Which AI startup is growing the fastest?
In our generative AI market deck, you will find everything you need to understand the market
SUMMARY
Anthropic is the fastest-growing AI startup today.
Its lead comes from the amount of new business it is adding, not just a spectacular percentage. Anthropic has added tens of billions of dollars in annualized revenue while still multiplying from an already large base.
The accounting debate with OpenAI narrows the gap but does not erase it. Even after applying OpenAI’s proposed adjustment to Anthropic’s cloud revenue, Anthropic remains well ahead of OpenAI’s latest disclosed annualized figure.
OpenAI still owns the stronger consumer distribution asset. ChatGPT’s reach gives it more ways to monetize later, but a huge free audience is not the same thing as the fastest current revenue expansion.
Claude Code appears to have been the spark rather than the whole engine. It brought developers into Anthropic’s ecosystem, then helped the company expand into APIs, cloud distribution and much larger enterprise accounts.
The sharpest evidence is coming from customer depth. Anthropic’s group of seven-figure business customers doubled to more than 1,000 in under two months, which is hard to explain as casual experimentation.
Cursor leads the application-software race. Its annualized revenue reportedly doubled from about $1 billion to more than $2 billion in roughly three months, with corporate customers now providing most of the business.
Lovable wins the race from launch. It reached more than $500 million in annualized revenue at remarkable speed, although its latest $100 million increase took longer than the jump before it.
Mercor belongs near the leaders, but its roughly $2 billion figure is gross marketplace revenue. Once contractor payments are removed, the comparison with software and API companies changes quite a lot.
The uncomfortable part is profitability. Anthropic is growing fastest, Cursor may have the strongest large application business, and Glean or Sierra may eventually produce steadier economics, but nobody has yet proved that today’s explosive AI revenue converts cleanly into durable profit.
This market map, featured in our generative AI market deck, highlights top companies and startups in the generative AI market
Which AI startups are actually growing fast enough to count?
Anthropic, OpenAI, Cursor and Lovable are currently the four serious contenders, although each one leads a different kind of race.
Anthropic and OpenAI operate frontier models, APIs and consumer assistants. Cursor sells an AI coding environment mainly to developers and companies. Lovable helps less technical users build complete applications through natural-language instructions.
Mercor also belongs in the discussion because its annualized customer spending has reached roughly the same scale as Cursor’s. We have kept it separate because Mercor passes most of that money to the specialists who train and evaluate AI models. Its gross revenue cannot be compared directly with software revenue.
Several familiar AI names fall outside the leading group. Safe Superintelligence has attracted enormous funding without launching a commercial product. Perplexity, Harvey, Glean and Sierra are growing quickly, but their disclosed revenue remains much smaller. xAI is harder to isolate after becoming part of a broader company.
The field is smaller than the AI funding headlines suggest. Once we require recent revenue evidence at meaningful scale, only a handful of companies have a credible claim.
| AI company | Latest public revenue measure | What gives it a claim |
|---|---|---|
| Anthropic | More than $47B run rate | Most new revenue added |
| OpenAI | About $25B annualized revenue | Largest consumer reach |
| Cursor | More than $2B annualized revenue | Fastest large AI application |
| Mercor | About $2B gross run rate | Exceptional marketplace growth |
| Lovable | More than $500M run rate | Fastest rise from launch |
| Glean | $300M ARR | Strong enterprise subscription growth |
What does “growing fastest” mean for an AI startup?
The fastest-growing AI startup should be the company adding the most revenue at exceptional speed, rather than whichever young company can produce the largest percentage.
A startup moving from $1 million to $10 million has grown tenfold. Another moving from $10 billion to $40 billion has grown fourfold. The first company wins the percentage comparison, while the second has created more than 3,000 times as much new revenue.
We therefore look primarily at net new annualized revenue. This measures how much additional yearly business a company’s latest monthly pace represents. We then check how quickly that figure grew, whether customers keep expanding their spending and how comparable the accounting is.
Annualized revenue still needs care. It takes one recent month or short period and extends it across a full year. A temporary surge can make the number look stronger than the revenue the company eventually records.
The measure is especially imperfect in AI because many customers pay according to usage. A company can see its run rate jump after a popular model release, then watch usage fall as customers optimize costs or move routine work to cheaper models.
Even with those limits, net new annualized revenue gives us the clearest answer. User registrations, valuation and funding can support the analysis, but none tells us how much demand customers are turning into actual spending.
As this chart shows, and as featured in our generative AI market deck, search interest in LLMs has surged
Is Anthropic the fastest-growing AI startup right now?
Anthropic is currently the fastest-growing AI startup because it has added roughly $38 billion of annualized revenue within a few months.
No close competitor has disclosed anything on that scale. Cursor added about $1 billion while doubling its annualized revenue over three months. Lovable added approximately $100 million between its two latest milestones. Glean added $200 million over fifteen months.
Anthropic has added more annualized revenue than those three companies generate altogether. The gap is huge.
Percentage growth points in the same direction. Anthropic increased its run rate more than fivefold from the end of 2025 to its latest funding announcement. Large companies almost never keep multiplying like a startup that has only just found its first customers.
The comparison with OpenAI is even more striking. The Financial Times reported that OpenAI reached about $25 billion in annualized revenue earlier in 2026. Anthropic was behind at that point, then moved ahead within weeks.
Run-rate revenue should never be confused with completed annual sales, and Anthropic uses a more generous accounting presentation for some cloud transactions. We account for both issues later. Neither is large enough to explain away tens of billions of dollars in new customer spending.
If you want more recent data on this point, please see our latest generative AI market report.
How fast is Anthropic really growing?
Anthropic’s revenue growth has accelerated sharply rather than rising at a steady pace.
The company began 2025 at roughly $1 billion in run-rate revenue and passed $5 billion by August. It finished the year near $9 billion, then crossed $30 billion during the first part of 2026.
Anthropic’s latest funding announcement said the figure had passed $47 billion. That implies a recent monthly pace of nearly $4 billion, compared with about $750 million at the end of 2025.
The jump from $9 billion to $30 billion was already extraordinary. Moving from $30 billion to $47 billion shortly afterward shows that demand was still accelerating instead of settling after one strong product launch.
We should still treat the latest figure as a snapshot. Anthropic would need to maintain that monthly pace for a full year before recording $47 billion of actual revenue. Usage-based businesses can change quickly.
Yet the sequence is difficult to dismiss as one abnormal month. Anthropic reported several separate milestones on the way up, and each one showed a substantially larger business than the last.
| Anthropic milestone | Run-rate revenue | Implied monthly pace |
|---|---|---|
| Beginning of 2025 | About $1B | About $83M |
| August 2025 | More than $5B | More than $417M |
| End of 2025 | About $9B | About $750M |
| Early 2026 | More than $30B | More than $2.5B |
| Latest disclosure | More than $47B | More than $3.9B |
This chart, featured in our generative AI market deck, shows annual VC investment in generative AI startups
Has Anthropic already overtaken OpenAI?
Anthropic appears to have passed OpenAI in annualized revenue, although the exact size of its lead remains debatable.
The Financial Times reported that OpenAI had reached approximately $25 billion in annualized revenue by the end of February. Anthropic was around $14 billion at the time, leaving OpenAI comfortably ahead.
Anthropic crossed $30 billion shortly afterward and continued climbing. Even allowing for some growth at OpenAI since its last reported figure, Anthropic’s latest disclosure places it well ahead on the available numbers.
OpenAI argues that the comparison overstates Anthropic’s position. According to the Financial Times, Anthropic records the full amount spent by customers purchasing Claude through Amazon and Google, then treats the cloud providers’ share as a cost. OpenAI records an amount closer to what it keeps.
An OpenAI executive estimated that this difference added roughly $8 billion to Anthropic’s reported run rate. Both companies say their accounting follows accepted standards.
Applying OpenAI’s full adjustment would still leave Anthropic at around $39 billion. OpenAI may have grown beyond its last disclosed pace, but it would need a large unreported jump to close that remaining gap.
Anthropic has taken the lead in disclosed annualized revenue. Declaring it permanently larger than OpenAI would go beyond the evidence, especially when the companies publish private financial data at different times and under different accounting presentations.
If you want more recent data on this point, please see our latest generative AI market report.
Did Claude Code cause Anthropic’s revenue explosion?
Claude Code gave Anthropic its breakthrough product, but the coding assistant accounts for only part of the company’s wider revenue surge.
Anthropic launched Claude Code publicly in May 2025. Six months later, the company said the product had reached $1 billion in run-rate revenue. That would make Claude Code a major software company on its own.
Coding is especially valuable for an AI model provider because developers use large amounts of computing capacity while working across files, testing changes and running repeated agent loops. A coding customer can generate far more usage than someone asking a chatbot a few questions.
Claude Code also gave Anthropic a direct route into companies. An engineer may begin with an individual subscription, after which the employer adopts Claude across a team and starts using the API in internal products.
Still, a $1 billion coding product could not explain a company run rate several dozen times larger. Anthropic’s later expansion must include heavy API consumption, cloud sales, enterprise subscriptions and Claude embedded inside customer products.
Claude Code looks more like the spark than the whole fire. It brought developers to Claude, demonstrated that the model could perform valuable work and helped Anthropic win broader corporate spending.
This chart, featured in our generative AI market deck, looks at OpenAI’s strategy in generative AI
Are companies really spending millions on Anthropic?
Large companies are now spending enough on Anthropic to support the revenue story, and the number of seven-figure customers has been rising unusually fast.
Anthropic said more than 500 business customers were each spending at least $1 million annually when it announced its February funding round. Fewer than two months later, that number had passed 1,000.
Even at the minimum threshold, those accounts would represent more than $1 billion of yearly spending. The true amount is higher because several customers will be far above the cutoff.
The broader customer base has expanded as well. Anthropic reported more than 300,000 business customers in late 2025, while the number generating over $100,000 in annualized revenue had increased nearly sevenfold within a year.
Cloud distribution helps explain the pace. Businesses can buy Claude through Amazon Web Services, Google Cloud and Microsoft Azure rather than creating a completely new vendor relationship. Each platform gives Anthropic access to companies that already have approved budgets and infrastructure.
This concentration in business usage makes Anthropic’s recent growth more convincing. A viral consumer chatbot can attract millions of curious users who disappear within weeks. Companies spending six or seven figures usually require security reviews, technical integration and approval from several departments.
Some of those businesses are undoubtedly experimenting. But experimentation becomes a weaker explanation when customers keep crossing higher spending thresholds and the largest-account group doubles in under two months.
Is Cursor still the fastest-growing AI software startup?
Cursor is currently the fastest-growing large AI application company, even though its overall revenue growth is far smaller than Anthropic’s.
Cursor passed $500 million in annual recurring revenue in June 2025. It crossed $1 billion in November and reportedly exceeded $2 billion about three months later.
That progression means Cursor added roughly $1.5 billion of annualized revenue in less than a year. Few software startups have ever moved through those milestones so quickly.
The company also changed the type of customer driving its business. Cursor initially spread among individual developers, creating the impression that its success depended on enthusiastic early adopters paying relatively small subscriptions.
Bloomberg later reported that corporate customers had grown to approximately 60% of revenue. Larger contracts helped Cursor continue growing even as some developers publicly moved to Claude Code.
That is why online discussion briefly made Cursor look weaker than its financial results suggested. Individual developers are visible on social media. A company expanding a contract across hundreds of engineers is not.
Cursor has forecast further rapid growth, but its last widely reported completed milestone remains above $2 billion. Forecasts should stay out of the ranking until customers generate the revenue.
Cursor wins the application-software category because it combines a multibillion-dollar base with a recent doubling. Lovable is younger and grew faster from zero, but it has not yet added revenue at Cursor’s current pace.
If you want more recent data on this point, please see our latest generative AI market report.
This chart, featured in our generative AI market deck, shows annual funding in generative AI startups
Is Lovable growing faster than Cursor?
Lovable has grown faster from launch than Cursor, while Cursor is currently adding more revenue in absolute dollars.
Lovable reached $10 million in annual recurring revenue within months of launching its commercial product. It passed $100 million eight months after its first $1 million, then moved through $200 million, $300 million and $400 million in quick succession.
The company later told TechCrunch that it had passed $500 million in annualized revenue. Users had created more than 50 million projects and were starting roughly one million new ones each week.
Those figures are exceptional for a company that has yet to reach its third anniversary. Lovable also serves a wider audience than a professional coding editor. Its customers include founders, designers, sales teams and small businesses building websites, stores and internal tools.
The latest revenue interval reveals some slowing. Lovable added $100 million in one month when it moved from $300 million to $400 million. Adding the next $100 million took several months.
Growth from $400 million to $500 million remains strong, but Cursor added about $1 billion during its latest reported three-month interval. Cursor is moving more dollars these days, even if Lovable reached its first major milestones sooner.
Lovable’s next challenge is keeping the software alive after the first build. A project can be created in an afternoon, whereas maintaining security, integrations and changing dependencies requires years of work. Project creation will become a much better growth measure once Lovable publishes retention and abandonment figures.
Does ChatGPT’s huge audience make OpenAI the real winner?
OpenAI still has the strongest consumer distribution in AI, but ChatGPT’s enormous audience does not make OpenAI the fastest-growing company by revenue today.
ChatGPT now reaches an audience measured in hundreds of millions of weekly users and around one billion monthly users, according to recent company and media reports. No other independent AI product has comparable global reach.
That audience gives OpenAI several advantages. It can introduce new subscriptions, workplace tools, commerce features and advertisements to users who already know the product. Anthropic, Cursor and Lovable must work with much smaller starting audiences.
User counts still tell us little about immediate revenue growth. Someone opening the free version of ChatGPT twice a month counts as a user. A company spending $5 million through an API may never open the consumer application.
OpenAI’s actual financial growth remains impressive. Audited figures reviewed by the Financial Times showed that yearly revenue reached around $13 billion in 2025, while revenue increased from roughly $1 billion per quarter in late 2024 to about $2 billion per month by the end of 2025.
Anthropic has lately converted enterprise and coding demand into revenue much faster. OpenAI keeps the better distribution asset, but Anthropic has the steeper commercial curve.
That could reverse again. OpenAI has been pushing harder into corporate sales and Codex as competition from Claude grows. For now, the user-growth crown and the revenue-growth crown belong to different companies.
This chart, featured in our generative AI market deck, compares the main business model options for generative AI SaaS platforms
Could Mercor be growing faster than Anthropic?
Mercor may be growing faster than any AI application startup in percentage terms, but its reported revenue is too different from Anthropic’s to make it the overall winner.
Mercor supplies AI laboratories with engineers, doctors, lawyers and other specialists who train models, evaluate answers and create expert datasets. Sacra estimates that its annualized gross revenue reached about $2 billion after ending 2025 near $760 million.
That is an increase of roughly 160% within six months. The company reportedly now pays more than $4 million per day to over 30,000 specialists.
The contractor payments explain the comparison problem. Mercor counts total customer spending before paying the people who perform the work. Sacra estimates that contractors receive roughly 60% to 70% of that amount.
A software company keeps a much larger share of each revenue dollar before paying for staff and infrastructure. Mercor’s $2 billion of gross revenue may translate into only $600 million to $800 million before its own operating costs.
Mercor also depends heavily on a small group of frontier AI laboratories. Its growth could fall quickly if those companies reduce spending on human evaluation or automate a larger share of the work.
Mercor is still one of the fastest-growing AI businesses below the frontier-model layer. Growing a marketplace from almost nothing to billions of dollars in customer spending is rare. But gross marketplace revenue cannot beat Anthropic’s software and API revenue in a clean comparison.
If you want more recent data on this point, please see our latest generative AI market report.
Can Glean, Harvey or Sierra catch the AI growth leaders?
Glean, Harvey and Sierra are growing into substantial enterprise software companies, but none is currently close to Anthropic, Cursor or Lovable in net new revenue.
Glean reached $300 million in annual recurring revenue fifteen months after passing $100 million. It needed nine months to move from $100 million to $200 million, then roughly six more months to add the next $100 million. The intervals suggest that growth has been accelerating.
Harvey ended 2025 near $190 million in annual recurring revenue after reaching approximately $100 million earlier that year. More recently, Business Insider reported that Harvey had added around $100 million of annualized revenue within one quarter as it expanded beyond law firms into corporate legal departments and asset management.
Sierra crossed $100 million in annual recurring revenue seven quarters after launching. Its customers include some of the world’s largest retailers, banks and healthcare companies, giving it room to expand contracts as its agents handle more customer interactions.
All three businesses have a credible path to billion-dollar revenue. They sell into large company budgets, solve specific workplace problems and can deepen usage after landing an initial customer.
Their current scale keeps them out of the top spot. Anthropic has recently added more revenue within weeks than these companies have generated throughout their lives.
| Enterprise AI startup | Disclosed revenue milestone | Recent growth pattern |
|---|---|---|
| Glean | $300M ARR | Tripled in 15 months |
| Harvey | More than $200M annualized revenue | Added about $100M in a recent quarter |
| Sierra | $100M ARR | Reached the milestone in seven quarters |
This chart, featured in our generative AI market deck, illustrates how revenue is distributed across customer segments in the generative AI market
Is all this AI startup revenue coming from real customers?
Most of the reported AI revenue comes from real customer spending, although easy funding and subsidized computing are helping the market grow faster than normal software.
The demand appears in several different forms. Anthropic has over 1,000 customers spending seven figures annually. Cursor says large companies now provide most of its revenue. Lovable has moved beyond hobby projects into internal tools and commercial websites. Glean has nearly doubled its Fortune 500 customer count within a year.
Those patterns are stronger than free registrations or temporary website traffic. Businesses have to approve budgets, connect internal systems and manage security before using these products at scale.
The subsidies remain large. Frontier-model companies raise tens of billions of dollars to buy chips and data-center capacity. Application startups purchase models from providers that are also competing aggressively on price. Customers may currently receive more AI usage than the long-term economics can support.
Heavy spending does not make the demand imaginary. It does mean today’s growth rates partly reflect prices designed to win market share before the industry settles.
The best test will come when prices rise or generous usage limits disappear. Products connected to measurable work, such as coding, legal review and customer service, should hold up better because buyers can compare the cost with employee time or outsourced work.
Generic AI subscriptions face a less predictable future. Users can switch between assistants easily, and several competing models can now handle ordinary writing, research and image tasks.
The revenue is real. The margins and long-term customer value remain much less certain.
Can these fast-growing AI startups ever make money?
Fast AI revenue growth has not yet produced a clearly profitable leader, and the frontier-model companies may need a very different cost structure before that changes.
OpenAI generated about $13 billion of revenue in 2025, according to audited figures reported by the Financial Times, while spending roughly $34 billion. Excluding several large non-cash items, its operating losses were still around $8 billion.
Anthropic also remains lossmaking while making huge infrastructure commitments. The company raised $65 billion at a $965 billion valuation in its latest round and said the money would help expand computing capacity for Claude.
Application companies avoid the full cost of training frontier models, but they pay model providers whenever customers use their products. A flat monthly subscription becomes dangerous when one heavy user can consume more computing than the subscription brings in.
Cursor has responded through usage limits, higher-priced plans, model routing and its own AI research. Lovable recently introduced subagents that can send simpler work to cheaper models. These changes show how closely the companies now watch the cost of every task.
Mercor faces a different problem. Most of its gross revenue goes directly to contractors, leaving a much smaller amount to cover sales, technology and operations.
Glean, Harvey and Sierra may eventually have cleaner software economics because they sell large contracts for narrow business problems. Their slower growth could come with better retention and more predictable margins.
Growth and business quality currently point to different winners. Anthropic is adding revenue fastest. Cursor may have the strongest large application business. Glean or Sierra could eventually produce the steadier enterprise economics.
This chart, featured in our generative AI market deck, shows how AI video generation technology has evolved over time
Which AI startup is growing the fastest today?
Anthropic is the fastest-growing AI startup today, and the gap is wide enough that the accounting caveats do not change the answer.
Anthropic has multiplied an already large revenue base while adding tens of billions of dollars in annualized customer spending. No other AI startup has documented a comparable increase over the same period.
Cursor takes the narrower title of fastest-growing large AI software application. Its move from hundreds of millions to billions in annualized revenue shows that an application company can keep growing even while competing directly with the model providers beneath it.
Lovable wins when we measure speed from launch. Few software companies have reached half a billion dollars of annualized revenue so early, although its latest growth has been slower than the month when it jumped from $300 million to $400 million.
OpenAI still owns the broadest consumer distribution. Mercor is growing at a remarkable rate before contractor payments. Glean, Harvey and Sierra are building strong enterprise businesses. Each one leads a useful secondary measure.
But the overall title has one answer. Based on the latest available revenue disclosures, the amount of new business created and the speed at which large customers are expanding, Anthropic is currently growing faster than every other major AI startup.
If you want more recent data on this point, please see our latest generative AI market report.
OUR METHODOLOGY
This analysis tests which AI startup is growing fastest using the freshest credible commercial evidence available. We compare net new annualized revenue, growth speed, starting scale, customer expansion and the comparability of each company’s reported figures.
We give the greatest weight to revenue already reflected in a recent run rate or ARR milestone. Percentage growth, product launches, user counts, funding and valuation are supporting evidence, not substitutes for customer spending.
Run-rate revenue is treated as a snapshot rather than completed annual sales. It extends a recent monthly or short-period pace across a full year, so we compare successive milestones and customer trends instead of relying on one isolated figure.
We also separate unlike revenue models. Mercor’s gross marketplace revenue includes money passed to specialists, while Anthropic, OpenAI, Cursor, Lovable, Glean, Harvey and Sierra report forms of software, API or subscription revenue. That difference is reflected in the ranking.
The Anthropic and OpenAI comparison includes the accounting disagreement reported by the Financial Times. We test Anthropic’s lead both before and after OpenAI’s proposed adjustment for cloud-provider revenue, rather than assuming the companies present those transactions in exactly the same way.
Private companies disclose milestones at different times. We use the latest completed figure we could support and keep forecasts out of the ranking until the revenue is actually generated.
We prioritized direct company disclosures and then used authoritative financial and business reporting for private figures, accounting context and independently reported milestones. Key sources include Anthropic’s Series H announcement, Anthropic’s Series G update, Anthropic’s $30 billion run-rate update, Anthropic’s seven-figure customer update, Anthropic on Claude Code’s $1 billion milestone, OpenAI’s latest funding and adoption announcement, OpenAI’s enterprise AI report, the Financial Times on OpenAI’s 2025 financials, and the Financial Times on Anthropic’s acceleration and accounting comparison.
Application-company milestones were checked against Bloomberg on Cursor’s $2 billion milestone, TechCrunch on Cursor’s corporate revenue mix, TechCrunch on Lovable’s $500 million run rate, Glean’s $300 million ARR announcement, Sierra’s $100 million ARR announcement, Business Insider on Harvey’s recent growth, and Sacra’s Mercor revenue estimate.
The conclusion is based on convergence rather than one headline number. Anthropic leads because the amount of new annualized revenue, the speed of the increase and the expansion of large customer accounts all point to the same company.
In our generative AI market deck, we identify pain points entrepreneurs should prioritize
