Which AI startup will reach $1B ARR first?
In our AI infrastructure market deck, you will find everything you need to understand the market
SUMMARY
Cognition is the AI startup most likely to reach $1 billion in annualized revenue next, assuming its latest roughly $900 million run-rate estimate is broadly accurate.
The race is much narrower than the headline suggests. Cursor, OpenAI and Anthropic have already crossed $1 billion, so the useful comparison is among private AI companies that are still below the threshold.
Cognition's advantage is mostly arithmetic. At roughly $900 million, it needs only about 11% more revenue, while ElevenLabs and Lovable still need close to another $500 million each.
The catch is evidence quality. Cognition's latest figure comes from a private-market estimate rather than a fresh company-confirmed ARR announcement, so its lead is large but not quite as clean as it looks.
ElevenLabs is the strongest verified challenger. Its revenue milestones have kept arriving faster, and voice agents can expand account value quickly as enterprise call volume grows.
Lovable remains a serious contender, but its recent progression from roughly $400 million to $500 million was less explosive than its earlier growth. The fact that its latest fundraising announcement still highlighted the $500 million run rate is also telling.
Replit is the wild card. Management has publicly targeted a $1 billion run rate by year-end, but the last widely disclosed starting point was around $150 million, leaving too much missing data to rank it ahead of Cognition.
Harvey, Glean, OpenEvidence and Sierra are all growing fast enough to become billion-dollar revenue businesses, but they still have much more distance to cover. Their main advantage is that a few large enterprise deployments can move revenue in big chunks.
AI coding has become the fastest route to this scale because adoption can start with one developer, spread across an engineering organization, and then rise again as usage increases. Cursor already proved that a coding company can move from $500 million to more than $2 billion at startling speed.
The $1 billion milestone is also less standardized than it sounds. Some companies mix subscriptions with consumption pricing, infrastructure usage or outcome-based charges, so annualized revenue should be treated as a scale marker rather than a perfectly comparable accounting metric.
The ranking changes only if Cognition's estimated $900 million run rate is materially overstated or a company such as Replit has already made a huge undisclosed jump. Based on the evidence available, Cognition is first, ElevenLabs is second, and Lovable and Replit form the next group.
This market map, featured in our AI infrastructure market deck, highlights top companies and startups in the AI infrastructure market
Which AI startup will reach $1B ARR first?
Has an AI startup already reached $1B ARR?
Yes. Several AI startups have already crossed $1 billion in annualized revenue, so the interesting race today is which company still below that level gets there next.
Cursor is the clearest comparison. Anysphere said Cursor had passed $500 million in ARR in June 2025. By November, the company was above $1 billion. Bloomberg later reported that Cursor's annualized revenue had doubled again to more than $2 billion by February 2026. In other words, Cursor added roughly $1.5 billion of annualized revenue in less than nine months after reaching the $500 million mark.
OpenAI and Anthropic crossed $1 billion much earlier and are now far beyond the range discussed here. Including them would make the question pointless.
So we are looking at private AI companies that have not yet clearly established annualized revenue above $1 billion. On that basis, Cognition is currently the company to beat.
| Company | Latest annualized revenue level | Position |
|---|---|---|
| Cursor | >$2B | Already crossed |
| Cognition | ~$900M estimate | Closest remaining |
| ElevenLabs | >$500M | Main challenger |
| Lovable | ~$500M | Main challenger |
| Harvey | >$350M | Further behind |
| Glean | $300M | Further behind |
| OpenEvidence | ~$300M | Further behind |
Is Cognition really the closest AI startup to $1B ARR right now?
Yes. Cognition currently has such a large numerical lead that another company probably wins only if its latest revenue estimate is materially wrong.
Sacra now estimates Cognition at roughly $900 million in annualized revenue, up from about $492 million in May. ElevenLabs' latest company-confirmed figure remains above $500 million. Lovable last disclosed approximately $500 million. Harvey has recently moved above $350 million, while Glean and OpenEvidence sit around $300 million.
That leaves Cognition roughly $100 million short of the threshold. ElevenLabs and Lovable each have about $500 million still to add. Harvey needs nearly $650 million.
The gap becomes clearer when we convert it into required growth. Cognition needs roughly 11% from its estimated current base. ElevenLabs and Lovable need close to 100%. Glean would need more than triple its current revenue.
Fresh disclosures have reinforced that ranking rather than overturned it. ElevenLabs recently updated its company page but still cites the $500 million-plus milestone reached earlier in the year. Lovable's latest $400 million fundraising announcement still refers back to its $500 million annualized run rate. Harvey has announced a string of major new deployments lately, but recent reporting still puts annualized revenue around $350 million.
No challenger has disclosed a fresh number that closes Cognition's apparent lead.
If you want more recent data on this point, please see our latest AI infrastructure market report.
As this chart shows, and as featured in our AI infrastructure market deck, search interest in AI infrastructure has risen sharply
Can we actually trust Cognition's $900M revenue estimate?
We should treat Cognition's roughly $900 million figure as credible enough to make it the favorite, but not solid enough to call the race finished.
The crucial distinction is where the numbers come from. Cognition itself disclosed that Devin went from $1 million ARR in September 2024 to $73 million by June 2025. After buying Windsurf, Cognition said the acquisition more than doubled its ARR and that combined enterprise ARR then grew another 30% in seven weeks.
Those figures establish the direction very clearly.
The newer numbers come from Sacra rather than Cognition. Sacra estimates annualized revenue at approximately $492 million in May and $900 million in August. If both estimates are close, Cognition added roughly $408 million of annualized revenue in three months.
That is enormous. The increase alone exceeds Glean's entire current ARR and is larger than Harvey's whole business.
But a $900 million estimate from a private-market data provider should carry less weight than a company-confirmed ARR announcement. We are highly confident Cognition has become a several-hundred-million-dollar business very quickly. We have lower confidence in the precise $900 million endpoint.
That uncertainty is basically the whole remaining case against Cognition.
How fast is Cognition growing compared with the other $1B ARR contenders?
Cognition's recent revenue growth appears much faster in absolute dollars than any other AI startup still below $1 billion.
The progression starts with Devin at $1 million ARR in September 2024 and $73 million nine months later. Cognition then bought Windsurf, whose ARR was around $82 million before the acquisition. The combined company subsequently moved into the hundreds of millions.
Sacra's latest estimates imply another jump from approximately $492 million in May to $900 million in August.
Compare that with ElevenLabs. The voice AI company finished 2025 at $350 million ARR and passed $500 million during the first four months of 2026, adding at least $150 million. Lovable moved from roughly $400 million in February to more than $500 million in June. Glean went from $200 million in December 2025 to $300 million in May.
All three are growing extremely fast. Cognition's estimated dollar increase is still in another category.
We should be careful about extrapolating three months indefinitely. At this point, though, Cognition barely needs any extrapolation. An 11% increase from $900 million gets the company across the line.
| Company | Earlier level | Latest level | Approximate increase |
|---|---|---|---|
| Cognition | ~$492M in May | ~$900M in August | +~$408M |
| ElevenLabs | $350M at end-2025 | >$500M within four months | +>$150M |
| Lovable | ~$400M in February | >$500M in June | +~$100M |
| Glean | $200M in December 2025 | $300M in May | +$100M |
This chart, included in our AI infrastructure market deck, shows annual VC investment in AI infrastructure startups
Could ElevenLabs still beat Cognition to $1B ARR?
ElevenLabs is the strongest verified challenger, but today it still looks too far behind Cognition to be our favorite.
ElevenLabs has one of the cleanest revenue trajectories in private AI. The company crossed $100 million ARR, then $200 million, finished 2025 at $350 million and passed $500 million within the first four months of 2026.
The intervals between those milestones kept shrinking. That is exactly what we want to see from a company trying to double from $500 million to $1 billion.
The business is also becoming broader. ElevenLabs started with synthetic speech, dubbing and voice generation. Enterprise customers now use ElevenAgents for customer support, sales, recruiting and marketing. Deutsche Telekom, Square, Revolut and other large organizations have deployed its technology.
Voice agents can create much larger accounts than individual creator subscriptions because usage grows with the number of calls or conversations handled. That gives ElevenLabs a plausible route to another several hundred million dollars of revenue.
Still, the arithmetic is difficult to ignore. ElevenLabs needs roughly another $500 million from its latest confirmed level. Cognition may need around $100 million.
If Cognition's $900 million estimate proves badly inflated, ElevenLabs becomes our favorite very quickly. With the information available now, it remains second.
If you want more recent data on this point, please see our latest AI infrastructure market report.
Is Lovable still growing fast enough to reach $1B ARR soon?
Lovable can absolutely reach $1 billion in annualized revenue, but its recent pace suggests it probably gets there after Cognition.
Lovable's rise has been extraordinary. The company passed roughly $400 million in annualized revenue in February and $500 million in June, less than three years after it was founded.
The adoption numbers remain huge. When Lovable announced a $400 million funding round lately, the company said its platform hosted about 60 million projects attracting roughly 900 million monthly visits. TechCrunch previously reported around one million new projects being created each week.
Lovable has also found users well beyond professional software engineers. Its own research says a large majority of users are non-technical, which gives the company access to founders, marketers, designers and small-business owners who previously would have needed developers to build software.
But revenue progression has become less explosive than it was during the earliest phase. Moving from around $400 million to $500 million took roughly four months. Another $500 million would have to be added to reach the target.
Lovable's latest financing also gives us a useful fresh check. The company raised at a $13.3 billion valuation and talked extensively about its growth, infrastructure and 60 million hosted projects, yet the revenue number highlighted publicly remained the $500 million run rate reached in June. If Lovable had already moved dramatically closer to $1 billion, that fundraising announcement would have been the obvious moment to say so.
For now, Lovable looks like a future $1 billion business rather than the likely next one.
This chart, included in our AI infrastructure market deck, shows why CoreWeave is winning in AI infrastructure
Could Replit suddenly win the $1B ARR race?
Replit is the hardest contender to rank because management openly expects $1 billion in run-rate revenue, but we still lack the current number needed to know whether that target is realistic.
Replit said it aims to reach $1 billion in annual revenue run rate by the end of 2026. The ambition deserves attention because AI coding companies have already shown that apparently ridiculous growth can happen: Cursor doubled from $1 billion to $2 billion in roughly three months.
The problem is the starting point. Replit was around a $150 million annual revenue run rate in late 2025. Getting from $150 million to $1 billion requires adding $850 million, or more than five times the starting revenue.
Replit has several advantages. It reaches both developers and people who barely code. More than 50 million users have joined the platform, and Replit says people inside 85% of Fortune 500 companies use it. The product can also capture more spending as users move from generating an application to actually hosting and operating it.
Yet the company has not published a fresh intermediate figure showing that the revenue curve has caught up with the $1 billion target.
A current disclosure around $600 million or $700 million would change our ranking immediately. Until we get one, putting Replit ahead of Cognition would mean trusting a forecast over a much higher reported current run rate.
If you want more recent data on this point, please see our latest AI infrastructure market report.
Is Harvey becoming a serious $1B ARR contender?
Harvey is becoming a huge AI software company surprisingly quickly, although $1 billion still looks too far away for Harvey to win this race.
Recent reporting puts the legal AI startup above $350 million in annualized revenue, up from roughly $190 million near the beginning of the year. That is close to a doubling in well under twelve months.
Harvey has also kept adding large customers. Its recent newsroom shows deployments at GE Aerospace, Nelson Mullins, Finnegan, Macpherson Kelley and Grupo Financiero Inbursa within a short period. More than 200,000 lawyers reportedly use the platform, and Harvey already sells to major corporate legal departments as well as law firms.
Those deployments make the $350 million figure easier to take seriously. We can see a steady stream of large organizations adopting the product rather than a single unexplained revenue jump.
Harvey's challenge is the remaining distance. A move from $350 million to $1 billion requires almost tripling the business.
Competition is getting harder too. OpenAI is pushing more directly into legal workflows, Anthropic is working with major law firms, and some firms are building their own tools. Harvey could still become the dominant standalone legal AI platform, but it currently has far more ground to cover than Cognition.
This chart, included in our AI infrastructure market deck, shows annual funding in AI infrastructure startups
Could Glean reach $1B ARR through big enterprise contracts?
Glean has one of the most believable enterprise AI businesses in this group, but its current $300 million ARR leaves too much distance for it to win the next crossing.
Glean moved from $100 million ARR to $300 million in only 15 months. The company reached $200 million in December 2025 and then added another $100 million by May.
Its customer pattern looks strong as well. Glean says its Fortune 500 customer count nearly doubled year over year and that more than 85% of customers use the product across at least five departments. Large contracts above $1 million have also been growing quickly.
The company has remained active lately, launching a global partner network and new enterprise AI products aimed at context, cost control and agent deployment. None of those recent announcements included a revenue figure beyond the previously disclosed $300 million.
There is also a measurement wrinkle. CEO Arvind Jain told TechCrunch that Glean uses both consumption pricing and hybrid pricing. Part of what Glean calls ARR therefore moves with customer usage rather than coming entirely from fixed subscriptions.
We still like the underlying business. Enterprise expansion appears broad, and large organizations have plenty of room to spend more with Glean. But another $700 million is a lot to add before Cognition finds its final estimated $100 million.
Could OpenEvidence surprise everyone and reach $1B first?
OpenEvidence is probably the most dangerous outsider because its revenue has doubled so quickly, but the healthcare AI company still starts too far behind.
Sacra estimates that OpenEvidence reached about $300 million in annualized revenue in July, up from roughly $150 million at the end of 2025. The company had generated only about $7.9 million on an annualized basis in 2024.
That means OpenEvidence appears to have gone from single-digit millions to hundreds of millions in roughly two years.
Usage has scaled with it. Sacra reports around 20 million clinical consultations per month early this year, compared with roughly three million per month one year earlier. On one day in March, doctors reportedly conducted one million consultations through the platform.
OpenEvidence also stands out economically. Reported gross margins are around 90%, while the company has been described as approximately cash-flow breakeven. Plenty of fast-growing AI businesses cannot say anything similar.
The remaining gap is still roughly $700 million. Even another doubling would put OpenEvidence near $600 million rather than $1 billion.
We would watch OpenEvidence closely for the race after the next winner. It would take another extraordinary acceleration to pass Cognition first.
If you want more recent data on this point, please see our latest AI infrastructure market report.
This chart, included in our AI infrastructure market deck, compares the main business model options for AI cloud infrastructure providers
Is Sierra too far behind to reach $1B ARR soon?
Sierra is growing extremely fast for enterprise software, but its last confirmed ARR puts the company well behind the current leaders.
Sierra said in February that it entered its third year above $150 million ARR after reaching $100 million in seven quarters. The company had also just recorded its first $50 million quarter.
Its customer base explains why future growth could come in large chunks. One-quarter of Sierra customers generate more than $10 billion of revenue themselves, and half generate more than $1 billion. These are organizations capable of running enormous numbers of customer interactions through AI agents.
Sierra has continued expanding since that revenue disclosure. It acquired Takeoff, which said it had grown from zero to nearly eight figures of revenue during 2026, entered new international markets, added a partnership with SoftBank and recently appointed a CFO.
Those are useful signs of a company preparing for much larger scale, but none establishes that Sierra is suddenly anywhere near $1 billion ARR.
Even if Sierra's current revenue has risen materially above $150 million, we would need a very large undisclosed jump before it entered the same numerical range as ElevenLabs or Cognition.
Does $1B ARR mean the same thing for Cognition, ElevenLabs, Lovable and Glean?
No. The $1 billion ARR headline mixes subscription revenue, enterprise contracts and usage-based spending, so we should read it as a scale milestone rather than a perfectly standardized accounting measure.
Classic SaaS ARR is straightforward when a customer signs a recurring annual software contract. AI products increasingly charge for what customers actually consume: tokens, calls, agent actions, generated audio, compute or successful outcomes.
That can make revenue move much faster.
A customer using ElevenLabs for millions of support calls can increase spending without buying millions of new seats. Cognition can earn more when developers delegate more work to coding agents. Lovable users consume more infrastructure as they build and run more applications.
Glean makes the problem especially clear because it offers consumption pricing alongside hybrid contracts. TechCrunch pointed out when Glean announced $300 million ARR that part of the figure behaves more like annualized current consumption than fixed recurring subscription revenue.
The distinction does not invalidate the comparison. A customer repeatedly consuming millions of dollars of AI can be very valuable. But a run rate based on recent usage can fall faster than a contracted subscription if customers optimize spending or activity drops.
For this article, we therefore care about two things at once: how much revenue a company is generating at today's pace and how believable that pace is.
Cognition leads on the first measure. ElevenLabs and Glean give us cleaner company-confirmed milestones on the second.
This chart, featured in our AI infrastructure market deck, shows the share of revenue generated by each customer segment in the AI infrastructure market
Why are AI coding startups reaching $1B revenue so much faster than everyone else?
AI coding has become the fastest route to $1 billion because developers can adopt these products immediately, companies can expand them across entire engineering teams, and heavier usage directly creates more revenue.
Cursor has already shown what the upper end looks like. The company went from more than $500 million ARR to more than $1 billion in about five months, then reached an annualized rate above $2 billion roughly three months later.
Cognition has followed a different route. Devin started as a coding agent, while Windsurf brought an IDE, more enterprise distribution and a separate customer base. Cognition said customer overlap between Devin and Windsurf was below 5% before the acquisition, giving the combined company a large cross-selling opportunity.
Before the deal, Devin was around $73 million ARR and Windsurf was reported around $82 million. Together, that would have produced roughly $155 million before accounting for any growth. Cognition's current estimated run rate is almost six times that amount.
So the acquisition explains the initial jump but nowhere near the full increase that followed.
Replit and Lovable stretch the market further by letting people who are not professional engineers build software. Coding AI can therefore sell upward into huge enterprises and outward to millions of people who previously would never have bought developer tooling.
That combination has already produced one $2 billion-plus company in Cursor. It is now pushing Cognition toward the same billion-dollar threshold.
What could actually stop Cognition from reaching $1B ARR first?
The biggest threat to Cognition is measurement error: if the reported $900 million run rate is substantially too high, the whole race becomes much closer.
Imagine Cognition is really at $650 million rather than $900 million. Suddenly the gap with ElevenLabs and Lovable is closer to $150 million, and both competitors have enough growth to make that interesting.
A sharp slowdown in AI coding spending would also hurt. Cognition competes with Cursor, Claude Code, OpenAI Codex, GitHub Copilot, Replit and a growing number of other products. Developers can change tools unusually quickly when a competitor releases a better model or workflow.
There is another possibility: Replit could already be much larger than its last public revenue figure suggests. Management's $1 billion year-end target becomes far more credible if the company has quietly reached several hundred million dollars since its last disclosure.
But we would need one of those things to happen. With the numbers we can verify or reasonably estimate today, Cognition's lead is large.
As seen above, its latest run-rate figure deserves more caution than ElevenLabs' company-confirmed $500 million milestone. Even after applying that discount, Cognition remains the company with the shortest visible path.
This chart, included in our AI infrastructure market deck, shows how GPU cloud infrastructure technology has evolved over time
Which AI startup is most likely to reach $1B ARR next?
Cognition is currently our clear pick to become the next AI startup to reach $1 billion in annualized revenue.
The strongest evidence is the size of the remaining gap. Sacra estimates Cognition around $900 million. ElevenLabs remains above $500 million on its latest company disclosure. Lovable last reported roughly $500 million. Harvey is above $350 million, while Glean and OpenEvidence are around $300 million.
Cognition therefore needs only about 11% more revenue if the estimate is broadly right. ElevenLabs and Lovable still need roughly twice their current businesses.
The freshest information has not produced a hidden challenger. ElevenLabs' recently updated materials still point to its $500 million-plus milestone. Lovable raised another $400 million at a $13.3 billion valuation lately without announcing a newer revenue figure above the $500 million run rate. Harvey has kept landing major customers but remains around $350 million. Glean and Sierra have announced product launches, partnerships and expansion without revealing revenue that changes the ranking.
Replit remains the wild card because it has publicly targeted a $1 billion run rate by year-end without giving us enough current revenue data to judge how close it really is.
Our ranking today is Cognition first and ElevenLabs second. Lovable and Replit form the next group, with Harvey, Glean, OpenEvidence and Sierra further back.
The confidence level is high enough to make a real call: Cognition should reach $1 billion first among the companies that have not already crossed it. The main reason we could be wrong is unusually specific: the latest $900 million Cognition estimate would have to be materially overstated. If that happens, ElevenLabs becomes the company we would back next.
If you want more recent data on this point, please see our latest AI infrastructure market report.
OUR METHODOLOGY
This analysis treats the question as a forecast rather than a simple ranking of the largest AI startups. We compare each company's latest annualized revenue level, remaining distance to $1 billion, recent growth rate, absolute revenue additions, evidence quality and the freshness of the latest disclosure.
We reconstructed recent revenue trajectories instead of relying only on headline ARR. That helps separate a company that is close to $1 billion but slowing from one that still has a larger gap but is closing it unusually quickly. Short bursts of exceptional growth are treated as momentum, not as a growth rate that can be extrapolated indefinitely.
We ranked the evidence as well. Direct company disclosures and first-hand fundraising or financial announcements carry the most weight, followed by recent reporting from established financial and technology publications. Private-company revenue estimates are useful when they are the freshest meaningful datapoint, but we treat the exact number with less confidence than a company-confirmed milestone.
That distinction matters most for Cognition. Its position at the top partly depends on a recent estimate of roughly $900 million in annualized revenue, so we also tested whether the ranking would still make sense if the true number were materially lower.
We treat ARR and annualized revenue as measures of current revenue scale, not as perfectly standardized accounting metrics. Some of these businesses mix subscriptions with consumption pricing, infrastructure usage and outcome-based charges, so we give extra weight to repeated revenue milestones, customer expansion and evidence that the spending is recurring in practice.
Acquisitions are separated from organic acceleration where possible. For Cognition, the combined Devin and Windsurf business is the company competing for the next $1 billion milestone, but we also reconstruct the pre-acquisition revenue bases so the initial step-up is not mistaken for organic growth.
We also ran a freshness check across recent fundraising announcements, company updates, customer deployments and financial disclosures to see whether any contender had published new information strong enough to change the ranking. Those signals support the revenue analysis; they do not replace it.
Key sources include Cognition on post-Windsurf growth and ARR expansion, TechCrunch on Cognition's recent revenue trajectory, recent reporting on Cognition's roughly $900 million annualized revenue level, ElevenLabs on crossing $500 million ARR, Lovable's latest financing announcement, Glean on reaching $300 million ARR, Replit on its $1 billion run-rate target, Sierra on its $150 million-plus ARR milestone, Harvey's newsroom for recent enterprise deployments, TechCrunch on OpenEvidence's scale and revenue, Cursor's $500 million ARR benchmark, Cursor on crossing $1 billion, and Bloomberg on Cursor moving above $2 billion in annualized revenue.
In our AI infrastructure market deck, we identify pain points entrepreneurs should prioritize
Related blog posts
- Which AI startups have reached $100M ARR the fastest?
- Which AI infrastructure startups generate the most revenue today?
- Which AI startups have the most durable revenue?
