools, identify current challenges, and solidify a true path for success thatincorporates AI in modern software delivery.We surveyed 500 engineering leaders and practitioners to explore Today's developers find themselves wearingmultiple hats– they're expected to be securityexperts implementing robust safeguards,operations specialists managing complexinfrastructure, performance engineersoptimizing system efficiency, and UX advocatesensuring seamless user experiences. It also doesn’t help that the majority ofdevelopers are burdened with legacy processesthat result in a deluge of toil.And while this consolidation of responsibilitiescan improve efficiency, it has dramaticallyincreased the cognitive load on developers.
 Developer toil disruptsproductivity 78% of developers spend at least 30% of their time on manual,repetitive tasks. This includes things like writing compliancepolicies, quality assurance testing, or handlingerror remediation.They also fail to quantify howdeveloper toil is hindering profitability.Organizations often exacerbatethese challenges by failing to recognize thecumulative toil of constant context switching andcognitive overload. For example, the average salary of the developerswe surveyed is $107,599. So if 30% of their job ismade up of redundant tasks, that equates to$32,280 of wasted investment in each developer.When you consider we surveyed organizations withat least 250 developers, we’re talking at least$8Min lost productivity annually per engineering teamwe surveyed. Beyond the financial waste, the drive for efficiencycan actually lead to burnout as most are workingovertime. And when asked about the impact working overtime has on their workplace wellbeing andpersonal lives, they were pretty clear that it… Creates an unhealthy work/life balance Increases burnout Increases stress and anxiety levels Steals time they can spend withfamily and friends Entices them to leave the organization AI success, setbacks &surprises To combat this predicament, the rise of AI-powered code generation tools represent asignificant shift in how developers manage their expanding responsibilities. At their core, AICodeGen tools offer a form of intelligent augmentation that addresses the growingcomplexity of modern software development. 98%of developers believe AI tools area great way to reduce burnout However, despite this positive sentiment, adopting AI codegen is far from being a risk-freeendeavor. 92% of developers stated that while AI tools increase the volume of code shippedinto production, it also increases the "blast radius" from bad deployments. And that’s just thetip of the iceberg. of developers spend more time debugging AI generated code of developers spend more time resolving security vulnerabilities Contrary to popular belief, developers must invest significant time in understanding,debugging, and refining AI-generated code. AI systems may also generate code that includesoutdated dependencies or insecure coding patterns, requiring developers to spend timeupdating and patching these vulnerabilities. of developers experienceproblems with deploymentsat least half of the time whenusing AI coding tools. The adoption of AI codegen tools thus far has actually resulted in a shift in the developerworkload–This increased verification overheadarguably offsets a considerable amount of the productivity gains.while AI accelerates initial code production, it creates new demands around codereview, security validation, and quality assurance. Engineering leaders also share some notable concerns about the increased use of AI codegeneration tools. When asked about the impact working overtime has on their workplacewellbeing and personal lives, they were pretty clear that it… Increase in vulnerabilities andsecurity incidents Increase in performance problems Increase in manual downstreamwork(QA, testing, integration) Increased risk of regulatory
non-compliance Shadow AI is the newshadow IT Perhaps the most alarming observation was around the use of company-approved codingtools - or lack thereof. The unauthorized adoption of AI codegen tools creates significantshadow IT challenges that extend far beyond immediate security concerns. Shadow AI usage raises serious