Is vibe coding getting too crowded?

Has vibe coding already become mainstream?

Vibe coding is already mainstream enough that we can no longer treat it as a niche way for developers to experiment with AI.

The clearest evidence is the volume of people building. Lovable currently says it hosts around 60 million projects attracting roughly 900 million visits per month. Earlier this year, the company reported that people were creating about one million new projects every week.

Lovable’s own survey of more than 14,000 users is even more revealing. Four in five respondents identified with non-technical roles, while nearly two-thirds worked outside technology. Founders were the largest group, but designers, salespeople, marketers, operations workers and consultants were all using the platform too.

Emergent has seen much the same pattern. The company reported more than six million users across 190 countries within its first year, with roughly 70% saying they had no previous coding experience. By the time Emergent crossed $100 million in annual revenue run rate, users had already created more than seven million applications.

Traditional developers are also using AI much more heavily. GitHub’s latest Octoverse counted more than 180 million developers, up by over 36 million in one year, while about 80% of new GitHub users tried Copilot during their first week.

The big change is who can realistically make software. A salesperson with an idea, a restaurant owner who needs an internal tool or a consultant who understands a client workflow can increasingly build something themselves instead of finding a technical co-founder or paying an agency.

Recent measure	Scale
Lovable projects hosted	~60 million
Lovable project traffic	~900 million visits/month
Emergent users	6+ million
Emergent apps created	7+ million
GitHub developers	180+ million
New GitHub developers using Copilot in week one	~80%

Are people creating vibe-coded apps faster than anyone can use them?

Yes. The supply of new software is now growing so quickly that generic vibe-coded apps are almost certainly being created faster than attention for them can grow.

Lovable alone previously reported roughly one million new projects a week. GitHub recorded 121 million new repositories during 2025, equivalent to around 230 every minute. Those figures include experiments, private projects and unfinished software, so they do not represent millions of new SaaS competitors. They still show how radically the cost of creating software has fallen.

Microsoft gave us an unusually fresh confirmation of the same pressure from a completely different part of the market. In a recent update about the Edge extension store, Microsoft said AI-assisted coding had made extensions easier to build and that submission volumes had continued rising enough to put extra strain on its review pipeline.

That is useful because the effect is reaching distribution infrastructure, not just survey data from companies selling coding tools.

Attention has not multiplied at the same rate. Lovable currently hosts around 60 million projects and says those projects receive roughly 900 million monthly visits. Dividing one by the other gives about 15 visits per project, although the real distribution will obviously be extremely uneven. A relatively small group of apps will capture a large share of those visits while millions receive almost none.

This is where the crowding becomes real. Making something has become cheap enough that publishing another app barely tells us anything about whether anybody wants it.

Are all the obvious vibe-coded app ideas already overcrowded?

A lot of the obvious ideas are overcrowded today, especially when the entire pitch can be understood from one sentence.

Another AI meeting summarizer, habit tracker, invoice tool, generic CRM, content generator, workout planner or productivity dashboard enters a market where competitors can reproduce most visible features very quickly.

The problem goes beyond the number of competitors. These apps increasingly start from the same ingredients: similar frontier models, React interfaces, Stripe payments, Supabase-style backends, standard authentication and the same familiar dashboard layouts.

A founder could once spend several months turning a fairly obvious SaaS idea into a working product. That development time created a small barrier in itself. Lovable, Base44, Replit or Emergent can now compress much of that work into days, sometimes hours for the first usable version.

“Already taken” is still too strong. Calendly entered a world full of calendars. HubSpot entered a world with Salesforce. Better execution, positioning and distribution can still create very large companies in busy categories.

What has become much less valuable is simply noticing the category.

“AI invoicing for freelancers” tells us almost nothing anymore. Knowing why a particular type of construction subcontractor loses money during invoicing, which documents cause the problem, who approves them and which accounting software needs the final data is far more useful.

Vibe coding has made obvious ideas abundant. Specific problems remain much harder to find.

Has vibe coding made copying competitors too easy?

Yes. Vibe coding has made visible software features dramatically easier to copy, so a polished interface or a long feature list offers much less protection than it used to.

Look at what modern builders can already generate. Authentication, payments, databases, dashboards, email notifications, CRUD workflows, AI integrations and deployment can all be assembled with far less manual engineering.

The platforms keep absorbing more of the difficult infrastructure too. Lovable has expanded its authentication, security, integrations and deployment stack. Base44 now controls more of its own backend and AI infrastructure. Emergent sells itself to small businesses as something close to an engineering team in a box.

A competitor therefore has fewer pieces to recreate independently.

Anything visible from the outside is particularly vulnerable. A founder can inspect a homepage, create an account, walk through an onboarding flow and ask an AI builder to reproduce the basic workflow. The copy may initially be worse, but the engineering cost of trying has collapsed.

Customer history, proprietary data, difficult integrations, regulatory approvals, operational know-how and an existing audience behave very differently. Those advantages accumulate outside the interface.

These days, “we built this feature first” is becoming a weak position surprisingly quickly.

If software becomes cheap to build, what is actually valuable now?

As vibe coding makes implementation cheaper, the valuable parts of a software business are moving toward customer access, domain knowledge, data, trust and difficult workflows.

We can see this in the people adopting these tools. Lovable says nearly two-thirds of its users come from outside technology, across industries including retail, healthcare, finance, education and real estate. Emergent says small businesses account for around 40% of its users.

Those people often arrive with something professional developers do not have: firsthand knowledge of an annoying process.

Emergent has described trucking companies building shipment-tracking systems, construction businesses creating ERP-style tools, factories developing internal software and property managers building their own customer-management systems. These are very different opportunities from launching the hundredth public AI note-taking app.

The coding may actually be the easy part of those products. Understanding how a trucking company handles exceptions, who needs access to a particular document and why the existing spreadsheet keeps failing requires experience inside the workflow.

That gives domain experts an unusual advantage. A dentist who deeply understands a dental-office problem can now compete with a developer who knows JavaScript but has never worked inside a dental practice.

The scarce resource is increasingly knowing what should happen inside the software.

Are Lovable, Base44, Replit and Emergent already fighting over a saturated market?

The vibe-coding platform market is crowded, but current revenue growth still looks far too strong for us to call it saturated.

Lovable said it passed $500 million in annualized revenue earlier this year. More recently, the company raised $400 million at a $13.3 billion valuation and said it was hosting around 60 million projects.

Base44 provides an even fresher test. The company crossed $100 million ARR, reached $150 million shortly afterward and then passed $200 million only five months after the $100 million milestone. Base44 also says it now has more than 10 million users.

Emergent reached roughly $120 million in annual revenue run rate after growing about 70% over four months, according to the company’s latest fundraising disclosures. It also reported more than 200,000 paying customers.

Replit has not published a newer revenue figure since reporting roughly $150 million in annualized revenue, so we should not pretend we know its current ARR. Investors nevertheless valued Replit at $9 billion when it raised $400 million earlier this year, three times the valuation from its previous round six months earlier.

Adding the latest disclosed run rates for Lovable, Base44 and Emergent alone gets us to roughly $820 million. Including Replit’s older $150 million milestone pushes the combined figure close to $1 billion, although the timing differs enough that this should only be treated as an order-of-magnitude comparison.

Several major platforms are still expanding together. The category is brutally competitive, but demand has so far been growing fast enough to support several winners.

There is an important caveat. Strong platform revenue proves that people will pay to build software. It says much less about how many of the resulting apps will ever make money. The tool seller gets paid even when the customer’s startup fails.

Platform	Latest useful business figure
Lovable	~$500M annualized revenue
Base44	>$200M ARR
Emergent	~$120M annual revenue run rate
Replit	~$150M annualized revenue at last disclosure
Replit latest primary valuation	$9B

Can a vibe-coded app still make $10,000 a month?

Yes. A vibe-coded app can absolutely still reach $10,000 a month, but simply launching something competent is much less likely to get you there than it was when software itself was harder to produce.

The customer counts required are actually quite small. At $49 per month, $10,000 MRR takes about 204 customers. At $199, it takes around 51. A specialized B2B product charging $500 per month needs 20.

Twenty customers is not a huge market. Neither is 51.

That is why crowding does not kill the small-software opportunity. A founder can make a meaningful living from a problem that would be irrelevant to a large venture-backed SaaS company.

What has become harder is acquiring those customers without a clear reason for them to choose you.

Lovable’s user survey helps explain why. More than half of respondents said they were building a business, while roughly another quarter hoped to monetize a side project. There are therefore huge numbers of builders chasing commercial outcomes at the same time.

The better odds are increasingly found in markets where the founder can describe the first customers before finishing the product. Maybe they already work in the industry. Maybe they run an agency serving those customers. Maybe they belong to a professional community where the problem keeps coming up.

Vibe coding has made reaching $10,000 MRR technically cheaper. Finding the right 20, 50 or 200 customers remains the real test.

Is distribution now harder than building the app?

Yes. Distribution is becoming the main bottleneck for many vibe-coded startups because generating a decent product has become much faster than earning attention for it.

This changes who has the advantage.

A few years ago, somebody with a large newsletter, a deep professional network or excellent sales skills could still be blocked by the inability to build software. They might need a technical co-founder, outside developers or significant upfront capital.

That barrier has shrunk sharply.

Lovable’s own data shows founders, designers and salespeople among the fastest-growing groups on the platform. Those people can bring skills that were previously downstream from product development directly into the founding team from day one.

A consultant can turn a service into software. An agency can build a product for a problem it sees across clients. A salesperson who already knows 100 buyers can create a simple tool for them.

Faster coding also changes how founders should use their time. If an MVP that once took eight weeks can be built in a few days, spending the next seven weeks adding features wastes much of the advantage. The better use is getting the product in front of customers, seeing where they stop, hearing what they refuse to pay for and rebuilding from there.

The speed advantage becomes valuable when it produces faster learning. Otherwise we simply end up with more software.

Are niche B2B apps the best place left for vibe coding?

Niche B2B software is probably one of the strongest opportunities in vibe coding right now because many business problems are too small and awkward for traditional SaaS companies but large enough for one or two founders.

Traditional SaaS economics pushed founders toward markets capable of supporting large teams, expensive engineering and venture returns. A workflow used by 500 specialized businesses might never have justified building a company around it.

That calculation changes when one person can build and operate far more software.

Imagine a tool with 100 customers paying $300 per month. That produces $360,000 in annual recurring revenue. It would barely register inside Salesforce, but it could support a very attractive small business.

The best niches also tend to be less visible. Anyone browsing Product Hunt can discover that people want AI writing tools. Discovering how insurance brokers manually reconcile a particular document or how a regional logistics company handles damaged shipments usually requires being much closer to the industry.

Emergent’s customer mix supports this direction. The company says many small businesses use its platform to replace workflows that previously lived in spreadsheets, email or messaging apps.

Replit and Lovable are moving further into workplace software too. Replit has discussed adoption inside companies such as Visa, while Lovable has expanded enterprise governance, integrations and security controls.

The interesting opportunity is often hiding in a spreadsheet somebody hates opening every Monday morning.

Will companies just build their own software instead of buying SaaS?

Companies will increasingly build simple internal software themselves, but mature SaaS products with deep integrations, accumulated data and important workflows still have a strong reason to exist.

The first category is already easy to imagine. A team needs a small approval dashboard, a pricing calculator, a staff portal or a custom reporting tool. Buying a full SaaS product may involve unnecessary features, per-seat pricing and a workflow that never quite fits.

Vibe coding makes custom software much more competitive in those cases.

This creates a genuine threat for lightweight SaaS whose whole product can be recreated from a short description. A simple internal tracker that costs a company $5,000 per year becomes harder to defend when somebody can generate a tailored alternative in an afternoon.

Replacing Salesforce, ServiceNow, Stripe or a mature accounting system is a very different proposition. Years of data, permissions, integrations, compliance work and employee habits sit inside those systems.

The more value a product accumulates while customers use it, the safer it becomes.

We should therefore expect the squeeze to be strongest around small, generic tools with shallow workflows. Custom internal apps can eat into that market without replacing complex SaaS as a whole.

Is trust becoming the new bottleneck for vibe-coded apps?

Yes. As vibe-coded software becomes easier to generate, proving that an app is secure, reliable and safe enough to use is becoming more valuable.

A weekend project can tolerate rough edges. A business storing customer records, processing payments or controlling an operational workflow cannot.

The major builders are responding accordingly. Lovable has added security scanning, governance controls, publishing permissions and more enterprise-focused infrastructure. Base44 has pushed deeper into its own backend and model stack. Replit is expanding enterprise integrations and controls.

There is a practical reason for all of this. AI can generate thousands of lines of plausible code much faster than most users can inspect those lines. Non-technical builders are especially dependent on the platform catching problems they would never know to look for.

GitHub’s latest developer research points in the same direction from the professional side. As AI writes more code, developers describe their work increasingly in terms of understanding, directing and verifying what the agent produces.

That verification step becomes much more important as the cost of generation approaches zero.

A crowded market full of shaky products can actually help builders who earn trust. Security certifications, a strong reputation, reliable support and years of clean operation become easier for buyers to appreciate once they have encountered enough bad AI-generated software.

Are app stores starting to push back against vibe coding?

Yes. App stores and extension marketplaces are already becoming a practical constraint as AI makes software submission much easier.

Apple gave us the clearest example earlier this year. It blocked updates to vibe-coding apps including Replit and Vibecode over concerns involving code execution rules. Anything was removed from the App Store twice before changing its approach.

Apple said it was applying existing rules rather than banning vibe coding itself. The distinction matters legally, but from a builder’s perspective the result is similar: generating an app does not mean the platform will allow that app to behave however the builder wants.

Microsoft has now provided a second example from browser extensions. The Edge team recently said AI-assisted coding was enabling developers to build extensions faster, while growing submission volume was putting additional pressure on review turnaround times. Microsoft responded by changing how it handles faster reviews and quality recognition.

Cheap software creation has a second-order effect here. Human review systems, security checks and platform policies were designed for a world where far fewer people could produce software quickly.

Vibe coding can increase supply faster than those systems expand.

That does not stop the market from growing, but it creates another hurdle between “I generated an app” and “people can safely install it.”

Are vibe-coded apps all going to look the same?

A lot of vibe-coded apps already look similar, especially at the first-draft stage, but the sameness becomes much less important once products contain unique workflows and real customer history.

The visual convergence is easy to see. Similar prompts produce familiar landing pages, cards, sidebars, gradients, dashboards and onboarding flows because many products draw from the same design conventions and component libraries.

Emergent’s founder has openly discussed this weakness in AI-generated products. Anyone who spends time browsing newly generated apps can see it too.

The bigger issue is functional sameness. If two founders ask for “an AI CRM for freelancers,” the builders have enough shared conventions to produce surprisingly similar starting points.

That convergence becomes weaker as real usage enters the product. One company connects unusual data sources. Another learns that customers require three approval steps nobody expected. One spends a year fixing edge cases around a particular industry. Another develops a dataset competitors cannot access.

After hundreds of those decisions, the original prompt matters much less.

For a new founder, this raises the bar. Good design still helps, but a prettier generated dashboard is unlikely to create much protection on its own.

So, is vibe coding getting too crowded?

Yes, vibe coding is getting crowded very quickly, and the generic end of the market already looks oversupplied. We do not see evidence that the broader opportunity has run out.

The supply numbers are hard to ignore. Lovable hosts around 60 million projects. Emergent users have created more than seven million apps. GitHub added 121 million repositories in one year. Microsoft is seeing enough AI-assisted extension submissions to put pressure on its review process.

Demand is still expanding too. Lovable passed roughly $500 million in annualized revenue. Base44 has since climbed beyond $200 million ARR. Emergent reached around $120 million in annual revenue run rate. Replit raised at a $9 billion valuation after previously reporting $150 million in annualized revenue.

Those two trends can coexist.

There are far too many easy-to-copy dashboards, generic AI wrappers and products whose whole idea fits into one prompt. Competing there has become much tougher because builders have very little technical scarcity left to protect them.

The picture changes once we move into narrow business workflows, proprietary data, difficult integrations, unusual customer groups or markets where founders already have distribution. Millions of these problems remain poorly served because traditional software economics made them unattractive.

Vibe coding has lowered the minimum size a useful software market needs to have.

So the answer is sharper than simply saying the space is “more competitive.” The easy layer really is overcrowded today. If the plan is to browse trending ideas, generate one quickly and hope Product Hunt or social media supplies the customers, the odds have deteriorated badly.

For founders who know a specific market unusually well, however, vibe coding still opens far more doors than it closes. The code has become abundant. Good problems, trusted access to customers and genuine insight into how people work have not.

Part of the market	Our judgment now
Generic AI wrappers	Heavily overcrowded
Simple consumer utilities	Crowded
Copyable horizontal SaaS	Increasingly difficult
Specialized B2B workflows	Still attractive
Internal company software	Growing opportunity
Products with proprietary data or distribution	Much better protected