What’s funding like in the Frontier AI Labs market right now?

All funding deals in the Frontier AI Labs Market over the last 24 months

Below is the table listing all the deals. You can find our methodology at the end of this page.

If you want a deeper understanding of the market and its current dynamics, get our report covering the Frontier AI Labs Market.

Is Frontier AI funding actually healthier now?

Frontier AI funding is healthier today in terms of company formation, breadth and typical round size, even though the money itself has become much more concentrated.

We counted 76 financings in the latest 12 months, up from 43 in the preceding period, while the number of funded companies doubled from 31 to 62. Median disclosed round size also climbed from $77 million to $210 million. More companies are getting funded, and the typical successful financing is substantially larger.

The catch is concentration. Usable disclosed capital jumped from $32.9 billion to $253.6 billion, with OpenAI's $122 billion round alone representing 48.1% of the current total. Three rounds absorbed 85.6% of all capital. We therefore have a broader market at the company level and a much steeper hierarchy at the top.

That makes Frontier AI stronger than it was a year earlier, but in a very specific way: startup formation has accelerated while a handful of established labs have moved into a capital category of their own.

Metric	Current L12M	Previous 12M	Change
Deals	76	43	+76.7%
Funded companies	62	31	+100.0%
Companies raising more than once	10	6	+66.7%
Median disclosed round	$210M	$77M	+172.7%
Usable disclosed capital	$253.62B	$32.91B	+670.6%
Largest round	$122B	$13B	+838%
Top 3 share of capital	85.6%	68.4%	+17.2 pp

Would Frontier AI funding still look stronger without the giant rounds?

Yes. Frontier AI funding would still look stronger without the giant rounds, although the underlying improvement is far smaller than the headline 671% jump in capital.

When we remove rounds above $50 million, capital rises from $455.6 million to $516.3 million, or 13.3%. With rounds above $100 million removed, the increase is 30.6%, from $1.09 billion to $1.42 billion.

Those figures are much less spectacular than $253.6 billion, but they still point upward. Deal activity rose sharply at the same time, and the median financing increased enough that this cannot be explained by one strange quarter or a couple of giant checks.

The difference between headline and underlying growth is still huge. OpenAI closed $122 billion, while Anthropic raised $30 billion and later $65 billion. Those three deals transformed the capital total.

So saying Frontier AI funding increased almost eightfold is technically right and analytically misleading. Ordinary funding improved moderately while the biggest labs entered an entirely different financing regime.

Are more Frontier AI companies getting funded now?

Yes, Frontier AI funding is reaching far more companies now, which is one of the clearest improvements we found.

The number of companies funded during the period doubled to 62. Because deal count rose less quickly than company count, the repeat-fundraising rate actually slipped from 19.4% to 16.1%.

That's useful. If activity had surged mainly because OpenAI, Anthropic, Moonshot AI and a few other names kept coming back for money, we would have seen the opposite pattern. Instead, new names expanded the funded-company base considerably.

We can see that breadth in very different parts of the market. World Labs raised $1 billion around spatial intelligence, Black Forest Labs raised $300 million around visual models, Deepgram raised $130 million for real-time voice AI, and a new group of agentic-model companies also entered the dataset.

As of now, Frontier AI has a much larger bench underneath the handful of labs dominating the financial headlines.

Are investors still creating new Frontier AI startups, or just backing the winners?

Investors are creating many more new Frontier AI companies these days, although the established winners still receive almost all of the dollars.

First financings rose from five to 21 and went from 11.6% to 27.6% of all deals. Median first financing increased from $40 million to $85 million, while known capital going into those initial rounds rose from $2.13 billion to $7.58 billion.

Investors are also entering earlier. Median company age at first financing fell from roughly 0.96 years to 0.75 years, while Seed-stage median age dropped from 1.29 years to 0.92 years.

At the other end, 54 follow-on financings absorbed about $246 billion. Their median size reached $250 million, up from $90 million previously.

The market is producing more startups without weakening the position of its biggest incumbents. If anything, the two trends are feeding each other: more labs are being formed, while proven labs can raise vastly more once investors believe they have a real chance of staying near the frontier.

Metric	Current L12M	Previous 12M
First financings	21	5
First-financing share	27.6%	11.6%
Known first-financing capital	$7.58B	$2.13B
Median first financing	$85M	$40M
Follow-on financings	54	38
Known follow-on capital	$245.97B	$30.79B
Median follow-on	$250M	$90M

Has a $100 million Frontier AI round become normal?

For Frontier AI labs, a $100 million financing is now common enough that it no longer feels exceptional.

In the current period, 48 disclosed rounds were worth at least $100 million, representing 63.2% of financings with usable size data. The comparable share was 41.9% a year earlier.

Meanwhile, the $20 million to $50 million band fell from 27.9% of deals to 11.8%. The middle of the distribution is getting thinner as more activity moves toward much larger checks.

The company examples make this easier to grasp. World Labs raised $1 billion. ElevenLabs closed a $500 million Series D. Black Forest Labs raised $300 million. Recursive Superintelligence emerged with a $650 million financing. Even Sakana AI, which openly argues that AI development needs a more resource-efficient model, announced a roughly $200 million Series B.

Frontier AI increasingly has a barbell financing structure: younger teams can still raise formation capital, while companies seen as plausible category leaders quickly move into nine-figure territory.

Round size	Current deals	Current share	Previous deals	Previous share
Under $5M	2	2.6%	0	0%
$5M–<$20M	6	7.9%	3	7.0%
$20M–<$50M	9	11.8%	12	27.9%
$50M–<$100M	9	11.8%	7	16.3%
$100M+	48	63.2%	18	41.9%
Above $50M, all deals	54	71.1%	24	55.8%
Above $100M, all deals	43	56.6%	16	37.2%

Are Frontier AI investors taking more risk now?

Frontier AI investors are taking more early-company risk now while becoming even more aggressive with the companies that clear their bar.

Early-stage transactions increased from 34.9% to 40.8% of deals, and we found far more first financings. Young labs are clearly getting a chance.

Yet early-stage rounds accounted for only about 2.2% of current capital. Later-stage financings captured roughly 96.7%, with five Growth Equity deals alone representing around 53% of all usable capital.

Investors seem comfortable placing many comparatively small bets at formation and then concentrating enormous amounts behind the names that establish technical, commercial or strategic importance.

xAI's $20 billion Series E is a good recent example of the second part. The company directly linked the financing to expanding Colossus, scaling compute infrastructure and deploying products. At the other end, new agentic-model companies have raised Seed rounds before establishing anything close to that commercial scale.

Frontier AI investors currently look adventurous when choosing teams and highly concentrated when allocating serious capital.

Which parts of Frontier AI are gaining real momentum?

Multimodal models, voice AI and agentic-model platforms currently show the clearest momentum beyond general-purpose language models.

Multimodal financing grew from nine deals to 17, while known capital increased from $2.7 billion to about $8.1 billion. Median round size almost doubled from $180 million to $355 million. World Labs' $1 billion financing fits the broader shift toward world models, spatial intelligence and systems that work across more than text.

Audio and speech also broadened, from seven deals to 15. Median financing rose from $22 million to $52 million. ElevenLabs' $500 million Series D and Deepgram's $130 million Series C show how far voice AI has moved beyond small specialist-model rounds. ElevenLabs later said it had crossed $500 million of ARR during 2026, driven heavily by enterprise use of voice agents.

Agentic Model Platforms went from no dedicated category in the previous period to nine transactions, with a $200 million median. That's a genuine new financing cluster in our dataset.

Video AI moved differently. We recorded slightly fewer deals, but its median financing jumped from $70 million to roughly $296 million. Money appears to be concentrating around a smaller group of well-capitalized video platforms rather than spreading evenly across the category.

Category	Current deals	Previous deals	Current capital	Current median
Language & Reasoning	25	16	$236.9B	$367M
Multimodal	17	9	$8.05B	$355M
Audio & Speech	15	7	$1.87B	$52M
Agentic Model Platforms	9	0	$1.69B	$200M
Video Generation	8	9	$4.69B	~$296M
Image Generation	2	2	$376M	$188M
Total	76	43	$253.62B usable	$210M overall

Is language-model funding still swallowing the Frontier AI market?

Language and reasoning models still swallow most Frontier AI capital today, although they no longer dominate company activity to the same degree.

Their share of deals fell from 37.2% to 32.9%, while the category absorbed about $236.9 billion, or more than 93% of current usable capital.

OpenAI and Anthropic explain most of that disconnect. Add large financings at xAI, Mistral AI, Safe Superintelligence, Reflection AI and several Chinese labs, and language-model companies are operating at a scale that voice, image and most multimodal startups have not yet reached.

As seen above, the rest of Frontier AI is still expanding. Multimodal and voice gained deal breadth. The difference is that the biggest language-model companies now require amounts of capital capable of overwhelming an entire category comparison.

The funding mix currently tells two stories at once: model innovation is spreading across modalities, while infrastructure-scale financing remains centered on the major general-purpose labs.

Are Frontier AI companies raising again faster than before?

Frontier AI companies are actually taking longer to raise again, but the next round tends to be bigger when they return.

Among follow-ons where we could identify a previous financing in the same market, the median gap increased from 5.6 months to 9.3 months. The share returning within 12 months fell from 75% to 62.5%.

Round progression moved the other way. Median step-up increased from 1.56x to 1.73x, and 40% of current follow-ons were more than twice the size of the previous known financing. Rounds below 0.8x of the prior amount fell from 38.1% to 23.3%.

That's a more interesting pattern than simply raising faster. Companies are giving themselves more time between rounds, yet successful returners are generally able to ask for more.

We should be careful with this metric because previous-round dates were observable for only part of the follow-on population. Still, within the comparable sample, the direction is pretty clear: fundraising slowed in cadence while strengthening in check size.

Do Frontier AI startups need more proof to raise today?

The proof required for Frontier AI funding now depends heavily on the size of the round. Brand-new labs can still raise very early, but the giant checks increasingly sit behind companies with serious adoption, revenue or infrastructure needs.

We can see the first half in company age. The median company financed overall barely changed, at about 2.7 years, while first-time financings actually moved younger.

The upper end looks very different. OpenAI said around the time of its latest financing that monthly revenue had reached roughly $2 billion, enterprise customers represented more than 40% of revenue, and its APIs were processing more than 15 billion tokens per minute. Anthropic said shortly before its latest round that run-rate revenue had crossed $47 billion. ElevenLabs reported more than $500 million of ARR after previously closing its $500 million Series D.

These figures do not establish a universal funding threshold, and we do not have a clean structured milestone field for every transaction. They do show what sits behind some of the market's largest checks now: technical leadership plus massive real-world usage.

At formation stage, pedigree and technical ambition can still unlock capital. At the top end today, investors increasingly want evidence that huge compute budgets can become products, customers and revenue.

Is Europe actually catching up in Frontier AI?

Europe is catching up in Frontier AI company activity, but North America still dominates the money by an enormous margin.

Europe's share of deals increased from 7.0% to 15.8%. Its capital share, however, declined from 6.6% to 2.6% because North American megadeals grew much faster than European funding.

Mistral AI shows both sides of that story. The company announced a €3 billion Series D during the current period, while continuing to push sovereign and open-weight AI from Europe. ElevenLabs also raised $500 million while expanding an increasingly international enterprise voice business.

North America, meanwhile, fell from 60.5% to 53.9% of deal activity but increased its capital share to 91.4%. OpenAI and Anthropic make that result almost inevitable.

Asia-Pacific also gained ground, moving from 27.9% to 30.3% of deals and from 3.6% to 5.9% of capital. Chinese financings remained numerous, while several rounds moved into the hundreds of millions or billions.

The geography of Frontier AI startups is becoming more international. The very largest pools of funding are still heavily tilted toward North America.

Region	Current deal share	Previous deal share	Current capital share	Previous capital share
North America	53.9%	60.5%	91.4%	89.3%
Asia-Pacific	30.3%	27.9%	5.9%	3.6%
Europe	15.8%	7.0%	2.6%	6.6%
Middle East	0%	4.7%	0%	0.4%
Europe change	+8.8 pp	-	-4.0 pp	-
APAC change	+2.4 pp	-	+2.3 pp	-
North America change	-6.5 pp	-	+2.1 pp	-

Are strategic investors becoming unavoidable in Frontier AI?

Strategic investors are becoming a central part of Frontier AI financing, with participation rising from 39.5% to 71.1% of deals.

This shift reaches well beyond late-stage rounds. We counted strategic investors across Seed, Series A, Series B, Series C, later-stage and Growth Equity financings. NVIDIA alone appeared in 18 current transactions, compared with seven previously.

The reason has become easier to see lately. The largest AI labs need capital, but they also need chips, cloud capacity, power, networking, memory and long-term infrastructure commitments. A strategic investor can potentially provide several of those things at once.

Anthropic's latest financing included Micron, Samsung and SK hynix and came alongside huge compute agreements with Amazon, Google, Broadcom and SpaceX. xAI said NVIDIA and Cisco participated in its Series E while it expanded the Colossus compute clusters. ElevenLabs brought NVIDIA into a later Series D close while scaling enterprise voice AI.

NVIDIA itself made the trend explicit in its latest earnings discussion: the company said frontier AI labs were growing faster than their balance sheets could support and disclosed nearly $50 billion of investments in frontier labs. It has also partnered with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR on platforms intended to mobilize more than $500 billion for AI infrastructure.

Frontier AI financing now sits much closer to the compute supply chain than conventional software venture capital does.

Investor measure	Current L12M	Previous 12M
Deals with strategic investors	54	17
Strategic participation	71.1%	39.5%
Unique disclosed investors	142	56
Investors appearing more than once	31	7
Deals with disclosed lead investors	55	27
NVIDIA participations	18	7
Median strategic-backed round	~$296M	$279M

Are fewer investors controlling Frontier AI funding?

No. The Frontier AI investor pool is much broader now even though company-level capital has become more concentrated.

We identified 142 disclosed investors in the current period versus 56 previously. The number appearing in more than one deal rose from seven to 31.

That mix is important. More institutions are entering the market, and a larger group is returning for multiple transactions. We're seeing venture firms, sovereign investors, asset managers, chip companies, cloud providers, telecom groups and major technology companies in the same financing ecosystem.

Sakana AI's investor base is a useful example of how broad these syndicates can become, spanning Japanese financial groups and corporates alongside Google, Salesforce Ventures, In-Q-Tel, Khosla Ventures and other international investors. Black Forest Labs' $300 million Series B brought together venture firms, strategic technology companies and creative-software groups.

So the concentration problem currently sits more on the recipient side than the investor side. Many more investors are active; a small group of companies can absorb extraordinary amounts of their capital.

What are Frontier AI companies actually raising all this money to do?

The biggest Frontier AI labs are increasingly raising to secure compute and scale real products, while smaller labs still use funding for core research and initial commercialization.

OpenAI currently describes durable compute access as one of its strategic advantages and has spread its infrastructure across Microsoft, Oracle, AWS, CoreWeave and Google Cloud, with silicon from NVIDIA, AMD, Trainium, Cerebras and its own Broadcom partnership.

Anthropic's latest financing was explicitly tied to safety and interpretability research, more compute, products and partnerships. The company has since disclosed an Amazon agreement for up to five gigawatts of capacity and a SpaceX arrangement giving it access to more than 220,000 NVIDIA GPUs at Colossus.

xAI said its $20 billion Series E would help expand infrastructure, deploy new products and continue research. ElevenLabs used its Series D to expand its enterprise voice-agent platform, international go-to-market footprint and underlying model research. Sakana AI, which follows a much more compute-efficient philosophy, said its Series B would fund R&D, commercial deployment and a broader Japanese AI ecosystem.

We cannot convert these examples into a market-wide percentage because use of funds was not consistently disclosed in a structured way. Still, the direction at the top is clear: funding frontier models now increasingly means financing the infrastructure and distribution required to run them at scale.

Is Frontier AI funding becoming an infrastructure market?

Yes. The upper end of Frontier AI funding now looks increasingly like infrastructure finance layered on top of venture equity.

OpenAI's current infrastructure portfolio spans multiple clouds, chip architectures and data-center partners. Anthropic has committed to multi-gigawatt compute agreements. xAI built Colossus around enormous GPU deployments. NVIDIA is working with some of the world's largest alternative-asset and financial institutions to create dedicated compute-financing structures.

The amounts involved explain the change. A traditional software company can usually scale revenue far faster than physical infrastructure. Frontier-model companies need both: software economics on the demand side and data-center economics on the supply side.

That creates a funding problem ordinary venture capital was never designed to solve alone. Equity remains crucial, but it increasingly sits beside cloud commitments, infrastructure partnerships, strategic investments and specialist pools of capital.

This is why the largest Frontier AI rounds should not be compared too casually with conventional SaaS financings. The money supports a much heavier operating stack.

Is Frontier AI funding strong every month, or are a few months doing all the work?

Frontier AI deal activity is consistently stronger these days, while the dollar totals still swing wildly depending on when the megadeals land.

Average monthly deal activity increased from 3.6 to 6.3 and the median rose from 3.5 to 5.5. Neither period had a zero-deal rolling month, while the busiest current window reached 16 financings.

Capital is far lumpier. Average monthly funding reached roughly $21.1 billion, compared with a median of only $3.84 billion. One rolling window contained about $123 billion because OpenAI's round landed inside it.

This is why deal count gives a cleaner view of current market activity than monthly capital. The flow of financings strengthened across the year, whereas the capital chart is dictated by the timing of a few gigantic checks.

The better evidence of a broader market is the steady stream of companies getting funded, not the shape of the monthly dollar graph.

What is the most misleading headline about Frontier AI funding right now?

The most misleading Frontier AI headline would be "funding rose 671%, so the market is eight times healthier."

That figure is mathematically correct, but it mixes ordinary venture rounds with OpenAI's $122 billion financing and Anthropic's extraordinary capital requirements. Remove the largest checks and growth drops into the tens of percent rather than the hundreds.

The opposite take would miss what's happening too. Calling the boom purely an OpenAI-and-Anthropic effect ignores the rise in company formation, larger first rounds, growth in multimodal and voice financing, and the emergence of an agentic-model funding cluster.

As pointed out above, the market has broadened underneath the megadeals. The important change is less dramatic than the headline capital increase but probably more durable: more teams are being funded, while companies that establish themselves near the top can now access amounts of money previously associated with major infrastructure projects.

What really changed in Frontier AI funding over the last 12 months?

Frontier AI funding is fundamentally stronger today, but the improvement came with much heavier concentration at the top.

The first big change is breadth. Twice as many companies received funding, and new-lab formation accelerated sharply. Investors are still willing to make early bets rather than simply protect the incumbents.

The second is financing scale. Nine-figure rounds now occupy the center of the market far more often, and the traditional $20 million to $50 million band has lost a lot of weight. Once a Frontier AI company is viewed as strategically important, investors are prepared to capitalize it very aggressively.

The third is category expansion. Multimodal and voice companies gained both activity and larger rounds, while agentic-model platforms appeared as a new cluster. Video funding became more selective, with fewer companies receiving substantially larger checks.

The fourth is the growing role of infrastructure. Recent OpenAI, Anthropic and xAI announcements all link capital directly to compute capacity and large-scale deployment. NVIDIA's recent disclosures push the point even further: financing the AI labs and financing the AI factories behind them are starting to overlap.

The fifth is investor breadth. Far more investors are participating now, so the market has not become dependent on a shrinking financing club. Strategic companies, however, have become much more prominent because access to compute, chips, cloud capacity and distribution can matter almost as much as the cash itself.

Finally, the giant funding total should be handled carefully. The $253.6 billion headline tells us how capital-intensive the leading Frontier AI labs have become. It is much less useful for judging the health of the average company. For that, the increase in funded companies, new entrants, median rounds and activity outside the biggest deals gives us a much clearer answer.