Is the vibe coding market big or just a hype spike?

Why did vibe coding suddenly become so popular?

Vibe coding has become one of the fastest-growing parts of the software industry because people can now build working applications by describing what they want in ordinary language.

The term took off after Andrej Karpathy described the practice in February 2025. But the technology had already been improving for months. Anthropic's Claude 3.5 Sonnet made a noticeable leap in code generation in 2024, while Cursor, Replit Agent, Bolt.new and Lovable started turning those capabilities into products that could build entire applications.

What changed was the amount of work AI could handle. Earlier coding assistants mainly suggested code or completed individual functions. The newer platforms could generate interfaces, connect databases, fix errors and help users deploy applications. Someone with almost no programming experience could suddenly build a working prototype in an afternoon.

The commercial response came quickly. Lovable reached approximately $100 million in annual recurring revenue by July 2025 and crossed $400 million in February 2026. Replit reinvented its business around AI agents, while newer companies such as Emergent started reporting substantial revenue within months of launching.

That explains why the market attracted so much attention. Vibe coding was making software creation accessible to a much larger audience, and customers were already paying for the convenience.

What should we actually count as the vibe coding market?

The vibe coding market is smaller than the overall AI coding market, and confusing the two makes the opportunity look much bigger than it actually is.

In its original sense, vibe coding means describing an application to an AI system and allowing it to generate most of the code. Lovable, Bolt.new, Replit Agent, Base44 and Emergent are good examples. Their products are designed to help people turn an idea into an application without manually programming every feature.

Cursor, Claude Code, GitHub Copilot and OpenAI Codex serve a partly different market. Professional developers use them to write, review, refactor and maintain software. These tools can generate entire applications, but their customers often remain involved in the underlying engineering.

The distinction becomes important when calculating revenue. Cursor was reportedly generating billions of dollars in annualized revenue by mid-2026, but counting all of that as vibe coding would include substantial spending by professional software engineers.

For this analysis, we focus primarily on natural-language application builders. We include professional AI coding tools when they help explain competition, customer behavior or how much bigger the market could eventually become.

If you want more recent data on this point, please see our latest vibe coding market report.

How big is the vibe coding market today?

The vibe coding market has already passed $1 billion in combined annualized revenue among several leading platforms, although nobody has published a reliable total for the entire category.

According to figures reported by TechCrunch in September 2026, Lovable exceeded $600 million in annualized revenue. Sacra estimated that Replit reached $525 million in April 2026, while Emergent reported passing $100 million ARR in February.

Adding those three figures gives approximately $1.225 billion in annualized revenue. The calculation mixes company-reported results and an external estimate from different dates, and Replit earns some revenue from activities beyond vibe coding. Still, it gives us a useful sense of the commercial scale already reached.

The wider market also includes Bolt.new, which reportedly reached $40 million ARR in March 2025, alongside Base44, v0 and numerous smaller platforms. Some revenue generated by these products is included in larger software businesses and cannot be separated reliably.

We should also be careful with the word ARR. These companies often report annualized revenue based on a recent period of subscriptions and usage. That tells us how much they are generating at a particular moment, rather than how much revenue they have already collected over a full year.

Platform	Revenue benchmark	Measurement	Included in narrow market?
Lovable	Over $600M	September 2026 annualized revenue	Yes
Replit	~$525M	April 2026 Sacra estimate	Partly
Emergent	Over $100M	February 2026 reported ARR	Yes
Bolt.new	~$40M	March 2025 reported ARR	Yes
Base44	Not disclosed separately	Acquired by Wix in 2025	Yes
v0 by Vercel	Not disclosed separately	Part of Vercel's business	Partly
Cursor	Multibillion-dollar run rate	Broader AI coding revenue	Mostly no
GitHub Copilot	Not disclosed separately	Part of Microsoft's business	Mostly no

Are vibe coding startups still growing incredibly fast?

Leading vibe coding startups are still growing at extraordinary rates, although their growth has naturally slowed from the first months when revenue was starting from almost zero.

Lovable provides one of the clearest examples. The company reported approximately $250 million in annualized revenue at the end of 2025, $400 million in February 2026 and $500 million in June. By September, it had exceeded $600 million.

That represents an increase of more than $350 million in annualized revenue within roughly nine months. The growth rate slowed as the revenue base became larger, but the absolute amount of new business remained impressive.

Replit followed a similar trajectory. Sacra estimated that its annualized revenue increased from $300 million at the end of 2025 to $525 million in April 2026, a gain of 75% in four months.

Emergent offers an even younger example. In February 2026, the company told TechCrunch that it had doubled its annualized revenue to $100 million in a single month, approximately eight months after launch.

These companies have reached revenue levels that many traditional SaaS startups take years to achieve. The tougher question is how much of that growth comes from customers who keep building and paying, rather than new users spending heavily during their first experiments.

Are millions of people really using vibe coding?

Millions of people are experimenting with vibe coding, but the headline user counts tell us much less about how many people build applications regularly.

In June 2026, Lovable said users had created more than 50 million projects and were adding approximately one million projects every week. Three months later, the company reported that applications built with Lovable were receiving close to one billion visits per month.

Replit had passed 50 million registered users by March 2026, according to Sacra. Some of those users came from its original coding platform, which existed years before the current AI boom.

Emergent reported more than six million users across 190 countries in February 2026. Among those users, approximately 150,000 were paying customers. That implies a paid-user conversion rate of around 2.5% at the time.

A registered user may never return. A person can generate several projects without deploying any of them. And an application receiving visits may have been built months earlier.

The most encouraging evidence comes from applications that attract actual users outside the coding platforms. Companies still disclose far more information about project creation than about how many projects remain active after six or twelve months.

Are people actually paying for vibe coding, or just trying the free versions?

Customers are spending real money on vibe coding, and the number of paying users is already large enough to support substantial businesses.

Emergent reported approximately 150,000 paying customers in February 2026. Against its $100 million annualized revenue figure at the time, that works out to roughly $667 in annualized revenue per paying customer, or $56 per month on average.

That calculation is an aggregate average, not the price of a typical subscription. Some customers spend considerably more because AI agents charge for the work they perform.

Lovable uses subscriptions and usage credits, while Replit combines subscription pricing with charges for agent activity and other services. Bolt.new also ties consumption to available tokens and paid plans.

This differs from conventional SaaS, where customers often pay a predictable amount for access to a finished product. With vibe coding, customers may spend much more while building an application and much less once the work is complete.

That creates a retention problem. A founder who spends $200 building a product may be delighted with the result but have little reason to keep paying $200 every month.

Customers have demonstrated that they will pay for application creation. Whether they generate enough recurring work to support traditional SaaS-style retention is still an open question.

Are businesses really using vibe coding now?

Businesses are already using vibe coding for internal tools and prototypes, and the early adoption extends well beyond individual hobbyists.

Lovable's co-founder Fabian Hedin said in September 2026 that employees at roughly two-thirds of Fortune 500 companies were using the platform. The company also named Microsoft, Nvidia and Deutsche Telekom among its customers.

The wording deserves attention. Employees using Lovable at a Fortune 500 company does not mean the entire organization has adopted it or signed a large enterprise contract.

Still, there are practical reasons for businesses to experiment. Marketing teams can build landing pages without waiting for developers. Operations employees can create internal dashboards, while sales departments can build simple customer-management tools.

Emergent's customer data gives us a closer look at who is adopting these products. According to its February 2026 disclosures, nearly 40% of users were small businesses, and approximately 70% had no previous coding experience. Many were building software to replace processes previously managed through spreadsheets, emails and messaging applications.

Large organizations may eventually spend much more, although security reviews, procurement and integration with existing systems can make adoption slower than the initial experimentation figures suggest.

Are professional developers actually interested in vibe coding?

Professional developers increasingly use AI coding tools, but most still do not consider vibe coding part of their normal work.

Stack Overflow's 2025 Developer Survey found that 84% of respondents were using or planning to use AI tools in their development process, compared with 76% the previous year. Among professional developers, 51% used AI tools daily.

Yet when asked specifically about vibe coding, approximately 72% said it was not part of their professional workflow, while another 5% strongly rejected the practice.

The distinction remained visible in Stack Overflow's 2026 research, involving more than 30,000 developers across 169 countries. Nearly two-thirds were regularly using AI assistants, but fewer than 10% fully trusted AI-generated answers.

Developers are clearly becoming comfortable using AI to generate and modify code. They are much less willing to hand over responsibility for everything that happens afterward.

That's understandable. An experienced engineer modifying a payment system, an authentication service or a large production codebase has different requirements from someone creating a simple website.

For the vibe coding market, the bigger opportunity may therefore come from bringing new people into software creation rather than persuading professional developers to abandon traditional engineering practices.

If you want more recent data on this point, please see our latest vibe coding market report.

Does vibe coding really make building software faster?

Vibe coding can make creating a new application dramatically faster, but controlled studies show that AI coding tools do not save time on every kind of development work.

A study published in September 2026 involving 30 participants found that vibe coding reduced task-completion time by 27% compared with traditional programming and by 12% compared with AI-assisted coding in the tasks examined.

However, research published by METR in July 2025 found almost the opposite result. In an experiment involving 16 experienced open-source developers and 246 tasks, participants working with AI tools took 19% longer than when they worked without them.

The developers had expected to become faster. Instead, time spent prompting, reviewing output, waiting for responses and correcting mistakes outweighed some of the time saved.

In February 2026, METR reported that newer tools probably improved productivity, but its follow-up experiment could not produce a reliable estimate because participants increasingly avoided working without AI and the measurements were affected by other biases.

These findings make more sense when we look at the tasks. Generating a new interface from a blank page is relatively easy to evaluate. Changing a mature software system requires understanding dependencies, preserving existing behavior and checking for unintended consequences.

Vibe coding's speed advantage is strongest during the first version of an application. Once a project becomes complex, testing, debugging and maintenance take up a much larger share of the work.

Can vibe coding create reliable software people actually use?

Vibe coding can already produce useful applications, but we still cannot assume that software generated through prompts will be reliable enough for sensitive business operations.

The platforms have improved well beyond simple demonstrations. Lovable can generate applications with authentication and databases, Replit can build and deploy full-stack software, and Bolt.new lets users create applications in a browser-based development environment.

The challenge appears when those applications need to operate safely for real customers.

A December 2025 research paper titled "Is Vibe Coding Safe?" examined 200 security-sensitive software engineering tasks. In one tested configuration using SWE-Agent and Claude 4 Sonnet, 61% of solutions were functionally correct, but only 10.5% met the researchers' security criteria.

A separate study published in June 2026 examined real-world vibe-coded applications and found recurring vulnerabilities involving exposed secrets, insufficient input filtering and placeholder security logic.

An application can appear to work while allowing unauthorized access to customer information or exposing private credentials. That's a serious problem for anyone putting generated software in front of paying customers.

The severity depends heavily on the use case. A marketing landing page, a restaurant menu and a small internal calculator can tolerate risks that would be unacceptable in banking, healthcare or customer-account management.

Type of vibe-coded application	Typical difficulty	Main production concern
Landing page	Low	Hosting and basic security
Personal portfolio	Low	Content and account protection
Event website	Low to moderate	Reliability during peak traffic
Internal dashboard	Moderate	Permissions and data access
Customer portal	High	Authentication and data isolation
E-commerce application	High	Payments and customer information
SaaS product	High	Security, maintenance and scaling
Healthcare or financial application	Very high	Compliance, privacy and security

The platforms that make testing, security and maintenance easier could unlock a much larger market than those focused mainly on generating attractive first versions.

Is vibe coding taking customers away from no-code platforms?

Vibe coding is becoming a serious competitor to no-code platforms such as Bubble, Webflow and Glide, particularly when users want more flexibility than visual builders provide.

Traditional no-code tools let users assemble software through predefined components and interfaces. They work well for standardized requirements but can become frustrating when a project needs something unusual.

With conversational application builders, users can describe the change they want and ask AI to implement it. This removes much of the work involved in manually configuring complex workflows.

Established software companies have already reacted.

In June 2025, Wix acquired Base44 for approximately $80 million in initial consideration, with additional contingent payments. The acquisition gave Wix access to natural-language application-building capabilities alongside its traditional website-building products.

Vercel has developed v0 around AI-generated interfaces and deployment, while Replit has expanded from an online coding environment into a platform for users who want to create complete applications.

No-code platforms will probably incorporate more AI-generated development. Some established products will lose customers, but others can combine conversational creation with the reliability, integrations and support they already provide.

We expect the distinction between no-code and vibe coding to become less meaningful as the products converge.

Could vibe coding make the software market much bigger?

Vibe coding could create a much larger market for custom software by making small applications affordable for people who would never have paid developers to build them.

Consider a small business that needs a basic inventory dashboard. Hiring a developer or agency might cost thousands of dollars, making the project difficult to justify. A subscription to an application builder can make an initial version affordable within hours.

The same logic applies to internal reporting tools, customer portals, temporary event applications and specialized workflows that do not justify a traditional software-development project.

There is already evidence that this demand exists. Emergent's February 2026 user disclosures showed that small businesses accounted for a substantial share of adoption, with many users replacing spreadsheets and manual processes.

Vibe coding can therefore capture spending from several places. Some customers will spend less on agencies, freelancers and conventional no-code platforms. Others will build tools that previously existed only as ideas because software development was too expensive.

We consider this expansion effect particularly important. Software demand has always exceeded the capacity of professional developers, especially for small and highly specific applications.

Making an application cheaper to build does not solve customer acquisition, ongoing support or product-market fit, though. Millions of new projects could be created without producing millions of viable software businesses.

The more convincing opportunity is a world where companies and individuals build substantially more software for their own use.

If you want more recent data on this point, please see our latest vibe coding market report.

Can vibe coding companies make good profits after paying for AI?

Vibe coding companies can build profitable businesses, but the cost of AI agents makes their economics less predictable than traditional software subscriptions.

Generating an application requires repeated calls to large language models. Complex tasks can involve planning, writing code, fixing errors, running tests and rebuilding components. Platforms also pay for hosting, databases and other infrastructure.

Replit shows why this matters. Sacra reported that its gross margins fluctuated between approximately 36% and negative 14% during parts of 2025, partly because of the cost of running AI agents. That is a very different economic profile from mature subscription software, where gross margins often exceed 70%.

The company subsequently pushed toward enterprise customers and consumption-based pricing. Sacra's May 2026 analysis pointed to approximately 85% gross margins in higher-value enterprise activity, illustrating how customer mix can change the economics.

Pricing is becoming an important part of the business model. Platforms increasingly charge according to the amount of work performed rather than offering unlimited generation for a fixed monthly fee.

That helps protect margins, but it can also make customers more sensitive to cost. A subscription that feels cheap when building a simple landing page may become expensive when an agent repeatedly fails to fix a complex application.

There is another commercial opportunity after the application has been created. Hosting, databases, security, monitoring and ongoing maintenance can generate revenue without requiring the same volume of expensive code generation.

Sacra's June 2026 analysis of Supabase is revealing here. The database company estimated annualized revenue of around $170 million, growing 221% year over year, as AI coding tools and agents drove demand for backend infrastructure. The report also noted that Lovable, Bolt and Replit had been expanding their own backend offerings.

The competition is moving toward capturing more of the application's lifecycle, where recurring revenue may become more attractive than one-off generation.

Could OpenAI, Anthropic and Google take over the vibe coding market?

OpenAI, Anthropic and Google are serious threats to independent vibe coding startups because the core technology is increasingly available through their own coding agents.

Anthropic has Claude Code, OpenAI offers Codex, and Google has integrated AI development into its wider ecosystem. Microsoft combines GitHub Copilot with Visual Studio Code, GitHub and Azure.

These companies control powerful models and already distribute software to enormous numbers of developers. As coding agents become better at editing files, running tests and building applications, some features currently sold by independent platforms will become standard capabilities.

The pressure is especially strong on products whose main advantage is providing a convenient interface to someone else's AI model.

Yet application builders have their own advantages. Lovable focuses on turning prompts into deployed products, while Replit provides an integrated environment for building, hosting and operating applications. Vercel connects generated interfaces to its established deployment platform.

The difference increasingly comes down to what customers need after the code has been generated.

A professional developer may prefer Claude Code or Codex because the tools fit an existing engineering workflow. A non-technical founder might prefer Lovable because the platform handles more of the setup and deployment process.

Independent application builders will face tougher competition as foundation models improve. Companies that own the customer relationship, simplify deployment and manage applications over time should be better positioned than those relying mainly on code generation.

Are vibe coding startups becoming overvalued?

Vibe coding startup valuations have become extremely ambitious, especially given how little we know about long-term customer retention and profitability.

Lovable raised $400 million in August 2026 at a $13.3 billion valuation, roughly double its $6.6 billion valuation from December 2025. Compared with its later annualized revenue benchmark, that valuation represents approximately 22 times revenue.

Replit raised $400 million in March 2026 at a $9 billion valuation. Based on Sacra's subsequent annualized revenue estimate, the company was valued at around 17 times that revenue level.

These are rough comparisons using figures from different dates, but they show the expectations investors have placed on the category.

A high revenue multiple can make sense for a company growing exceptionally quickly, particularly if investors believe it will eventually produce strong margins. The problem is that the market's early revenue growth tells us relatively little about what customer spending will look like after several years.

Vibe coding also has a relatively low barrier to launching competing products. Companies can use similar foundation models, offer comparable interfaces and compete aggressively on subscription prices.

Meanwhile, the largest technology companies can bundle AI development tools with products their customers already use.

Investors are effectively betting that the leading application builders will become important software platforms rather than interchangeable interfaces for AI models. Some companies may justify those expectations, but the current valuations leave limited room for growth to disappoint.

If you want more recent data on this point, please see our latest vibe coding market report.

What could make the vibe coding boom slow down?

The biggest threat to the vibe coding boom is that customers may continue creating applications without spending enough to support today's revenue growth and valuations.

The first warning sign would be weaker repeat spending. A user might pay for several months while building an application and then cancel once the project is finished. If this happens frequently, platforms need a constant flow of new customers to maintain revenue.

Another risk is frustration with code quality. Stack Overflow's 2025 survey found that 46% of developers distrusted the accuracy of AI-generated output, compared with 33% who trusted it. Around 66% complained about AI solutions that were almost correct but still required fixes.

These problems can become more expensive as applications grow. A project that begins with a simple interface may eventually require database changes, new permissions, integrations and reliable testing. Fixing one part can accidentally break another.

Competition could also reduce revenue per customer. As coding models become cheaper and application builders introduce similar features, customers may switch to lower-priced alternatives.

And the number of projects created could eventually grow much faster than the number of applications people continue using.

We would pay particular attention to paid customer retention, revenue generated after initial application creation, gross margins and the percentage of deployed applications that remain active over time. Those figures would tell us much more about the market's durability than another record number of generated projects.

How big could the vibe coding market become over the next five years?

The vibe coding market could reasonably become a several-billion-dollar annual business, although the available evidence does not support a precise five-year forecast.

Using approximately $1.2 billion in combined annualized revenue from three major platforms as an illustrative starting point, we can see how different growth rates would change the scale of the business.

At 20% annual growth, that revenue base would reach approximately $3 billion after five years. At 35%, it would reach around $5.4 billion. Sustaining 50% annual growth would push the figure above $9 billion.

Those are scenarios rather than forecasts, and the starting figure is a partial company sample, not a complete industry total. They demonstrate that the market could become substantial even if growth slows considerably from its earliest years.

The higher end would require stronger adoption by businesses, more reliable AI-generated applications and meaningful recurring spending on hosting, maintenance and other services.

A more limited outcome would emerge if vibe coding remained primarily a tool for prototypes, small websites and one-off projects. That could still support a multibillion-dollar business, but the opportunity would be much smaller than the broader AI software-engineering market.

We also expect increasing overlap with no-code software, cloud development and conventional AI coding assistants. Over time, the category may become harder to measure separately because conversational application creation will be built into products customers already use.

Is the vibe coding market big or just a hype spike?

The vibe coding market is already big enough to be taken seriously, and its growth is backed by paying customers. The hype comes from assuming that this early success guarantees enormous profits and lasting dominance for today's leading startups.

The numbers establish that there is genuine demand. Several application-building platforms have reached substantial annualized revenue, millions of people are experimenting with software creation, and businesses are beginning to use these tools for practical work.

But enthusiasm is getting ahead of the evidence in some important areas. Professional developers remain cautious about fully delegating engineering work. Security research continues to expose weaknesses in generated applications. And the leading platforms have not disclosed enough information to establish how well customers are retained over several years.

The most promising part of the market is making custom software affordable for people and businesses that previously could not justify building it. That opportunity goes well beyond hobby projects and could support a large industry.

However, the companies that win may look quite different from today's simple application generators. Hosting, secure deployment, databases, testing and maintenance are becoming increasingly important as customers move from creating prototypes to running real software.

Our judgment is that vibe coding is a durable new way of building software, with a genuinely large commercial opportunity. The speculative part is how much of that opportunity today's highly valued startups will ultimately capture.

The technology has already changed who can build software. We still need stronger evidence before concluding that the companies leading this first boom will become equally important, profitable businesses over the long term.

If you want more recent data on this point, please see our latest vibe coding market report.