Is Vibe Coding Here to Stay? (30 Data to Understand)

We research digital businesses every day, if you're building in this space, get our market clarity reports

Vibe coding tools like Cursor, GitHub Copilot, and Replit are growing faster than any developer tool in history, with valuations jumping 25x in under a year and user bases exploding from 1 million to 20 million in just two years.

But beneath the hype, there's a troubling pattern: developers report declining trust even as adoption rises, security vulnerabilities remain stubbornly high despite model improvements, and code quality metrics are collapsing across fundamental software engineering practices.

We analyzed 30 data points spanning usage metrics, investment trends, productivity studies, code quality research, developer surveys, and market forecasts to answer one critical question: is vibe coding sustainable, or are we heading for a correction? Our market research report about AI Wrappers covers similar adoption dynamics across the broader AI tooling landscape.

In our 200+-page report on AI wrappers, we'll show you which ones are standing out and what strategies they implemented to be that successful, so you can replicate some of them.

Is Vibe Coding Here to Stay? 30 Data Points Reveal the Truth

• 1. Vibe coding tool Cursor's valuation jumped 25x in 10 months

  What happened:

  Cursor's parent company Anysphere raised $900 million at a $9.9 billion valuation in June 2025, up from $400 million in August 2024. Bloomberg called it "the fastest growing startup ever," with the company surpassing $500 million in annual recurring revenue and revenue doubling approximately every two months.

  Why it matters:

  This 25x valuation increase in less than a year is unprecedented even in venture capital history. If sustainable, it signals genuine transformation of software development, but such explosive growth often precedes market corrections. The sustainability depends on whether productivity gains justify premium pricing long term, or whether we're witnessing bubble dynamics driven by fear of missing out.

  Source: TechCrunch

• 2. AI coding assistant GitHub Copilot reached 20M users in 2 years

  What happened:

  Microsoft CEO Satya Nadella announced GitHub Copilot reached 20 million all-time users by July 2025, adding 5 million new users in just three months. This represents 20x growth in approximately two years, making it one of the fastest-adopted developer tools in history, with 90% of Fortune 100 companies now using it.

  Why it matters:

  The raw user numbers show massive adoption velocity, but the metric is "all-time users" not active users, which could be significantly lower. The gap between trial and retention is crucial for assessing whether this is sticky transformation or experimental churn, though the fact that 90% of Fortune 100 use it suggests genuine enterprise stickiness rather than fleeting experimentation.

  Source: TechCrunch

• 3. Vibe coding platform Replit grew revenue 10x in 5.5 months

  What happened:

  Replit CEO Amjad Masad announced the company crossed $100 million in annual recurring revenue in June 2025, up from just $10 million at the end of 2024. This 10x growth in under six months represents one of the fastest B2B scaling trajectories ever recorded in software.

  Why it matters:

  This extraordinary growth suggests vibe coding platforms are tapping massive pent-up demand from non-technical users wanting to build software. However, the question is whether this represents a one-time land grab of new users or sustainable recurring revenue. The speed itself may indicate market saturation approaching rapidly, which we've seen in other AI wrapper categories covered in our report covering the AI Wrapper market.

  Source: Growth Unhinged

• 4. AI tool trust dropped from 77% to 72% despite rising use

  What happened:

  Stack Overflow's 2024 developer survey showed vibe coding tool favorability declined from 77% in 2023 to 72% in 2024, even as usage increased from 44% to 62% of developers. Only 43% trust AI tool accuracy, and this trust metric has barely moved despite massive improvements in models.

  Why it matters:

  This is a critical warning signal because developers are using tools they increasingly distrust. This suggests either resignation to inevitability or that pressure from management is driving adoption rather than developer satisfaction. Declining sentiment during rising adoption is historically associated with unsustainable trends and eventual backlash.

  Source: Stack Overflow


In our 200+-page report on AI wrappers, we'll show you the real challenges upfront - the things that trip up most founders and drain their time, money, or motivation. We think it will be better than learning these painful lessons yourself.

• 5. 66% spend more time fixing vibe-generated code issues

  What happened:

  Stack Overflow's survey revealed that 66% of developers report spending more time debugging and fixing AI-generated code that is "almost right, but not quite" compared to writing code themselves. This represents the largest single frustration with AI coding tools among professional developers.

  Why it matters:

  This data point reveals the hidden cost of AI productivity gains that marketing materials don't discuss. If two-thirds of developers are spending more time on fixes, the net productivity gain may be far smaller than advertised. This suggests we're in a transition period where tools create new types of work rather than eliminating existing work, potentially limiting long-term sustainability.

  Source: Stack Overflow

• 6. 45% of AI-generated code contains security vulnerabilities

  What happened:

  Veracode's 2025 GenAI Code Security Report analyzed 80 coding tasks across 100+ LLMs and found that 45% of AI-generated code introduced security vulnerabilities from the OWASP Top 10, which represents the most critical web application security risks. Java had a particularly bad 72% security failure rate.

  Why it matters:

  This represents a massive technical debt accumulation that could undermine long-term sustainability. Security vulnerabilities compound over time and are expensive to fix retroactively. If nearly half of AI code is insecure, organizations may face a reckoning when security incidents mount, potentially triggering regulatory responses that limit AI coding tool usage.

  Source: Veracode

• 7. Vibe coding security failures stayed flat from 2023-2025

  What happened:

  While AI models dramatically improved at generating syntactically correct code, Veracode found security performance remained flat across all model sizes and release dates from 2023 to 2025. Newer, "smarter" models were no better at writing secure code than older versions, despite massive improvements in other capabilities.

  Why it matters:

  This is perhaps the most damning finding for long-term sustainability of vibe coding. It suggests security vulnerabilities are a fundamental limitation of current AI approaches, not a temporary problem that will be solved by better models. This could create an insurmountable barrier to enterprise adoption at scale if regulators or insurers intervene, especially as security incidents accumulate.

  Source: Veracode

• 8. Code refactoring collapsed 60% since AI assistants rose

  What happened:

  GitClear analyzed 211 million changed lines of code from 2020 to 2024 and found that code refactoring (moving or reusing code) plummeted from 25% of changed lines in 2021 to less than 10% in 2024. This represents a 60% decline in a fundamental software engineering practice.

  Why it matters:

  Refactoring is how technical debt is controlled in software systems. This massive decline suggests vibe coding tools are optimizing for short-term speed at the expense of long-term maintainability. If this trend continues, codebases will become increasingly difficult to maintain, potentially creating a crisis point where maintenance costs exceed new development costs and force organizations to reconsider their AI adoption strategies.

  Source: GitClear

• 9. Copy-pasted code surged 48% with vibe coding tools adoption

  What happened:

  The same GitClear study found that copy-pasted code increased from 8.3% to 12.3% of changed lines between 2021 and 2024, a 48% relative increase. For the first time in software history, 2024 saw copy-pasted lines exceed moved or refactored lines, violating the fundamental "Don't Repeat Yourself" principle.

  Why it matters:

  This represents a fundamental shift away from software engineering best practices toward "hacking" approaches. Copy-paste code creates maintenance nightmares because bugs must be fixed in multiple places, and changes must be synchronized across duplicated sections. This trend is unsustainable for any codebase intended to last more than a few years, suggesting a major quality crisis brewing beneath the surface.

  Source: GitClear


In our 200+-page report on AI wrappers, we'll show you the ones that have survived multiple waves of LLM updates. Then, you can build similar moats.

• 10. Vibe-assisted code churn projected to double by year-end

  What happened:

  GitClear found that "code churn" (code that is revised or deleted within two weeks of being written) increased from 3.1% in 2020 to 5.7% in 2024, with projections suggesting it could double the 2021 baseline by year-end. This indicates code is increasingly being written incorrectly or incompletely the first time.

  Why it matters:

  Doubling code churn means developers are spending twice as much time fixing recent mistakes. This directly contradicts productivity gain claims and suggests vibe coding tools are creating a cycle of rapid low-quality output followed by expensive correction. If churn continues rising, it could erase all productivity gains and make the trend unsustainable, which mirrors patterns we've documented in our market clarity report covering AI Wrappers.

  Source: GitClear

• 11. AI coding assistant completed tasks 55.8% faster in study

  What happened:

  A 2023 peer-reviewed randomized controlled trial published by researchers from MIT and GitHub found developers using Copilot completed a JavaScript HTTP server implementation task 55.8% faster than the control group (71 minutes versus 161 minutes, with statistical significance of p=0.0017).

  Why it matters:

  This is the strongest scientific evidence for AI coding productivity gains because the controlled methodology and peer review lend credibility. However, the task was relatively simple and well-documented online, and speed doesn't account for quality, maintenance, or debugging costs. Real-world gains are likely smaller, but this establishes a floor for potential benefits.

  Source: arXiv

• 12. Enterprise AI tool users showed 8.69% more pull requests

  What happened:

  An Accenture enterprise study with randomized controlled trials found developers using GitHub Copilot had 8.69% more pull requests, 15% higher merge rates, and 84% increase in successful builds. Additionally, 90% felt more fulfilled in their jobs and 95% enjoyed coding more.

  Why it matters:

  The Accenture results are more modest than lab studies but show real enterprise productivity gains. The satisfaction metrics are particularly important for retention. However, the 8.69% pull request increase is far below the 55% speed claims, suggesting real-world complexity limits gains. The sustainability depends on whether quality holds up over time as technical debt accumulates.

  Source: GitHub

• 13. 84% of developers using or planning to use vibe coding tools

  What happened:

  Stack Overflow's 2025 survey showed 84% of developers are using or planning to use AI coding tools, up from 76% in 2024 and 70% in 2023. This represents consistent year-over-year growth in both adoption intentions and actual usage among professional developers.

  Why it matters:

  The continued acceleration in adoption despite declining satisfaction is puzzling and suggests organizational pressure or fear of missing out driving adoption. However, crossing 80% adoption indicates market maturity and suggests vibe coding tools are becoming standard rather than experimental. The question is whether this represents genuine value or a bubble that will deflate.

  Source: Stack Overflow

• 14. 75% of engineers predicted to use AI code assistants by 2028

  What happened:

  Gartner predicts 75% of enterprise software engineers will use AI code assistants by 2028, up from less than 10% in early 2023. This represents the fastest enterprise adoption curve of any developer technology in history, faster than Git (5 to 7 years) and Docker (6 to 7 years).

  Why it matters:

  Gartner's forecast, if accurate, suggests vibe coding is here to stay in some form. However, Gartner has been wrong before on adoption curves. The prediction assumes no major quality crises, security incidents, or regulatory interventions. The speed of forecasted adoption (10% to 75% in 5 years) itself suggests we may be in a hype cycle peak.

  Source: Gartner

• 15. Vibe coding tools market projected to hit $26.03B by 2030

  What happened:

  The vibe coding tools market is projected to grow from $4.86 billion in 2023 to $6.11 billion in 2024, reaching $26.03 billion by 2030. This represents a 27.1% compound annual growth rate, which would make it one of the fastest-growing enterprise software categories.

  Why it matters:

  The market size projections suggest massive economic opportunity and indicate analyst confidence in sustainability. However, these forecasts were made during peak AI hype and assume continued exponential growth. If quality issues emerge or productivity gains prove illusory, actual growth could be far lower, making current valuations unsustainable.

  Source: Grand View Research


In our 200+-page report on AI wrappers, we'll show you the real user pain points that don't yet have good solutions, so you can build what people want.

• 16. Vibe coding startup valuations hitting 25-70x revenue multiples

  What happened:

  AI coding startups are commanding valuation multiples of 25 to 70x annual recurring revenue, compared to typical SaaS multiples of 10 to 15x. Cursor at $9.9 billion on $500 million annual recurring revenue is roughly 20x, while at earlier stages companies were valued even higher relative to revenue.

  Why it matters:

  These extreme multiples indicate investor belief that vibe coding represents a category-defining shift. However, they also suggest frothy valuations driven by fear of missing out and fear of missing transformative technology. These multiples are unsustainable if growth slows or competitive moats fail to materialize, suggesting significant downside risk for late-stage investors.

  Source: TechCrunch

• 17. AI coding assistant GitHub Copilot hit $400M revenue in 2024

  What happened:

  Satya Nadella stated that GitHub Copilot reached $400 million in annual recurring revenue in 2024 and is now "a larger business than all of GitHub was when we acquired it" in 2018. GitHub was acquired for $7.5 billion, suggesting Copilot alone could be worth more than that today.

  Why it matters:

  This validates that vibe coding tools can generate massive revenue at scale, not just user adoption numbers. It suggests developers and enterprises are willing to pay premium prices ($10 to $19 per month individual, $39 per month enterprise), which supports sustainability. However, it also raises questions about whether GitHub can defend market share against Cursor and other competitors.

  Source: CIO Dive

• 18. 90% of Fortune 100 deployed vibe coding tool Copilot

  What happened:

  Microsoft reported that 90% of Fortune 100 companies have deployed GitHub Copilot. This represents unprecedented enterprise adoption for a tool that launched to general availability only in 2022. Over 50,000 organizations total are using the platform.

  Why it matters:

  Enterprise penetration at Fortune 100 level suggests vibe coding has moved beyond early adopters to mainstream business-critical deployment. Large enterprises are notoriously conservative, so 90% adoption indicates confidence in the technology. This is one of the strongest signals for sustainability, though it doesn't reveal depth of usage or return on investment.

  Source: TechCrunch

• 19. AI assistant license utilization reached 80% vs typical 30-40%

  What happened:

  GitHub Copilot shows 80% license utilization once deployed in enterprises, meaning 80% of employees with licenses actively use the tool. This compares to typical enterprise software utilization rates of 30 to 40%, indicating significantly higher engagement and stickiness than most enterprise tools.

  Why it matters:

  High utilization rates are one of the strongest indicators of genuine value and sustainability. When enterprises buy software licenses, low utilization is common because employees avoid poor tools. 80% utilization suggests Copilot is actually solving problems developers face daily, making it sticky and likely to renew, which is a pattern we've observed across successful AI wrappers in our report to build a profitable AI Wrapper.

  Source: Second Talent


In our 200+-page report on AI wrappers, we'll show you dozens of examples of great distribution strategies, with breakdowns you can copy.

• 20. 67% of AI coding assistant users active 5+ days weekly

  What happened:

  Among GitHub Copilot enterprise users, 67% are active 5 or more days per week, 81.4% install the IDE extension the same day they receive their license, and 96% start accepting suggestions the same day. These metrics indicate immediate adoption and sustained engagement beyond initial experimentation.

  Why it matters:

  These behavioral metrics are extremely strong signals of product-market fit. Same-day installation and usage, plus 5+ days weekly activity, suggests the tool quickly becomes part of daily workflow rather than sitting unused. This level of engagement makes adoption "sticky" and much more likely to be sustainable long-term, even if competitors emerge with better features.

  Source: GitHub

• 21. 80% of companies using third-party vibe coding tools

  What happened:

  JetBrains' 2024 developer survey found that 80% of companies are using third-party AI coding tools, and 85% of individual developers report using AI coding assistants in their work. This indicates widespread adoption across both organizational and individual levels.

  Why it matters:

  The JetBrains data provides independent confirmation of high adoption rates beyond GitHub's self-reported metrics. The convergence of company-level (80%) and developer-level (85%) adoption suggests both top-down mandates and bottom-up enthusiasm are driving usage, which typically indicates sustainable adoption rather than pure hype or management pressure alone.

  Source: JetBrains

• 22. 85% of students using AI coding assistants

  What happened:

  Stack Overflow's survey found 85% of students are using AI coding tools, a higher adoption rate than among professional developers. Major universities including Stanford, UC San Diego, and NUS are integrating AI coding tools into their curricula, with free access for students since 2022.

  Why it matters:

  Student adoption suggests vibe coding will be "native" to the next generation of developers, similar to how millennials were "mobile native." This demographic inevitability makes the trend harder to reverse even if current tools prove flawed. However, it also raises concerns about students learning to code without understanding fundamentals, which could create knowledge gaps.

  Source: Stack Overflow

• 23. 51% of professionals using vibe coding tools daily

  What happened:

  Stack Overflow's data shows 51% of professional developers use AI coding tools on a daily basis, indicating the tools have become integrated into regular workflow rather than occasional experimental use. This represents a significant milestone in habit formation and routine practice adoption.

  Why it matters:

  Daily usage by a majority of professionals suggests vibe coding has crossed the "habit threshold" where it becomes part of routine practice. Habits are sticky and hard to break even if alternatives emerge. This indicates the trend has momentum, though it doesn't prove the value proposition is sustainable long-term or that the quality issues won't eventually cause backlash.

  Source: Stack Overflow

• 24. Vibe coding assistant Cursor hit 1M+ daily users in 16 months

  What happened:

  Cursor reached over 1 million daily active users within just 16 months of its launch, with 360,000 paying customers. For context, Slack took 2+ years to reach 1 million daily actives. This represents one of the fastest consumer-to-enterprise adoption curves in software history.

  Why it matters:

  The speed to 1 million daily actives suggests Cursor found genuine product-market fit extremely quickly. However, the question is whether this represents displacement of GitHub Copilot (zero-sum competition) or market expansion. If it's displacement, it suggests winner-take-all dynamics may favor best-in-class solutions, making the space volatile and potentially unsustainable for most players.

  Source: Bloomberg

• 25. 37-40% of AI assistant Claude conversations are coding tasks

  What happened:

  Anthropic's Claude AI assistant reports that 37 to 40% of all conversations involve coding or mathematical tasks, making it one of the most popular use cases for the general-purpose model. Claude has 16 to 18.9 million monthly active users.

  Why it matters:

  The fact that coding represents 37 to 40% of all Claude usage (which isn't even coding-specific) demonstrates massive organic demand for vibe coding help beyond specialized tools. This suggests the market is much larger than just professional developers and could include analysts, scientists, and hobbyists, supporting long-term sustainability and growth beyond traditional software engineering roles.

  Source: Anthropic


In our 200+-page report on AI wrappers, we'll show you the best conversion tactics with real examples. Then, you can replicate the frameworks that are already working instead of spending months testing what converts.

• 26. 88% of AI-generated code retained in editor

  What happened:

  GitHub's research found that developers retain 88% of Copilot-generated characters in their code editor, meaning only 12% is deleted or significantly modified. This high retention rate suggests developers find the suggestions useful and mostly correct on first generation.

  Why it matters:

  High retention rates are a proxy for quality and usefulness. If developers were deleting most suggestions, it would indicate poor quality and frustration. 88% retention suggests the code is "good enough" to keep most of the time, supporting productivity claims. However, this doesn't account for bugs or security issues that may not be caught until later in the development cycle.

  Source: GitHub

• 27. Top 10 vibe coding startups raised $2B+ in 2024

  What happened:

  In 2024 alone, the top 10 vibe coding startups raised over $2 billion in venture capital funding, including Cursor's $900 million round, Codeium's raises toward a $3 billion valuation, and multiple other nine-figure rounds. This represents one of the largest capital deployments in any enterprise software category.

  Why it matters:

  Massive VC investment indicates sophisticated investors believe vibe coding is a generational opportunity. However, it also creates pressure for rapid growth and eventual exits, potentially driving unsustainable practices. The capital influx could fuel a competitive race that benefits users short-term but leads to consolidation and higher prices long-term.

  Source: Crunchbase

• 28. Vibe coding tool Codeium valuation grew 6x in 13 months

  What happened:

  Codeium saw its valuation rise from $500 million in January 2024 to $1.25 billion in August 2024, with reports of discussions at approximately $3 billion in February 2025. This represents a 6x increase in just 13 months.

  Why it matters:

  Codeium's valuation trajectory, while less extreme than Cursor's, still shows extraordinary growth that suggests genuine market demand. However, the speed of valuation increases raises questions about whether fundamentals support these prices or if we're in a bubble. The rapid appreciation across multiple companies suggests sector-wide exuberance rather than company-specific performance.

  Source: TechCrunch

• 29. 41% of developers use AI tool ChatGPT for coding

  What happened:

  Stack Overflow found that 41% of developers use ChatGPT for coding tasks, making it the most popular AI tool among developers, ahead of GitHub Copilot and other specialized coding assistants. This despite ChatGPT being a general-purpose tool not specifically designed for coding.

  Why it matters:

  The fact that a general-purpose chatbot is the most-used AI coding tool suggests developers value versatility and natural language interaction over specialized features. This raises questions about the moats of specialized vibe coding tools: if ChatGPT or Claude can do "good enough" coding while also handling other tasks, why pay for a specialized tool? This could limit long-term sustainability of coding-only products.

  Source: Stack Overflow


In our 200+-page report on AI wrappers, we'll show you what successful wrappers implemented to lock in users. Small tweaks that (we think) make a massive difference in retention numbers.

• 30. AI models show 86% failure rate on XSS prevention

  What happened:

  Veracode's analysis found that AI models failed to defend against Cross-Site Scripting vulnerabilities (CWE-80) in 86% of relevant code samples. XSS is one of the most common and dangerous web application vulnerabilities, and has been well-documented for decades in security literature.

  Why it matters:

  The 86% failure rate on a well-known vulnerability type is particularly alarming because it suggests AI models aren't learning from decades of security best practices available in training data. This isn't an edge case but rather fundamental security. If AI can't handle basic security patterns, it calls into question whether current architectures can ever be trusted for production code without extensive review.

  Source: Veracode

In our 200+-page report on AI wrappers, we'll show you which areas are already overcrowded, so you don't waste time or effort.