Frontier AI Labs Funding: A Quarter-by-Quarter Analysis

All funding deals in the Frontier AI Labs Market over the last years

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 slowing down?

Frontier AI Labs funding is cooling sharply in dollar terms, but more companies are getting financed than during most of 2025.

We tracked 100 financing transactions between Q2 2025 and Q3 2026. Investment peaked at $157.2 billion in Q1 2026, declined to $97.32 billion in Q2 and fell again to $16.97 billion in Q3. That represents an 89% decline from the peak in just two quarters.

Deal activity tells a rather different story. We counted 12 transactions in Q1, 32 in Q2 and 23 in Q3. Despite the latest slowdown, Q3 still had more than twice as many deals as Q3 2025, when only nine financings were recorded.

Round sizes also changed considerably. The median reached $600 million in Q1 2026 before dropping to $234 million in Q2 and $76 million in Q3.

There is also a major difference between committed capital and money actually deployed. Some of the largest financings involve structured commitments, making headline totals less useful for measuring how much fresh equity reached companies during a particular quarter.

Quarter	Deals	Companies	Recorded capital	Median round
Q2 2025	12	12	$4.53B	$32M
Q3 2025	9	8	$16.32B	$140M
Q4 2025	12	12	$4.74B	$192.5M
Q1 2026	12	12	$157.20B	$600M
Q2 2026	32	28	$97.32B	$234M
Q3 2026	23	23	$16.97B	$76M

Was the Frontier AI funding boom basically just OpenAI and Anthropic?

OpenAI and Anthropic explain much of the extraordinary growth in Frontier AI funding, and their billion-dollar rounds have created a market where a handful of laboratories receive almost all the capital.

On March 31, 2026, OpenAI announced a $122 billion financing involving committed capital at an $852 billion post-money valuation. That single transaction represented approximately 77.6% of all recorded frontier AI investment during Q1.

Only 12 financings took place, exactly the same number as in Q4 2025. Yet total capital jumped from $4.74 billion to $157.2 billion.

The concentration was already visible in the previous year. Anthropic announced a $13 billion Series F on September 2, 2025, accounting for almost 80% of Q3 funding. A few months later, the market entered another period dominated by exceptionally large rounds.

The pattern has persisted throughout our six-quarter analysis. In Q2 2025, the three largest financings accounted for 95.1% of recorded capital. Q1 2026 was similarly extreme, with its three largest transactions representing 97.3% of funding.

Even after the latest correction, the five largest Q3 2026 rounds accounted for approximately 90% of total investment.

One reason for this concentration is the amount of infrastructure required to build and operate frontier models.

OpenAI's March announcement described computing capacity, enterprise deployment and consumer distribution as central parts of its strategy. Anthropic's May 2026 Series H similarly outlined plans to expand computing capacity, fund safety research and support growing enterprise demand. The company also disclosed major infrastructure agreements involving Amazon, Google, Broadcom and SpaceX.

There is an important distinction between a financing announcement and an entirely new injection of capital. Anthropic confirmed that $15 billion of its Series H consisted of previously committed investments from hyperscalers, including $5 billion from Amazon.

Simply adding every announced financing can therefore overstate the amount of genuinely new capital available during a quarter.

Our calculations also show that Q1 capital from transactions worth $50 million or less was just $30 million, compared with $100 million in the previous quarter.

The defining feature of frontier AI funding today is the distance between the largest platforms and ordinary venture-backed model developers.

Quarter	Largest deal share	Top 3 share	Top 5 share
Q2 2025	44.1%	95.1%	97.4%
Q3 2025	79.6%	94.9%	97.6%
Q4 2025	42.2%	71.7%	83.4%
Q1 2026	77.6%	97.3%	98.4%
Q2 2026	66.8%	84.4%	92.1%
Q3 2026	29.5%	70.7%	90.1%

Why did so many Frontier AI startups suddenly raise money in Q2 2026?

Frontier AI startup fundraising expanded dramatically in Q2 2026, with more companies attracting capital across several model categories.

Transactions jumped from 12 to 32, an increase of 166.7%. The number of funded companies rose from 12 to 28, confirming that the additional activity reached substantially more businesses.

Early-stage financings increased from two to 15. First financings rose from two to 12. Language and reasoning model deals increased from three to 13, while multimodal funding rounds climbed from three to eight.

Commercial demand was also developing quickly around the same period.

In April 2026, OpenAI reported that more than 40% of its revenue came from enterprise customers. Its Codex product had reached three million weekly active users, while the company's APIs were processing more than 15 billion tokens per minute.

Anthropic's February 2026 Series G announcement had already connected its $30 billion financing to demand for enterprise AI and coding products.

These company disclosures help explain why investors were increasingly interested in financing different parts of the frontier AI ecosystem. The largest laboratories were demonstrating substantial commercial demand, while smaller companies were pursuing specialized capabilities and products.

We cannot attribute the quarterly jump to one announcement or product launch. But Q2 was clearly the first quarter in our analysis when the increase in funded businesses became as interesting as the capital totals themselves.

Did Frontier AI startups lose momentum again in Q3 2026?

Frontier AI fundraising weakened in Q3 2026, particularly for new companies and large transactions, although smaller rounds held up surprisingly well.

Deal volume fell 28.1%, from 32 to 23 transactions. The median financing declined from $234 million to $76 million.

First financings suffered the sharpest drop, falling from 12 to five. Follow-on deals were more resilient, declining only from 19 to 18.

There was also a substantial change in the size of the largest transactions. The biggest Q2 financing was Anthropic's $65 billion Series H, announced on May 28, 2026. The largest Q3 transaction was $5 billion.

With fewer enormous financings, the market's headline capital fell dramatically.

However, investment from rounds worth $50 million or less increased from $158.8 million to $227.5 million.

Q3 was also stronger than several earlier quarters when measured by funded companies. Its 23 companies compared with just eight in Q3 2025 and 12 in Q4 2025.

The latest quarter therefore shows a genuine slowdown following Q2's expansion, with the most noticeable weakness among first-time financing recipients and companies seeking very large checks.

Are smaller Frontier AI startups finally getting more funding?

Smaller Frontier AI financing rounds have recovered since the beginning of 2026, and that improvement continued even during the latest market slowdown.

Funding from transactions worth $50 million or less was exceptionally limited during much of 2025. It stood at $169.7 million in Q2, disappeared entirely in Q3 and recovered to $100 million in Q4.

Activity then fell to just $30 million in Q1 2026.

The following two quarters marked a change. Smaller-round funding increased to $158.8 million in Q2 and $227.5 million in Q3.

Interestingly, the number of rounds in this smaller financing bracket remained at eight in both quarters. The amount invested increased without a corresponding rise in transaction count.

Investors were still willing to write meaningful checks to smaller companies even as the largest financing commitments became less frequent.

We now have two consecutive quarters showing a more active smaller-financing segment. That's an early sign of recovery, although the market remains far from demonstrating uninterrupted growth.

Are investors becoming more willing to fund young Frontier AI labs?

Early-stage Frontier AI companies are getting a much larger share of financing deals than they did at the beginning of 2026, although established laboratories still receive almost all the money.

Early-stage transactions represented just 16.7% of deals in Q1 2026. Their share rose to 46.9% in Q2 and 47.8% in Q3.

That is a substantial change from Q3 2025, when all nine recorded financings were later-stage transactions.

In Q2 2026, we identified 15 early-stage deals and 13 later-stage deals. Q3 recorded 11 early-stage and ten later-stage financings, with two transactions lacking a reliable stage classification.

Capital allocation remained much less balanced. In Q3, early-stage companies raised $430 million, compared with $15.82 billion for later-stage companies.

Company age and financing maturity are different things in frontier AI, though.

A recently founded laboratory can raise an exceptionally large first round if it has an experienced research team, valuable intellectual property or major strategic backing. An older company may still operate with comparatively limited financing.

Our founding-year review also uncovered several unreliable dates, including spreadsheet serial values and apparent 1905 entries affecting companies such as Mistral AI, DeepSeek and Z.ai. We excluded those malformed values from precise age calculations.

An important development in March 2026 helps illustrate how the boundaries between general-purpose models and more specialized development were evolving.

Mistral introduced Forge, a system designed to help enterprises build AI models using proprietary company knowledge. The same month, it released Mistral Small 4, combining reasoning, multimodal understanding and coding capabilities.

These launches show how model development was moving toward specialized enterprise applications alongside the continuing race for general-purpose intelligence.

For now, the funding evidence is clear: early-stage companies are gaining participation, even though their financial resources remain tiny relative to the leading laboratories.

Is the boom in newly funded Frontier AI labs already over?

First-time Frontier AI fundraising cooled sharply after its Q2 2026 peak, making the supposed boom in new laboratories look increasingly temporary.

We identified 12 first financings in Q2, representing 37.5% of all transactions. That was a substantial increase from just two in Q1.

But Q3 recorded only five first financings, reducing their share to 21.7%.

The size of those transactions also changed. Median first financing fell from $210 million in Q2 to $40 million in Q3. Meanwhile, the median follow-on round declined from $301.3 million to $125 million.

As seen above, follow-on activity remained comparatively stable despite the overall drop in deal volume.

First financing should not automatically be interpreted as the creation of a new company. Some frontier AI laboratories begin with experienced research teams, established commercial relationships and unusually large initial commitments. Our classification reflects identifiable financing history rather than a complete reconstruction of every company's capital formation.

Quarter	First financings	Follow-on financings	First-financing share
Q2 2025	3	9	25.0%
Q3 2025	0	9	0%
Q4 2025	3	9	25.0%
Q1 2026	2	10	16.7%
Q2 2026	12	19	37.5%
Q3 2026	5	18	21.7%

Are language models still taking most of the Frontier AI funding?

Language and reasoning model developers still dominate Frontier AI fundraising, capturing far more capital than their share of deals would suggest.

In Q2 2026, the category accounted for 13 of 32 transactions but 94.4% of investment.

That imbalance had been visible throughout the period. Language and reasoning models attracted 96.4% of capital in Q3 2025 and 97.1% in Q1 2026.

Their share fell to 72.6% in Q3, although the category remained the largest recipient of funding.

The huge commitments to OpenAI and Anthropic explain much of this dominance. These companies are building widely used models and platforms that require major investments in training, inference, distribution and infrastructure.

However, the category is also becoming more varied.

Mistral's September 2025 Series C, worth €1.7 billion, was led by semiconductor equipment manufacturer ASML. The partnership connected frontier model development with industrial engineering and specialized commercial applications.

More recently, Mistral announced a €3 billion Series D in September 2026, led by Samsung Electronics, supporting research, computing infrastructure and international growth.

Language-model investment now extends well beyond the largest American platforms, even though their financing scale continues to dominate the global picture.

Are multimodal AI startups becoming the next funding hotspot?

Multimodal AI is emerging as one of the strongest areas for new frontier-model financing, although its share of capital remains modest.

The category recorded three transactions in Q1 2026, eight in Q2 and six in Q3.

That means multimodal deal activity remained twice as high in Q3 as it was at the beginning of the year, even after the broader market slowed.

Capital share also improved across the same three quarters, rising from 1.7% to 4.0% and then 8.4%.

Technological developments provide useful context.

Mistral Small 4, released in March 2026, brought reasoning, image understanding and agentic coding capabilities into a unified model. More recently, Mistral Large 4 expanded the company's multimodal approach at a much larger model scale.

Google has also been improving AI models that work across text, images and video. These developments reflect a broader move toward systems that can handle different types of information within the same workflow.

Multimodal AI currently has stronger evidence of sustained growth in deal participation than most other specialist model categories, even though its funding amounts remain far below those flowing to leading language-model platforms.

Are AI agents really attracting more startup funding?

Agentic AI funding is growing from a small base, but the quarterly figures show a volatile market rather than steady acceleration.

We recorded no agentic platform deals in Q2 or Q3 2025, followed by one in Q4 and another in Q1 2026.

Activity then jumped to five transactions in Q2 before declining to two in Q3.

The category attracted 15.6% of financing events in Q2 but only 0.9% of capital. Investors were financing a wider set of relatively smaller companies rather than making enormous commitments to agentic specialists.

Commercial activity around AI agents was becoming more concrete during the same period.

OpenAI launched its Frontier enterprise agent platform in February 2026 and expanded its enterprise deployment efforts during the following months. In April, the company also introduced shared workspace agents designed to work across business tools and organizational processes.

Its April announcement about Codex adoption named customers including Cisco, Ramp, Notion and Rakuten, illustrating how agentic tools were entering practical software-development workflows.

These developments provide a plausible commercial reason for growing investor attention. But the deal-count decline in Q3 shows that enthusiasm had not yet translated into a consistently growing financing market.

Why did video AI suddenly attract so much investment?

Video-generation AI became a much larger part of Frontier AI funding in Q3 2026 because two unusually large financings pushed the category's capital share sharply higher.

The category represented just 0.3% of investment in Q2, compared with 16.5% in Q3.

Those two Q3 deals accounted for approximately $2.8 billion in aggregate funding.

Transaction activity increased only modestly, from one deal to two, so the jump in capital reflects financing size much more than a broad increase in funded companies.

The technology had been progressing quickly before this funding change.

Google introduced Veo 3 in May 2025, combining video generation with synchronized audio. Its Veo 3.1 updates in January 2026 improved creative control, character consistency and output quality, including support for higher-resolution video.

These advances help explain the investment interest. Video models are moving toward capabilities that can support professional content creation and commercial applications.

Audio and speech companies, meanwhile, recorded four Q3 2026 financings yet captured less than 1% of capital.

Video AI clearly attracted larger checks during the latest quarter, although two transactions provide limited evidence of a lasting funding boom.

Is Frontier AI investment moving away from the United States?

Frontier AI fundraising is becoming more international, although North America still leads in deal activity and remains home to many of the largest recipients.

We identified six North American deals in Q1 2026, followed by 18 in Q2 and 12 in Q3.

Asia-Pacific followed a similar progression, with four, nine and seven transactions. Europe recorded two, five and four respectively.

The regional capital distribution changed more dramatically.

North America's share fell from approximately 85.5% in Q2 to 36.0% in Q3. Asia-Pacific attracted $7.24 billion in the latest quarter, while Europe received $3.62 billion.

These changes partly reflect the timing of extraordinary American financing announcements. A single enormous transaction can temporarily make North America appear much more dominant than its underlying share of deal activity suggests.

There are nevertheless reasons to take the international expansion seriously.

Mistral's recent financing and infrastructure plans highlight Europe's growing investment in domestic frontier AI capabilities. In August 2026, the company outlined plans to support regional model deployment and expand European computing capacity, with a longer-term infrastructure target reaching up to one gigawatt.

The interest in sovereign AI, regional infrastructure and greater control over model deployment could support more international investment over time.

Our funding figures already show more Asia-Pacific and European deal activity than during most of 2025. The durability of their capital gains remains less certain because of the influence of individual large rounds.

Are more venture capital firms entering the Frontier AI market?

More identifiable lead investors participated in Frontier AI funding during 2026, suggesting that financing activity is reaching a wider investor base.

We counted 11 distinct lead or co-lead names in Q4 2025, rising to 27 in Q1 2026 and 37 in Q2.

The count remained relatively high at 32 in Q3, despite the decline in transactions.

Several investors appeared repeatedly.

Menlo Ventures was identified more than once in Q4 2025, while D. E. Shaw Ventures and MGX appeared repeatedly in Q1 2026. Sequoia Capital, GV and Khosla Ventures were among the recurring names in Q2, followed by NVIDIA in Q3.

Large financing announcements also show how varied the investor groups have become.

Anthropic's February 2026 Series G involved investors such as GIC, Coatue, Founders Fund and MGX, while its May Series H included additional participation from major asset managers, growth investors and infrastructure-related partners.

There are limits to what the lead-investor data establish. We have not fully normalized every investor alias, and disclosure coverage varies across quarters. A firm missing from one quarter may still have participated without being publicly identified as a lead.

Nevertheless, the increase in distinct disclosed names suggests that investor participation broadened during the stronger financing quarters.

Why are corporate investors so interested in Frontier AI labs?

Corporate investors are deeply involved in Frontier AI because owning, financing or partnering with model developers can also support their computing, infrastructure and commercial businesses.

Strategic-backed transactions represented 25% of deals in Q2 2025. That share increased for three consecutive quarters, reaching 91.7% in Q1 2026.

Participation subsequently declined to 71.9% in Q2 and 65.2% in Q3.

The latest figures still show considerable corporate involvement, with 15 of 23 Q3 financings including a database-classified strategic investor.

Several company announcements help explain the interest.

ASML led Mistral's September 2025 financing, connecting frontier AI research with semiconductor manufacturing and industrial technology.

Anthropic's May 2026 Series H brought together major financial investors and infrastructure partners, including Micron, Samsung and SK hynix. The company also described substantial commitments involving cloud and computing capacity.

In September 2026, Samsung Electronics led Mistral's €3 billion Series D, adding another example of an industrial technology group backing frontier-model development.

Corporate participation was also changing across company stages. During Q2 2026, strategic investors appeared in ten early-stage and ten later-stage financings. By Q3, nine strategic-backed transactions were later-stage, compared with five early-stage deals.

Not every strategic relationship represents an equivalent cash equity investment. Compute credits, supplier relationships and commercial partnerships can have very different financial structures.

Still, strategic capital and corporate involvement have become regular features of the market, remaining present in most recorded financings despite retreating from the Q1 peak.

What is the biggest problem facing smaller Frontier AI labs today?

The biggest funding challenge for smaller Frontier AI labs is their limited access to the large amounts of capital needed to compete in model development, infrastructure and commercial deployment.

Leading developers are securing multibillion-dollar commitments, while many other model companies rely on transactions worth tens or hundreds of millions.

That gap affects training capacity, inference costs, product performance, customer distribution and the ability to keep developing successive generations of models.

Recent corporate activity makes the divide especially visible. OpenAI has emphasized the importance of durable computing access, while Anthropic has expanded its infrastructure agreements with several major technology providers.

There are alternative strategies. Mistral's enterprise-specific model development and work on sovereign AI infrastructure show how companies can build differentiated positions without necessarily following the same commercial approach as the largest American platforms.

The funding challenge now is whether a growing population of smaller laboratories can build sufficiently valuable products to secure successive rounds.

Q2 2026 demonstrated that investors were willing to back many more companies. The decline in first financings during Q3 raises the possibility that access to initial capital could narrow again.

For smaller laboratories, the ability to demonstrate commercial demand and secure further capital may increasingly determine which companies can remain competitive.

What is the Frontier AI Labs funding trend actually telling us?

Frontier AI Labs funding is becoming broader in the number of companies receiving investment, while the largest laboratories continue to control an extraordinary share of capital.

The evolution from Q2 2025 through Q3 2026 reveals several important changes.

Funding concentration has become a persistent feature of the market. The largest transactions dominated every quarter we examined. OpenAI and Anthropic demonstrate how financing requirements for leading platforms can reach a scale far beyond ordinary venture funding.

One major commitment can change quarterly funding by tens of billions of dollars without materially increasing the number of companies obtaining financing.

Q2 2026 was the clearest expansion in startup funding. The jump in deal volume, funded companies, early-stage activity and first financings was substantial.

The following quarter reversed part of that growth, so we cannot yet describe the expansion as a sustained trend. Nevertheless, the number of companies financed remained much higher than during most of 2025.

Smaller financing rounds are showing the most interesting resilience. Capital from rounds worth $50 million or less increased in both Q2 and Q3 2026.

That improvement occurred even as the market experienced a major decline in headline investment. If smaller-round activity continues, it could support a more diverse generation of frontier AI developers.

Language models still dominate capital, while multimodal AI is gaining breadth. Language and reasoning developers remain the principal beneficiaries of the largest financing commitments.

Multimodal AI has shown a more consistent increase in transaction activity, while agentic platforms remain volatile and video-generation funding is highly sensitive to exceptional deals.

Early-stage funding is recovering, but later-stage companies retain the financial advantage. The early-stage share of transactions increased considerably in 2026.

However, later-stage companies continue receiving almost all investment dollars. The market is funding a wider range of companies without distributing its financial resources proportionately.

Investor participation is broader, and strategic capital remains important. More distinct lead investors appeared during the high-volume financing quarters.

Strategic-investor participation stopped accelerating after Q1 2026 but remained present in most transactions. Computing infrastructure, enterprise deployment and commercial partnerships help explain why frontier AI attracts investors beyond traditional venture capital.

The next challenge is keeping the new company cohort funded. The Q2 2026 expansion brought a larger number of companies into the financed market. Q3 showed that first-time fundraising can contract quickly even when follow-on activity remains relatively stable.

The most important change is the emergence of a wider financing market beneath a small number of exceptionally well-capitalized platforms.

For now, we see a cooling market in terms of headline capital, but one with more active companies and a somewhat stronger smaller-round segment than in 2025. Whether that improvement lasts will depend on continued access to capital and the commercial progress of the newly financed laboratories.