What are the top Frontier AI Labs startups by total funding raised?

Top startups in the Frontier AI Labs Market ranked by total funding raised

Here is an updated table that ranks the top startups in the Frontier AI Labs Market based on the total amount of funding they have raised to date.

The table also includes the total number of funding rounds, the date and size of the latest round, the financing type (e.g. Series A, equity financing), key investors, the startup’s current status (active, IPO, acquired, or shut down), and a confidence score based on the data collected (we excluded startups with very low data confidence, to make sure everything is reliable).

If you need to dig deeper and get more detailed data, please check our report covering the Frontier AI Labs Market.

Which Frontier AI labs have raised the most funding?

OpenAI and Anthropic are currently in a funding league of their own.

Our reconstructed ranking puts OpenAI at about $183 billion in cumulative equity funding and Anthropic at roughly $114.8 billion. xAI is a distant third at $37.1 billion, while the next group drops abruptly to around $8 billion: Z.ai, Moonshot AI and Safe Superintelligence.

The top three account for about $335 billion of the roughly $417 billion raised by all 97 companies in our dataset. That means three organizations represent around 80% of the cumulative equity capital in the entire ranking.

Even within those three, OpenAI and Anthropic dominate. Their combined $297.8 billion is more than seven times the funding of xAI and more than 35 times the funding of a company sitting near the bottom of the top ten.

Rank	Company	Cumulative equity funding
1	OpenAI	$183.0B
2	Anthropic	$114.8B
3	xAI	$37.1B
4	Z.ai	$8.1B
5	Moonshot AI	$8.1B
6	Safe Superintelligence	$8.0B
7	DeepSeek	$7.4B
8	Mistral AI	$6.7B
9	Reflection AI	$4.6B
10	MiniMax	$4.2B

Is OpenAI really that much better funded than every other AI lab?

Yes. OpenAI's funding advantage is enormous even by the standards of the current AI boom.

OpenAI's latest major financing alone brought in $122 billion of committed capital. OpenAI said the round was anchored by Amazon, NVIDIA and SoftBank, with Microsoft also participating. That single financing is larger than the entire cumulative funding history of every company in our ranking except Anthropic.

The company had already accumulated substantial capital before that round, bringing our reconstructed equity total to about $183 billion. The scale reflects the economics of OpenAI's strategy: frontier-model research, consumer distribution through ChatGPT, enterprise products, APIs and increasingly capital-intensive compute infrastructure.

OpenAI also said when announcing the financing that its revenue had reached roughly $2 billion per month. That helps explain why investors are willing to fund the company at a scale that would be almost impossible for an experimental AI lab with little commercial activity.

How close is Anthropic to OpenAI in funding?

Anthropic is the only independent Frontier AI company that comes remotely close to OpenAI's fundraising scale.

Our reconstruction gives Anthropic roughly $114.8 billion of cumulative equity funding, including its $65 billion Series H. Anthropic said that round valued the company at $965 billion post-money and was led by Altimeter Capital, Dragoneer, Greenoaks and Sequoia Capital.

That $65 billion Series H alone is larger than the cumulative funding of xAI, Mistral AI, Moonshot AI, DeepSeek, Safe Superintelligence and most other well-known AI labs.

Anthropic also gave investors a commercial justification for that capital intensity. At the time of the Series H, the company said its revenue run rate had crossed $47 billion. It had also committed to huge amounts of compute through Amazon, Google, Broadcom and SpaceX infrastructure.

Anthropic now looks much more like a hyperscale technology company being financed before reaching the public markets than a conventional venture-backed startup.

Is xAI still one of the three biggest-funded Frontier AI labs?

Yes, although xAI is much closer to the rest of the market than to OpenAI or Anthropic.

xAI completed a $20 billion Series E backed by investors including Valor Equity Partners, Fidelity, Qatar Investment Authority, MGX, NVIDIA and Cisco Investments. That financing brought our reconstructed cumulative total to about $37.1 billion.

At $37.1 billion, xAI has raised more than four times as much as the companies clustered around $8 billion. OpenAI, though, has raised almost five times as much as xAI, while Anthropic has raised roughly three times as much.

There is also an important status difference now. SpaceX announced that it had acquired xAI, so xAI should no longer be treated exactly like an independent startup competing for another venture round. Its models, compute assets and products now sit inside a much larger corporate structure.

How concentrated is Frontier AI funding right now?

Frontier AI funding is extraordinarily concentrated at the top.

Across the 97 companies in our reconstructed dataset, cumulative equity funding totals roughly $417 billion. OpenAI, Anthropic and xAI account for around $335 billion of that amount.

That leaves only about $82 billion for the other 94 companies combined.

Only 19 companies have raised at least $1 billion. Thirty have crossed $500 million, while 63 have raised at least $100 million.

Funding level	Number of companies
$100B+	2
$30B+	3
$5B+	8
$1B+	19
$500M+	30
$100M+	63
Below $100M	34

Which Frontier AI startups are in the second funding tier?

Moonshot AI, Safe Superintelligence, DeepSeek, Mistral AI and Z.ai currently form the clearest second tier.

Their cumulative funding totals sit between about $6.7 billion and $8.1 billion, far below xAI's $37.1 billion but comfortably ahead of most other independent labs.

They are not similar businesses, though. Mistral is building sovereign and open-weight AI from Europe. Moonshot AI is one of China's highest-profile foundation-model companies. DeepSeek built its reputation around highly efficient open-weight reasoning models. Safe Superintelligence remains focused on fundamental superintelligence research rather than a broad commercial product portfolio.

Similar funding totals are financing several very different theories of how a challenger to the largest American labs might emerge.

Is Mistral AI now Europe's best-funded Frontier AI startup?

Yes. Mistral AI has created a funding gap between itself and the rest of Europe's independent frontier-model ecosystem.

Mistral recently announced a €3 billion Series D led by Samsung Electronics, with Scaleup Europe Fund and PSG Equity among the co-leads. The company described it as the largest equity fundraising ever completed by a European technology company.

Our reconstruction puts Mistral at roughly $6.7 billion of cumulative equity funding.

Mistral plans to use the capital for frontier research, compute infrastructure, international expansion and commercial growth. The company said it already supports more than 125 large enterprises, including Airbus, ASML and HSBC.

Europe therefore does have a heavily financed frontier-model contender. But the capital gap with OpenAI and Anthropic is still enormous.

How much funding has China put behind its biggest Frontier AI startups?

China now has several multi-billion-dollar Frontier AI companies rather than one obvious national champion.

Our ranking places Z.ai and Moonshot AI at roughly $8.1 billion each, DeepSeek at about $7.4 billion and MiniMax at around $4.2 billion. StepFun follows at approximately $3.2 billion.

Moonshot AI is a useful example of how quickly Chinese financing has moved. Bloomberg reported that the company closed a $3.5 billion financing at a valuation of roughly $35 billion, with China's National Artificial Intelligence Industry Investment Fund among the investors.

The pattern differs from the United States. American funding is massively concentrated in OpenAI and Anthropic, with xAI behind them. China has a broader cluster of heavily funded model developers within a narrower capital range.

Company	Cumulative funding
Z.ai	$8.1B
Moonshot AI	$8.1B
DeepSeek	$7.4B
MiniMax	$4.2B
StepFun	$3.2B
Kling AI	$2.8B
ShengShu Technology	$942M

Is Safe Superintelligence really worth treating as a top-funded AI lab without a major public product?

Yes, because investors are funding Safe Superintelligence primarily as a research bet rather than a product company.

SSI has raised roughly $8 billion in our reconstruction. Its early $1 billion round came from investors including Andreessen Horowitz, Sequoia, DST Global, NFDG and SV Angel. More recently, NVIDIA committed around $5 billion as part of a strategic partnership expected to increase SSI's compute capacity dramatically.

SSI has not built a broad consumer assistant, launched a major enterprise platform or disclosed the type of revenue scale that companies such as OpenAI, Anthropic and ElevenLabs discuss publicly.

Investors are pricing the probability that Ilya Sutskever and his team can produce a major technical breakthrough. Very few companies could raise several billion dollars on that premise alone.

Are investors mostly funding general-purpose AI models?

No. A growing share of large Frontier AI rounds now goes to specialized foundation models.

The top three remain broad general-purpose model developers, but further down the ranking the approaches become much more specialized.

World Labs is pursuing spatial intelligence and world models. Physical Intelligence and Generalist AI are focused on foundation models for robotics. ElevenLabs specializes in voice and audio. Suno builds generative music models. Black Forest Labs focuses on image generation. Runway, Luma AI, PixVerse and Higgsfield are heavily exposed to generative video.

Investors no longer appear to believe that every valuable AI lab has to reproduce ChatGPT. Capital is increasingly flowing to companies trying to own one modality, one type of interaction or one physical domain.

Which specialized AI labs have attracted the most funding?

Physical Intelligence, World Labs, Luma AI, ElevenLabs and Suno are among the clearest examples of specialized labs reaching billion-dollar or near-billion-dollar funding scale.

Physical Intelligence sits around $2.1 billion in our reconstruction and is developing general-purpose robot foundation models. World Labs has raised about $1.2 billion while pursuing spatial intelligence. Luma AI is around $1.1 billion. ElevenLabs has raised $781 million and Suno roughly $775 million.

The gap between these companies and most specialized startups remains huge. A $500 million financing still puts a company near the top 30 of our entire 97-company ranking.

Specialization lowers the need to compete across every modality, but frontier video, spatial, robotics and voice systems still require large research teams, proprietary data pipelines and substantial compute.

How did ElevenLabs become one of the best-funded voice AI companies?

ElevenLabs turned rapid commercial growth into unusually strong fundraising power.

The company raised a $500 million Series D led by Sequoia Capital, bringing total funding to $781 million according to ElevenLabs itself. Andreessen Horowitz and ICONIQ increased their positions, while Lightspeed, BOND and other investors joined the round.

The financial trajectory explains much of that enthusiasm. ElevenLabs said it finished 2025 with more than $330 million in annual recurring revenue and then passed $500 million ARR during the first four months of the following year.

More recently, ElevenLabs said its voice agents were handling more than 15 million conversations every week and that the ARR attached to ElevenAgents had more than tripled since February.

That is a very different funding case from a research lab whose valuation depends mainly on future technical breakthroughs.

Why are investors putting hundreds of millions into AI music and video labs?

Investors are willing to finance creative AI at scale because some of these products are already reaching mass consumer usage.

Suno, for example, raised more than $400 million in its Series D at a $5.4 billion post-money valuation. The company says more than 100 million people have used its tools to create music.

Video has produced an even larger cluster of funded startups. Runway has raised around $860 million in our reconstruction, PixVerse about $568 million, Higgsfield roughly $538 million and Black Forest Labs around $451 million on the image side.

The opportunity is broader than novelty consumer apps. These companies are trying to become production tools for advertising, entertainment, games, social media and professional creative workflows.

Are robotics foundation-model startups attracting serious money yet?

Yes. Frontier AI for robots has already become a billion-dollar funding category.

Physical Intelligence has raised around $2.1 billion in our reconstruction, making it one of the most heavily funded specialized AI labs. Generalist AI follows at approximately $738 million, while several world-model startups also target applications that eventually connect AI reasoning to physical environments.

The funding levels suggest investors increasingly view robotics intelligence as a separate model layer rather than simply another robotics software feature.

The bet is that if foundation-model techniques can generalize across robotic tasks, the winning platform could sit across many robot manufacturers and industries. Whether that generalization works reliably enough in real environments is still unresolved, but the capital is already there.

Which Frontier AI startups recently raised the biggest rounds?

The biggest recent rounds show how far the financing ceiling has moved.

OpenAI closed $122 billion of committed capital. Anthropic followed with a $65 billion Series H. xAI raised $20 billion. Safe Superintelligence secured a multibillion-dollar strategic NVIDIA investment, while Mistral announced €3 billion and Moonshot AI closed roughly $3.5 billion.

Below those giant financings, several younger labs have still raised rounds that would have looked extraordinary only a few years ago. World Labs raised $1 billion. ElevenLabs raised $500 million. Suno raised more than $400 million.

A $100 million AI funding round is no longer exceptional near the frontier. Sixty-three companies in our ranking have already crossed $100 million in cumulative equity funding.

Company	Recent major round
OpenAI	$122B
Anthropic	$65B
xAI	$20B
Safe Superintelligence	~$5B strategic investment
Moonshot AI	$3.5B
Mistral AI	€3.0B
World Labs	$1.0B
ElevenLabs	$500M
Suno	$400M+

Who keeps investing in Frontier AI startups?

NVIDIA, Sequoia, Andreessen Horowitz and several large sovereign or strategic investors appear repeatedly across the Frontier AI funding map.

NVIDIA is especially visible. It has invested in or partnered financially with companies including OpenAI, xAI, Safe Superintelligence, Mistral, Thinking Machines Lab, ElevenLabs and several smaller model developers.

Sequoia appears across Anthropic, SSI, Reflection AI, ElevenLabs, Magic and other companies. Andreessen Horowitz has backed companies ranging from SSI and ElevenLabs to visual, voice and open-model startups.

The investor mix has also broadened well beyond venture capital. Amazon, Microsoft, Samsung, NVIDIA and other technology companies now commit strategic capital, while sovereign investors and government-linked funds play a growing role in the Middle East and China.

At the top end, Frontier AI funding is now partly venture capital and partly an infrastructure and industrial-policy race.

Is NVIDIA becoming the most important investor in Frontier AI?

NVIDIA is one of the most influential financial backers in Frontier AI, largely because its investment strategy reinforces its core compute business.

NVIDIA participated in xAI's $20 billion Series E and became a major strategic partner of Safe Superintelligence. It also appears among the investors or strategic partners of OpenAI, Mistral, Thinking Machines Lab, World Labs, ElevenLabs and numerous smaller companies.

AI labs need enormous quantities of accelerated computing. NVIDIA can invest capital into selected labs, deepen relationships with their technical teams and potentially strengthen demand for NVIDIA systems as those companies scale.

It does not need to pick one winner. NVIDIA can gain exposure to several frontier-model approaches while selling infrastructure across most of the market.

Are smaller Frontier AI labs still able to raise meaningful money?

Yes, but "small" in Frontier AI can already mean tens of millions of dollars.

Our ranking includes companies such as Standard Intelligence at $75 million, Hume AI at $68 million, Nuance Labs at $60 million, Deep Cogito at $56 million, Elorian AI at $55 million and Fish Audio at $52 million.

Further down, startups such as Symbolica AI, Rime, Subquadratic, Sapient Intelligence and Smallest.ai have raised between roughly $20 million and $33 million.

These are still meaningful rounds for technically ambitious model research.

The problem is relative scale. A $30 million startup now sits in a market where the largest competitor has raised more than 6,000 times as much capital. Smaller labs need a sharper technical or market wedge instead of trying to duplicate the full OpenAI stack.

Can a Frontier AI startup compete without raising billions?

Yes, but competing directly with OpenAI or Anthropic across the entire frontier stack is becoming increasingly unrealistic.

A smaller lab can still win by narrowing the problem. ElevenLabs did not need OpenAI-level capital to build a leading position in voice. Suno concentrated on music. Black Forest Labs focused on image models. LangChain built infrastructure around agents rather than training the largest general-purpose model.

Open-source or highly efficient architectures can also reduce the capital required to achieve useful performance.

What becomes much harder is simultaneously funding frontier pre-training, inference infrastructure, global distribution, enterprise sales, consumer products and large research teams.

General-purpose frontier competition is narrowing to a small group of heavily capitalized companies. Specialist competition is much more open.

Does raising more money mean an AI lab has better models?

No. Funding measures financial capacity, not model quality.

DeepSeek is a useful reminder. Its reported funding total is much smaller than OpenAI's or Anthropic's, yet its models have repeatedly influenced the broader discussion around reasoning performance, open weights and training efficiency.

Mistral has also built a major technical reputation with far less capital than the two American leaders. Meanwhile, some companies with enormous financings have disclosed much less publicly about model performance.

Capital matters because compute, researchers and infrastructure are expensive. But the relationship between dollars raised and technical leadership is imperfect.

A funding ranking tells us who has the greatest financial firepower. It cannot tell us on its own which company has the strongest model.

Does raising more money make an AI startup more likely to survive?

Up to a point, yes, because Frontier AI has unusually high fixed costs.

Large funding cushions allow companies to buy compute, recruit expensive researchers, absorb product failures and continue training new model generations before previous investments have fully paid back.

Huge funding also creates huge expectations. A company that raises several billion dollars eventually needs enormous commercial value, an acquisition, public-market access or a technical breakthrough important enough to justify that capital.

Cohere illustrates the commercial pressure. The company has raised around $1.6 billion in our reconstruction and continues positioning itself around secure enterprise and sovereign AI rather than trying to outspend OpenAI in consumer distribution.

Money buys time and options. It does not remove the need for a durable business.

Is the Frontier AI funding boom spreading beyond the United States?

Yes. The market is becoming more geographically diverse even though the United States still dominates the very top.

Europe has Mistral AI and companies such as ElevenLabs. China has Moonshot AI, DeepSeek, Z.ai, MiniMax, StepFun and several large multimodal labs. Japan has Sakana AI. India has Sarvam AI.

Sarvam is a useful example of the new sovereign-AI financing model. The company announced a $234 million first close of a planned $300 million Series B to build models and infrastructure aimed at Indian languages, enterprises and government use cases.

Sakana AI has assembled an unusually broad investor group in Japan, including major banks, industrial companies and technology groups alongside international investors.

Many of these companies are being financed around language, sovereignty, regulation, local infrastructure and national strategic priorities rather than a simple attempt to copy the biggest US labs.

Is sovereign AI becoming a real funding category?

Yes. Sovereign AI has moved from a policy slogan into a genuine investment theme.

Mistral explicitly frames part of its strategy around European technological sovereignty. Sarvam is building a full-stack AI platform for India. Sakana AI has attracted backing from a long list of major Japanese financial and industrial groups. China's largest model labs are increasingly supported by domestic technology companies and state-linked capital.

Frontier models are becoming infrastructure. Governments and large enterprises increasingly care about where models are trained, where data is processed, which chips and cloud providers they depend on and whether they can control the technology long term.

The biggest American labs still have enormous economic advantages, but sovereign demand creates room for regional champions that do not need to beat OpenAI globally to become valuable companies.

Are AI agent companies part of the Frontier AI funding race?

Some are, although the distinction between model labs and agent infrastructure is becoming blurry.

LangChain has raised about $160 million in our reconstruction after a $125 million Series B. The company said its LangChain and LangGraph projects were seeing around 90 million monthly downloads combined and that 35% of the Fortune 500 used its services.

Companies such as H, Rogo and several coding-focused startups are also attracting substantial funding around agentic systems.

The technical value sits in a different place. OpenAI and Anthropic train massive base models, while agent companies often build orchestration, tools, memory, observability or domain-specific workflows on top of those models.

Both can become large businesses, but they have very different capital requirements. That is why a company such as LangChain can become strategically important without needing tens of billions of dollars for model training.

Has Frontier AI funding become too expensive for normal venture capital?

At the very top, yes. The largest Frontier AI rounds now exceed what traditional venture funds can finance on their own.

A $65 billion Anthropic round or $122 billion OpenAI financing requires capital from strategic technology companies, sovereign investors, major asset managers and other institutional investors alongside venture firms.

Even $3 billion to $5 billion rounds stretch beyond conventional startup financing.

Traditional venture capital still matters enormously at the early stages and inside syndicates. Firms such as Sequoia, Andreessen Horowitz, Lightspeed and others continue to appear throughout the ranking.

Once a lab reaches hyperscale, though, financing starts to look closer to infrastructure or late-stage corporate capital than classic venture investing.

Will the biggest Frontier AI labs keep raising even more money?

Probably. Current financing patterns suggest the capital race still has room to run.

OpenAI says durable compute access is central to its strategy. Anthropic has announced multi-gigawatt infrastructure commitments. xAI built massive GPU clusters and used its latest financing partly to expand compute. Mistral, Sarvam and other challengers also explicitly connect new funding to model training and infrastructure.

The economics keep pushing in the same direction: more users create more inference demand, stronger models require more training resources, and companies want enough capacity to avoid being completely dependent on one infrastructure supplier.

Efficiency improvements could change that curve, so every future model generation does not automatically need proportionally more capital.

For now, the financing behavior of the largest labs gives little sign that the funding race is nearing a natural stopping point.

So who is actually winning the Frontier AI funding race?

OpenAI is clearly winning on cumulative funding, with Anthropic the only company close enough to sit in the same financial class.

Our ranking puts OpenAI at about $183 billion, Anthropic at $114.8 billion and xAI at $37.1 billion. After that, funding drops below $10 billion.

That gap is the main conclusion from the dataset. Frontier AI may look crowded when we count companies, models and product launches, but the financial structure is remarkably narrow. A tiny group controls most of the capital.

Funding still does not settle the technical race. DeepSeek has shown that smaller budgets can produce influential models, while specialist companies such as ElevenLabs, Suno, World Labs and Physical Intelligence are building businesses around narrower forms of intelligence.

OpenAI and Anthropic dominate the general-purpose capital race. The broader market remains open wherever a company can avoid competing with them on every dimension at once.