Which AI coding startup is growing the fastest?

Why is it suddenly so hard to say which AI coding startup is growing fastest?

The AI coding race has split into several different businesses, so the company growing fastest in percentage terms is no longer necessarily the one adding the most revenue right now.

A year ago, Cursor would have been the obvious company to watch. Its annualized revenue eventually passed $4 billion after doubling from $2 billion in only a few months. But SpaceX has since completed its acquisition of Cursor, so we no longer consider it an independent startup.

Meanwhile, Lovable and Replit have turned AI coding into something much broader than a developer tool. Millions of people now use them to create applications without writing much code themselves. Cognition is taking a different route with Devin, selling AI agents into large engineering organizations. Factory is targeting similar enterprise work, while CodeRabbit concentrates on code review.

Then there is Claude Code. It is already bigger than most standalone coding startups by the last hard revenue figure Anthropic disclosed, but Claude Code belongs to a much larger AI company.

So we need to be quite precise about the question. We are looking for the independent AI coding startup whose business is expanding fastest today, rather than the company with the biggest user base, the fastest launch or the most dramatic percentage claim.

What should “growing fastest” actually mean for an AI coding startup?

For an AI coding startup, the best definition of “growing fastest” combines percentage growth with the amount of new revenue being added and the size of the business where that growth is happening.

Percentage growth by itself can be deceptive. Moving from $1 million to $10 million is 900% growth, while moving from $500 million to $900 million is only 80%. Yet the second company has added $391 million more annualized revenue.

Looking only at dollars creates the opposite distortion because the biggest company usually wins. We therefore want to see rapid percentage growth continuing after a startup has already reached meaningful scale.

We also give more weight to actual revenue and customer spending than registrations, app creations, valuations or funding rounds. Those numbers can help explain growth, but none of them proves that customers are paying more.

What we measure	Why we use it	What can go wrong
Percentage revenue growth	Shows how quickly the business is expanding	Tiny companies can produce enormous percentages
Revenue added	Shows the actual size of the expansion	Naturally favors larger companies
Starting revenue base	Tests whether growth is still impressive at scale	Often missing from startup disclosures
Customer spending and usage	Helps check whether revenue growth has real adoption behind it	Public data are much scarcer

If you want more recent data on this point, please see our latest AI code assistant market report.

Does Cursor still count as an AI coding startup?

Cursor should no longer count as an independent AI coding startup because SpaceX has completed its acquisition of Anysphere, the company behind Cursor.

That removes what would otherwise be the strongest contender.

Before the sale, Cursor's growth was extraordinary. Forbes reported annualized revenue above $4 billion in early June, compared with $3 billion in late April and $2 billion in February. Cursor had therefore added more than $2 billion of annualized revenue in a little over three months.

The business was also becoming much less dependent on individual developers. Forbes reported in late May that roughly three-quarters of Cursor's run-rate revenue came from business customers. That is a big change for a product that originally spread largely through individual programmers.

Cursor itself confirmed in mid-August that the acquisition had closed and that the company was now part of SpaceX.

We will keep Cursor as the benchmark. No independent AI coding startup we can verify is currently adding revenue at anything close to Cursor's pre-acquisition pace. But putting Cursor first in a startup ranking today would give the wrong answer to the question.

Is Cognition growing faster than every other AI coding startup right now?

Cognition is currently the strongest answer because its annualized revenue has nearly doubled in roughly one quarter from a base that was already close to half a billion dollars.

Cognition said in late May that its run-rate revenue had reached $492 million. Last week, Bloomberg reported that the company was bringing in more than $900 million on an annualized basis.

That works out to growth of at least 83% in roughly three months and more than $408 million of additional annualized revenue.

The Information independently reported a figure around $900 million, equivalent to roughly $75 million of revenue per month at the latest pace, and said Cognition's run rate had more than tripled since the start of the year.

Customer activity backs up the financial numbers. Cognition said enterprise usage of Devin had increased more than tenfold since the beginning of the year. TechCrunch separately reported that enterprise usage had been growing around 50% month over month for six consecutive months.

Those two statements line up surprisingly well. Six months of 50% compounded monthly growth produces an increase of roughly 11.4 times.

Cognition is also doing this with customers such as Goldman Sachs, Mercedes-Benz, Santander, Citi and NASA. These are organizations that can start with a limited deployment and then spread the product across much larger engineering teams.

Investors have noticed the change. Bloomberg reported last week that Cognition was close to raising roughly $1 billion at a valuation around $47 billion, only a few months after raising at $26 billion. Investor interest in the new round reportedly approached $10 billion.

Funding enthusiasm does not prove growth. Here, though, it follows an unusually large jump in actual revenue.

If you want more recent data on this point, please see our latest AI code assistant market report.

Did Cognition simply buy its growth by acquiring Windsurf?

The Windsurf acquisition gave Cognition a big head start, but it is far too small to explain how quickly Cognition is growing now.

When Cognition acquired Windsurf's remaining business in 2025, Cognition said Windsurf had $82 million of annual recurring revenue, more than 350 enterprise customers and hundreds of thousands of daily users. Windsurf's enterprise ARR was also doubling quarter over quarter.

The Information later reported that Cognition itself was generating revenue in roughly the same ballpark around the time of the transaction. In simple terms, the two businesses together were probably somewhere around the mid-$100 millions in annualized revenue when they came together.

Cognition then reported that the Windsurf deal had more than doubled its ARR and that combined enterprise ARR increased another 30% during the seven weeks after the acquisition.

Windsurf clearly helped. Cognition acquired an established coding environment, hundreds of enterprise accounts, an experienced sales organization and a product that developers were already using every day.

But the combined business has since become several times larger. Even the entire $82 million of ARR that Windsurf brought into the deal is small beside the amount of new annualized revenue Cognition has added lately.

The acquisition accelerated Cognition's expansion. It does not explain away that expansion.

Is Lovable still growing faster than Cognition?

Lovable is still growing remarkably fast today, but Cognition has pulled ahead on the rate at which it is adding revenue.

Lovable's latest confirmed figure was more than $500 million of annualized revenue in June, according to the company and TechCrunch. That was up from roughly $200 million in November 2025.

The company subsequently told investors that it expected to approach a $600 million run rate by the end of August. We should treat that $600 million number as a company target rather than a confirmed result until Lovable publishes a newer milestone.

Even using the confirmed figure, the scale-up is exceptional. Lovable added more than $300 million of annualized revenue in roughly seven months.

The distribution behind it is even harder to ignore. Lovable says people have created more than 60 million projects on the platform, and applications built with Lovable now attract around 900 million visits per month. Employees at nearly two-thirds of Fortune 500 companies reportedly use the product.

Lovable also raised $400 million at a $13.3 billion valuation in August, double its valuation from eight months earlier.

Where Lovable loses the current comparison is the slope. Going from roughly $200 million to $500 million in seven months is enormous. Cognition recently added more annualized revenue over a much shorter interval while starting from a larger base.

Lovable still has a good claim to one of the fastest launches in software history. It just does not appear to be growing fastest today.

Is Replit catching Cognition in AI coding?

Replit is still a serious contender, but the freshest usable revenue evidence does not put Replit ahead of Cognition today.

Replit's turnaround has been spectacular. Before AI Agent changed the business, Replit said it had roughly $2.8 million of annualized revenue. Less than a year later, it had reached around $150 million.

Sacra estimates that Replit then climbed from roughly $300 million at the end of 2025 to $525 million in annualized revenue by April 2026. If that estimate is close, Replit added approximately $225 million in four months, or around 75%.

Usage is already enormous. Replit says it has more than 50 million users, and people inside 85% of Fortune 500 companies use the platform. The company raised $400 million at a $9 billion valuation earlier this year and has publicly talked about reaching a $1 billion revenue run rate.

The $1 billion figure is an ambition, though, rather than a reported current result.

That distinction becomes important in a market moving this quickly. Replit's last widely used revenue estimate is several months older than Cognition's latest number, and it comes from Sacra rather than Replit itself.

Replit could close the gap. For now, we simply do not have fresh enough evidence to say that it already has.

Could Factory actually be growing faster than Cognition?

Factory may have the fastest percentage revenue growth in AI coding, but its disclosure is too incomplete for us to call it the overall winner.

Factory made one of the most striking claims in the entire market when it raised $150 million at a $1.5 billion valuation in April. The company said it had doubled revenue every month for six consecutive months.

Taken literally, six doublings multiply revenue by 64.

Factory also said hundreds of thousands of developers were using its Droids daily across companies including Nvidia, Adobe, EY, Palo Alto Networks and Adyen. Since then, Factory has expanded internationally, launched a partner network backed by a $100 million commitment and continued pushing its Droids deeper into enterprise engineering workflows.

The missing number is the one we need most: revenue.

A 64-fold increase starting from $100,000 produces $6.4 million. Starting from $1 million produces $64 million. Starting from $10 million produces $640 million.

Factory has never publicly given us the starting point needed to resolve that difference.

So Factory could be the percentage-growth winner. Until we know the revenue base, putting it ahead of a company adding hundreds of millions of dollars would require us to fill the most important gap with an assumption.

If you want more recent data on this point, please see our latest AI code assistant market report.

Are Emergent and CodeRabbit growing fast enough to threaten the leaders?

Emergent and CodeRabbit are growing quickly, but neither is large enough yet to challenge Cognition, Lovable or Replit on absolute revenue growth.

Emergent is the larger of the two. TechCrunch reported in July that the startup had reached a $120 million annual revenue run rate, up around 70% over four months, with more than 200,000 paying customers.

That is a strong pace for a company launched only a little over a year earlier. Emergent is also interesting because many of its customers are not traditional software developers. Trucking firms, construction companies, factories and property managers use it to build their own business applications.

CodeRabbit is smaller but growing quickly inside a much narrower job: AI code review. Sacra estimates that CodeRabbit reached roughly $50 million of ARR in July, double its estimated $25 million at the end of 2025. Axios separately reported that revenue had grown more than fivefold year over year when CodeRabbit raised at a $1.5 billion valuation in August.

That fivefold annual growth is faster in percentage terms than several much larger competitors. The dollar increase is still measured in tens of millions rather than hundreds of millions.

Emergent and CodeRabbit belong in the conversation because they show how many parts of software development are producing fast-growing standalone businesses. Neither changes the leader today.

Which AI coding startup is adding the most revenue right now?

Cognition is adding the most annualized revenue among the independent AI coding startups for which we have recent enough numbers to make a useful comparison.

The table also shows why we should resist ranking the companies purely by percentage growth. CodeRabbit can double while adding roughly $25 million of ARR. Cognition can grow by a smaller percentage while adding more than $400 million.

We also need to mark the quality of the figures. Cognition's latest number comes from Bloomberg and The Information. Lovable's $500 million figure was confirmed by the company, while the later figure near $600 million was still a target. Replit and CodeRabbit rely on Sacra estimates.

Cursor would dwarf everyone in the table if we included its final independent months, but its acquisition means we exclude it from the current startup ranking.

AI coding startup	Recent revenue change	Approximate increase	Evidence quality
Cognition	$492M → more than $900M in roughly 3 months	More than $408M	Company disclosure followed by Bloomberg/The Information reporting
Lovable	~$200M → more than $500M in roughly 7 months	More than $300M	Company-confirmed milestones
Replit	~$300M → ~$525M in roughly 4 months	~$225M	Sacra estimate
Emergent	~70% growth to $120M in 4 months	Roughly $49M implied	CEO disclosure reported by TechCrunch
CodeRabbit	~$25M → ~$50M in roughly 7 months	~$25M	Sacra estimate; Axios separately reported more than fivefold YoY growth

Are big companies now deciding who wins the AI coding race?

Enterprise customers are increasingly deciding which AI coding startups can keep growing after the first rush of developers and casual builders.

Cognition is the clearest example. Devin's enterprise usage increased more than tenfold from the start of the year, according to Cognition, while TechCrunch reported roughly 50% month-over-month expansion for six months. Customers include major banks, manufacturers and other companies with very large engineering organizations.

Lovable is also moving upmarket, although its route is different. The platform first exploded by letting founders, designers and non-programmers create software through natural-language prompts. Employees at nearly two-thirds of the Fortune 500 now use Lovable, according to information released around its latest financing.

Replit says its users already reach 85% of the Fortune 500. Its move toward enterprise plans and cloud marketplaces also gives larger organizations an easier way to buy and deploy the product.

Cursor showed how powerful that transition can become before the acquisition. Its business started with individual programmers, yet corporate customers eventually generated most of its revenue.

Large companies can spend far more per account and expand a successful deployment across hundreds or thousands of workers. That is one reason Cognition's current enterprise-heavy growth can add revenue so much faster than a product whose adoption is mostly coming from individual subscriptions.

Does real customer spending back Cognition's growth story?

Yes. The latest independent spending data supports Cognition's acceleration rather than leaving us dependent on Cognition's own numbers.

Brex tracks card and bill-payment activity across more than 35,000 companies and periodically ranks software vendors by a mix of spending momentum, growth velocity and acceleration.

In its spring ranking, Lovable was 14th, Cognition 16th and Replit 17th. The three companies were sitting remarkably close together.

The newest Brex ranking looks different. Cognition has climbed to seventh among all software vendors. CodeRabbit appears 18th. Lovable and Replit no longer appear in the top 25.

We should not turn that into a global market-share ranking. Brex customers skew toward startups and technology companies, and the benchmark measures how quickly spending is changing rather than total industry revenue.

But the timing is useful. Cognition's reported enterprise usage accelerated, its revenue jumped, and spending from an independent pool of thousands of companies moved sharply upward at the same time.

That gives us a much better basis for a current judgment than relying on a funding announcement or a CEO quote alone.

If you want more recent data on this point, please see our latest AI code assistant market report.

Can we really compare all these AI coding revenue numbers?

We can compare AI coding revenue well enough to identify the leader, but the numbers are less standardized than they initially look.

Cognition usually talks about run-rate or annualized revenue. Lovable uses annualized revenue. CodeRabbit's $50 million figure is an estimate of ARR. Replit's latest widely cited figure is also an outside estimate. Factory gives us a growth rate without the underlying dollars.

Annualized revenue also differs from revenue actually earned during the previous 12 months. A company doing $75 million in its latest month can describe that pace as $900 million annualized even if it made much less over the preceding year.

For a fast-growing startup, that gap can be enormous.

We therefore use these figures to measure the current speed of the business rather than pretend they are audited trailing revenue. Repeated observations are much more useful than a single number. If a company's run rate moves from one large figure to another a few months later, something substantial has changed even if the accounting definitions are not perfectly identical.

We also lower our confidence when the number comes from an estimate rather than the company or a well-sourced financial report. That is why Replit's apparent growth deserves attention without giving it the same evidentiary weight as Cognition's fresher figures.

Is Claude Code actually bigger than all these AI coding startups?

Claude Code is already larger than the independent AI coding startups on its last confirmed revenue milestone, but Anthropic is far too broad a company for us to classify it as an AI coding startup.

Anthropic said in February that Claude Code had surpassed $2.5 billion in run-rate revenue and had more than doubled since the start of the year. That figure was already several times Cognition's current scale.

Anthropic itself has continued growing extraordinarily quickly. Reuters reported in August that the company's overall annual revenue run rate had passed $65 billion by the end of July, up from $47 billion in May and roughly $9 billion at the end of 2025. Reuters specifically identified the success of Claude's coding agent as one of the drivers of that enterprise growth.

We do not have an equally clean, recent company disclosure breaking out Claude Code by itself. Third-party trackers estimate a much higher figure today, but those models depend on assumptions about how Anthropic's total revenue should be allocated among products. We would rather keep the hard $2.5 billion milestone than present an estimate as fact.

Claude Code therefore changes how we interpret the market even though it does not change our startup ranking.

Cognition may currently lead the independent startups, while one of its biggest competitors sits inside an AI laboratory generating tens of billions of dollars in annualized revenue. The pressure from Anthropic and OpenAI is much greater than the startup leaderboard alone suggests.

Can Cognition really keep growing this fast?

Cognition will almost certainly slow down from its current percentage growth rate, but it can slow dramatically and still remain one of the fastest-growing AI coding companies.

The mathematics makes that unavoidable.

Growth of 50% every month compounds to roughly 11 times after six months. Extend the same rate for a full year and the business would multiply by about 130. No company can maintain that pace once it reaches billion-dollar scale.

Cognition's revenue expansion has the same problem. As we saw above, its run rate nearly doubled in roughly one quarter. Repeating that several more times would quickly push the business into many billions of dollars.

We should therefore expect deceleration rather than treat it as evidence that something has gone wrong.

The more useful question is how sharply Cognition slows. Lovable continued growing after its early explosion, but the pace moderated. Emergent's growth also settled after its first sprint. Factory may eventually face the same arithmetic if its six consecutive monthly doublings came from a meaningful base.

Cognition currently has one advantage here: its strongest growth is appearing after it has already reached hundreds of millions of dollars in annualized revenue. That is harder to achieve than producing the same percentage increase during the first few million.

Which AI coding startup is growing the fastest today?

Cognition is the fastest-growing independent AI coding startup today based on the combination of current revenue growth, dollars added and enterprise adoption.

The conclusion is stronger after checking the latest evidence rather than weaker.

Cognition recently added more than $400 million of annualized revenue in roughly one quarter while starting from close to half a billion dollars. As pointed out above, that same period came with unusually fast enterprise expansion, and Brex's newest real-world spending data has Cognition climbing from 16th to 7th among its fastest-growing software vendors.

Lovable remains the most impressive challenger on broad adoption. Its rise from roughly $200 million to more than $500 million of annualized revenue in around seven months would normally be enough to dominate a software growth ranking. Replit may also be above $500 million based on Sacra's estimate, although we need a fresher company number before moving it higher.

Factory is the one company that could overturn the answer. Six consecutive months of doubled revenue is faster than anything Cognition has disclosed in percentage terms, but Factory has never told us what revenue level those doublings started from. We cannot rank an unknown base ahead of several hundred million dollars of observed growth.

Cursor would have won on absolute expansion before SpaceX bought it. Claude Code is already a larger coding business on its last confirmed figure, but Anthropic is a general AI company rather than an AI coding startup.

That leaves Cognition in the clearest position for now. It has reached the point where growth becomes difficult to fake with a tiny denominator, and the business is still expanding unusually fast. Among independent AI coding startups with revenue we can actually measure, nobody else currently combines its scale with the same rate of acceleration.

If you want more recent data on this point, please see our latest AI code assistant market report.

Who is the author of this content?

NEW MARKET PITCH TEAM

We track new markets so founders and investors can move faster

We build living "market pitch" documents for emerging markets: AI, synthetic biology, new proteins, and more. Instead of outdated PDFs or hallucinated LLM answers, our clients get a clean, visual, always-updated view of what's really happening: key players, deals, regulations, and signals that matter. Learn more about us.