Lovable Valuation Tops Replit by $4.3B as AI App Builders Split

The gap, explained

In the AI app-builder market, the startup with fewer users is now worth more. Lovable, the Stockholm company that turns plain-language prompts into working web apps, raised $400 million in a Series C announced August 12, 2026, at a $13.3 billion valuation18. Replit's most recent priced round, a $400 million Series D on March 11, 2026, valued it at $9 billion1113. That leaves Lovable $4.3 billion ahead.

The gap is a reversal. When Replit closed its Series D, Lovable's last confirmed valuation was the $6.6 billion it set in December 202524. In about five months, Lovable went from roughly $2.4 billion behind Replit to $4.3 billion ahead. Replit's own valuation rose quickly too: it tripled from $3 billion in only six months13.

Bolt.new, from StackBlitz, is far behind both. Its only confirmed valuation is about $700 million, from a $105.5 million Series B in January 2025 led by Emergence Capital and GV2228. No newer price has been confirmed28.

Why Lovable pulled ahead

The main reason is the speed of its revenue growth. Lovable reports that its annualized revenue went from $200 million in November 2025 to $400 million in February 2026 and past $500 million in June24. In September it said it had crossed $600 million27. Business Insider reported that CEO Anton Osika described annual recurring revenue as nearly tripling since December8.

The company also beat its own fundraising target. Forbes reported in June that Lovable was in talks at about $12 billion, but the round closed at $13.3 billion7. Menlo Ventures and the EQT-managed Scaleup Europe Fund co-led, with Tencent, Salesforce Ventures, HubSpot Ventures, DST Global and CapitalG also investing1. One analysis argues that a round closing above its rumored price usually means two things: revenue data that came in during due diligence beat expectations, and investors were competing to get in1.

Lovable is also very efficient per employee. It reported about $400 million in ARR in early 2026 with only 146 full-time staff, or roughly $2.7 million of revenue per employee37. One tracker says headcount has since grown past 1,0007, so that ratio is now probably lower.

Replit's numbers are harder to pin down

Replit has many more users but has been less specific about revenue. At its Series D, the company said it had customers at 85% of the Fortune 500 and expected to reach $1 billion in run-rate revenue by the end of 202611. TechCrunch noted that Replit did not give a current ARR figure alongside the round13.

Outside estimates differ widely:

• One profile works from $150 million in ARR, which implies a valuation of about 60 times revenue16.
• Another estimate puts ARR at about $265 million at the end of 202515.
• Research firm Sacra put it at about $300 million at the end of 2025, and later estimated $525 million by spring 20261412.

The gap between these estimates matters. At about $525 million, Replit's $9 billion valuation is roughly 17 times revenue12. If it reaches its $1 billion target, the multiple falls to about 9 times, which is ordinary for software16. Lovable, by comparison, is valued at about 22 times its forward revenue run rate1.

The likely reason Lovable is worth more despite a much smaller user base is that it earns more per user. One analysis credits Replit with more than 30 million users against Lovable's roughly 8 million, but says Lovable's credit-based pricing earns more from each user1. That pricing combines $25 to $50 monthly subscriptions with paid usage top-ups1. Replit itself cites more than 50 million users19. On that reading, the extra $4.3 billion reflects revenue per user, not audience size.

Replit still has strong backing. Visa made a strategic investment in May 2026 on undisclosed terms14. In June, Replit announced a Microsoft Fabric integration aimed at companies that already use Microsoft's data stack12. Secondary-market trading on Forge implied a value of about $9.4 billion in late July20, which suggests buyers have not marked Replit down even though Lovable has moved ahead.

Bolt.new: fast start, little news since

Bolt.new had one of the fastest launches in the category. StackBlitz launched it on October 3, 2024, and it reportedly reached $4 million in ARR within four weeks, $20 million within about three months, and about $40 million within five2325. Founder Eric Simons has said StackBlitz was close to shutting down at roughly $80,000 in ARR before Bolt25.

The $40 million figure dates from March 2025, and the most recent widely cited revenue number is still that third-party estimate28. Bolt had passed 7 million registered users by December 202528. However, its traffic in March 2026 was estimated at 5.43 million visits, compared with 20.91 million for Replit and 25.49 million for Cursor28. Sacra estimated its gross margin at about 40% as of May 202529. Lovable's reported ARR is now more than ten times Bolt's last public figure2428.

One review says StackBlitz is widely thought to have raised again at a higher valuation23, but other trackers say no newer price has been confirmed28. Until Bolt publishes newer numbers, it is best treated as a smaller competitor with good browser-based technology rather than a contender for the top spot.

The developer-tool side: Cursor's sale and Windsurf's rebrand

The consumer app builders sit beside a separate group of AI coding tools aimed at professional developers, and that group has consolidated quickly.

Cursor. SpaceX agreed on June 16, 2026 to buy Anysphere, the company behind the Cursor editor, in an all-stock deal worth $60 billion41. The deal closed on August 14, and Cursor now sits in a new SpaceXAI division4246. Coverage says Cursor's annualized revenue grew from about $100 million in January 2025 to about $4 billion by June 2026, which values the deal at about 15 times revenue41.

Some details differ between reports. Most date Cursor's $29.3 billion Series D to November 20254347, while one investor guide places it in January 202644. One analysis says Cursor's main selling point to companies is that it can route work to different AI models from OpenAI, Anthropic and Google task by task. In contrast, Claude Code and Codex each run only on their own company's models47.

Windsurf. Windsurf's story is more confusing. In July 2025, OpenAI's planned roughly $3 billion purchase fell apart. Google then paid a reported $2.4 billion to license the technology and hire Windsurf's leaders. Days later, Cognition, the maker of the Devin coding agent, bought the remaining product and team3440. One review dates that purchase to December 202531, but most accounts put it in July 20253435.

Under Cognition, the editor changed a great deal:

• Pricing (March 19, 2026): the Pro plan rose from $15 to $20 a month, matching Cursor, and daily and weekly quotas replaced credits3138.
• Windsurf 2.0 (April 15, 2026): added Codemaps, which maps how a codebase fits together; Cognition's in-house SWE-1.5 coding model; embedded Devin agents; and an Agent Command Center for managing agents32.
• Rebrand (June 2, 2026): an automatic update renamed the editor Devin Desktop, and windsurf.com now redirects to devin.ai3334.
• New agent (July 1, 2026): the original Cascade agent was retired and replaced by Devin Local, which runs on Cognition's own SWE-1.6 model36.

Cognition says the combined company's revenue run rate grew from $73 million to more than $500 million40, and it raised money at a $26 billion valuation in May12.

What it means

The market is splitting into two groups that investors value differently.

• Tools for professional developers. Cursor and Devin Desktop are full code editors built for engineers. Their companies are worth tens of billions, they increasingly use their own AI models, and their roadmaps now focus on teams of AI agents. Both Cursor 3 and Windsurf 2.0 shipped in April with a focus on managing multiple agents32.
• App builders that work in the browser. Replit, Lovable and Bolt.new let people build apps without needing to know how to code. In this group, investors are rewarding the companies that earn the most revenue, not the ones with the most sign-ups.

On that basis, Lovable's lead over Replit looks justified, but it also carries risk. Both valuations depend on revenue figures the companies report themselves. Lovable had disclosed $500 million but had not confirmed reaching $600 million when its round closed1, and Replit's $1 billion target has not been independently checked12. Replit's $9 billion price will look modest if it reaches $1 billion in ARR by year-end. If it falls well short, the $4.3 billion gap will probably grow. The two companies' next funding rounds will show which happened.