Who are the top investors in frontier AI labs?







Every funding round, every investor, and what the labs promised in return, in one document. Have a look ↓
Summary
NVIDIA is the most active frontier AI lab investor, joining 43 rounds in our coverage. Andreessen Horowitz follows with 17 rounds and AMD Ventures with 16. The chipmaker appears in more model-lab financings than any VC firm.
Joining rounds and leading them are different jobs. NVIDIA is often part of the syndicate, while Andreessen Horowitz is named lead much more often. That difference changes who a founder should call first.
Traditional funds are still firmly involved. Sequoia Capital and Lightspeed Venture Partners back original-model labs, even as corporate investors join more and more of their rounds.
NVIDIA is far from the only corporate backer. AMD Ventures, Alibaba, Tencent and Salesforce Ventures support competing model families and can gain commercial relationships alongside an equity stake.
The leading investors have stayed active. NVIDIA joined 23 rounds this year versus 13 last year, and 19 of the 20 investors in our most active group have taken part in a round this year.
They show up at different stages. Andreessen Horowitz backs the most seed rounds; NVIDIA leads participation at Series A, B, C and D+; SoftBank Vision Fund is most active in Growth Equity.
A huge round can have several investors with different check sizes. The announced round total cannot tell us how much any participant contributed.
Look at the syndicates and some names keep appearing together. AMD Ventures and NVIDIA often share deals, as do Alibaba and Tencent. These are recurring relationships across multiple labs.
Smaller funds still win meaningful roles. Radical Ventures, Khosla Ventures, Index Ventures and Lux Capital often help lead research bets rather than joining every big round.
We find a mixed investment market: NVIDIA joins the widest range of deals, VCs often lead, and strategic backers come back repeatedly. Participation is visible; an individual investor's check size usually is not.
Written by the New Market Pitch teamWe track new markets so founders and investors can move faster. About us
70+pages
This is just a preview page. Don’t worry if a number looks old: the report gets refreshed, and you’ll always get the latest version.
Which VCs invest the most in frontier AI labs right now?
NVIDIA is the most active investor in frontier AI labs, with 43 rounds joined, ahead of Andreessen Horowitz at 17 and AMD Ventures at 16.
Among traditional VC firms, Andreessen Horowitz is the most active. NVIDIA is a chipmaker whose investments can also support its computing ecosystem.
NVIDIA backs OpenAI, Anthropic, xAI, Safe Superintelligence and Reflection AI. Andreessen Horowitz backs OpenAI, Safe Superintelligence and Thinking Machines Lab. NVIDIA's NVentures program offers companies technical support alongside financing, while Andreessen Horowitz has argued publicly for funding research teams such as Mirendil. AMD Ventures also backs rival model developers, so the investor market extends beyond the familiar VC names.
The table below ranks the 20 most active frontier AI labs investors by the number of equity rounds they joined, with the rounds they led, their main stages, the startups they backed and the biggest round they joined.
If you want to go more into detail and get more data to understand what investors are betting on now, please check our frontier AI labs report.
| # | Investor | Rounds | Led | Main stages | Frontier AI labs startups backed | Biggest round joined | Latest round |
|---|---|---|---|---|---|---|---|
| 1 | NVIDIA ▲Corporate | 43 | 5Mostly follows (led 14% of the 37 rounds with a known lead) |
Series B 10Series A 8 | OpenAI, Anthropic, xAI, Safe Superintelligence, Reflection AI + 29 more | $122BOpenAI · 2026 | Oct 2026 |
| 2 | Andreessen Horowitz ▲VC | 17 | 14Named as lead in every round found (lead known for 14 of 17 rounds) |
Seed 7Series A 5 | OpenAI, Safe Superintelligence, Thinking Machines Lab, ElevenLabs, World Labs + 10 more | $122BOpenAI · 2026 | Nov 2026 |
| 3 | AMD Ventures ▲Corporate | 16 | 2Mostly follows (led 22% of the 9 rounds with a known lead) |
Series B 4Series D+ 3 | xAI, Cohere, Thinking Machines Lab, World Labs, Hark + 8 more | $6.0BxAI · 2024 | Aug 2026 |
| 4 | Sequoia Capital ▲VC | 14 | 9Usually leads (82% of the 11 rounds with a known lead) |
Series A 5Seed 4 | OpenAI, Anthropic, Safe Superintelligence, Reflection AI, ElevenLabs + 7 more | $65.0BAnthropic · 2026 | Nov 2026 |
| 5 | Alibaba ▲Corporate | 12 | 6Usually leads (75% of the 8 rounds with a known lead) |
Series B 5Series C 3 | Moonshot AI, ShengShu Technology, Baichuan AI, PixVerse, MiniMax + 1 more | $3.5BMoonshot AI · 2026 | Jul 2026 |
| 6 | Lightspeed Venture Partners ▼VC | 10 | 8Usually leads (89% of the 9 rounds with a known lead) |
Series A 4Seed 2 | Anthropic, Reflection AI, Skild AI, Suno, Ineffable Intelligence + 3 more | $13.0BAnthropic · 2025 | Oct 2026 |
| 7 | Tencent ▲Corporate | 10 | 1Mostly follows (led 25% of the 4 rounds with a known lead) |
Series B 3Series A 1 | DeepSeek, Moonshot AI, StepFun, Baichuan AI, Zhipu AI | $15.9BDeepSeek · 2026 | Oct 2026 |
| 8 | Salesforce Ventures ▼Corporate | 8 | 1Mostly follows (led 14% of the 7 rounds with a known lead) |
Series D+ 3Series B 2 | Anthropic, Cohere, Mistral AI, Hark, Black Forest Labs + 2 more | $3.5BAnthropic · 2025 | Jun 2026 |
| 9 | Khosla Ventures ▲VC | 7 | 7Named as lead in every round found | Series A 5Seed 2 | General Intuition, Sakana AI, Scaled Cognition, Moonvalley, Recraft + 1 more | $320MGeneral Intuition · 2026 | Jun 2026 |
| 10 | SoftBank Vision Fund ▲VC | 7 | 5Usually leads (83% of the 6 rounds with a known lead) |
Growth 4Series A 1 | OpenAI, Skild AI | $122BOpenAI · 2026 | Nov 2026 |
| 11 | GV ▲VC | 7 | 3Mostly follows (led 43% of its rounds) |
Seed 3Series A 2 | Flapping Airplanes, Recursive, Flourish, Odyssey | $1.0BFlapping Airplanes · 2026 | Oct 2026 |
| 12 | Amazon ▲Corporate | 7 | 0Never led (lead known for 5 of 7 rounds) |
Series B 3Seed 1 | OpenAI, Luma AI, Upstage, Odyssey, Twelve Labs + 1 more | $122BOpenAI · 2026 | Jul 2026 |
| 13 | Radical Ventures ▲VC | 6 | 5Named as lead in every round found (lead known for 5 of 6 rounds) |
Series B 2Seed 1 | Cohere, World Labs, Decart, Generalist AI, Writer + 1 more | $500MCohere · 2025 | Jul 2026 |
| 14 | Lux CapitalVC | 5 | 5Named as lead in every round found | Series A 3Seed 2 | Physical Intelligence, Flapping Airplanes, Sakana AI, Phonic | $1.0BFlapping Airplanes · 2026 | Oct 2026 |
| 15 | New Enterprise AssociatesVC | 5 | 5Named as lead in every round found | Series A 3Series B 1 | World Labs, Sakana AI, Twelve Labs, Genmo | $200MSakana AI · 2024 | Jul 2026 |
| 16 | Index Ventures ▲VC | 5 | 4Usually leads (80% of its rounds) |
Seed 2Series A 2 | Flapping Airplanes, Simile, Cartesia | $1.0BFlapping Airplanes · 2026 | Oct 2026 |
| 17 | Kleiner Perkins ▲VC | 4 | 4Named as lead in every round found | Series A 3Seed 1 | Mind Robotics, Cartesia, Mirendil | $1.0BMirendil · 2026 | Nov 2026 |
| 18 | Menlo Ventures ▼VC | 4 | 4Named as lead in every round found | Seed 2Series C 1 | Anthropic, Suno, Elorian, Inception Labs | $750MAnthropic · 2024 | Apr 2026 |
| 19 | ICONIQ Growth ▼Investment firm | 4 | 4Named as lead in every round found | Series C 2Series D+ 2 | Anthropic, ElevenLabs, Writer | $30.0BAnthropic · 2026 | Feb 2026 |
| 20 | General CatalystVC | 4 | 4Named as lead in every round found | Seed 2Series B 2 | Mistral AI, General Intuition, Moonvalley | $660MMistral AI · 2024 | Oct 2025 |
No investors match this search.
Data from January 2024 through 13 November 2026, which is enough to show who is active today and who has been for years. Rounds: equity rounds in which the investor is named. Led: rounds it led or co-led. ▲ more rounds in 2026 so far than its 2025 pace, ▼ fewer.
70+pages
Just a preview. If a number here looks old, no worries: your report comes with the latest data.
Is NVIDIA backing frontier AI labs like a normal VC?
NVIDIA does not back frontier AI labs like a conventional VC: it also wants more demand for its computing platform.
NVIDIA participated in financings involving 34 startups, but was named lead in just 5 rounds. Its lead rate is 14% among rounds with a known lead. NVIDIA usually joins other investors instead of leading the round. The company appears in financings involving Anthropic, xAI, Safe Superintelligence and other laboratories pursuing different model strategies.
NVIDIA's September 2025 agreement with OpenAI focused on computing systems for future models. NVentures also describes technical help and introductions as benefits for portfolio companies. A frontier lab may benefit from access to hardware expertise, software libraries and deployment partners even when another investor sets the deal's terms. These investments do not prove equipment exclusivity or reveal NVIDIA's individual checks. They give the company exposure to a wider group of future AI-computing customers.
Does Andreessen Horowitz lead frontier AI rounds or just join them?
Andreessen Horowitz is the busiest named lead investor in frontier AI labs, even though NVIDIA joins more rounds overall.
We recorded 14 rounds in which Andreessen Horowitz was named lead. Its status was published for those rounds, and the firm was named lead in each one. Its portfolio includes OpenAI, Safe Superintelligence, Thinking Machines Lab, ElevenLabs and World Labs, a range that reaches from general reasoning research to generative voice and world models.
Those are deliberate research bets across different kinds of models. In its June 2026 Mirendil investment announcement, the firm argued that research tooling could let more engineers and scientists participate directly in frontier-model development. Its 2025 discussion of Thinking Machines Lab similarly emphasized the founding team's research experience. The firm is backing researchers and their ability to train new models. Andreessen Horowitz's lead record says more about who starts deals than a list of occasional co-investors would.
70+pages
This is a preview page. Spotted an outdated number? That’s normal here. The report you get is always the freshest version.
Is Sequoia Capital a serious frontier AI lab investor now?
Sequoia Capital is now a major VC backer of frontier AI labs and often helps lead their rounds.
Sequoia Capital joined 14 rounds across 12 startups and was named lead in 9. The lead rate is 82% among its 11 rounds with a published lead. The portfolio includes OpenAI, Anthropic, Safe Superintelligence, Reflection AI and ElevenLabs. That mix gives Sequoia exposure to both broad model platforms and narrower technical approaches.
The firm's April 2025 feature on Reflection AI set out the case for independently trained open frontier models. Anthropic's May 2026 financing announcement separately identified Sequoia among its lead investors, linking the firm's activity to labs with significant enterprise adoption as well as research ambition. Sequoia backs both newer open-model teams and commercial model platforms. It trails NVIDIA in participation and Andreessen Horowitz in named lead rounds, but its place in this market is established.
Which frontier AI labs investors are still investing this year?
NVIDIA is still the busiest frontier AI lab investor this year, joining 23 rounds compared with 13 last year.
Of the 20 investors in our most active group, 19 have joined a round this year. We counted 103 rounds with named investors this year, versus 58 last year.
AMD Ventures, Sequoia Capital, GV and Tencent are active too. Some back a wide range of labs, while others prefer fewer deals where they can lead. New research approaches, including Mirendil's work on automated model development, still attract VC attention alongside the major platforms.
The table below shows, year by year, how many rounds 12 of the most active frontier AI labs investors joined, so it is easy to see who is still investing and who has slowed down.
If you want to go more into detail and see which investors are still writing checks and which have slowed down, please check our frontier AI labs report.
| Investor | 2024 | 2025 | 2026* | Total |
|---|---|---|---|---|
| NVIDIA | 7 | 13 | 23 | 43 |
| Sequoia Capital | 3 | 3 | 8 | 14 |
| AMD Ventures | 4 | 5 | 7 | 16 |
| Andreessen Horowitz | 8 | 3 | 6 | 17 |
| Alibaba | 5 | 2 | 5 | 12 |
| Tencent | 4 | 1 | 5 | 10 |
| GV | 2 | 5 | 7 | |
| Index Ventures | 1 | 4 | 5 | |
| Lightspeed Venture Partners | 3 | 4 | 3 | 10 |
| SoftBank Vision Fund | 2 | 2 | 3 | 7 |
| Amazon | 2 | 2 | 3 | 7 |
| Radical Ventures | 2 | 1 | 3 | 6 |
* 2026: January to 13 November. Darker cells mean more rounds that year. Data from January 2024 through 13 November 2026.
Is AMD Ventures catching up with NVIDIA in frontier AI investing?
AMD Ventures backs major frontier AI labs, but NVIDIA still participates in far more rounds.
AMD Ventures joined rounds involving 13 startups and was named lead in 2. Its portfolio includes xAI, Cohere, Thinking Machines Lab, World Labs and Hark. NVIDIA joined far more rounds, but the investment logic overlaps: a semiconductor supplier benefits when additional model developers have viable choices for building and serving AI.
AMD has also pursued technical partnerships that are distinct from minority equity positions. Its September 2025 collaboration announcement with Cohere emphasized infrastructure for secure enterprise and sovereign deployments; AMD's ROCm ecosystem material discusses running Cohere models on AMD accelerators. Those relationships can improve a model developer's hardware options without implying that a venture check is a purchase agreement. NVIDIA has a wider investment network. AMD Ventures can still help model developers adopt AMD hardware through targeted investments and partnerships.
Do Alibaba and Tencent back the same Chinese AI labs?
Alibaba and Tencent do back some of the same Chinese frontier AI labs, even while competing in other businesses.
Alibaba joined 12 rounds involving 6 startups and was named lead in 6. Tencent joined 10 rounds involving 5 startups and was named lead in 1. Alibaba's tracked companies include Moonshot AI, ShengShu Technology, Baichuan AI, PixVerse and MiniMax; Tencent's include DeepSeek, Moonshot AI, StepFun, Baichuan AI and Zhipu AI. The different lead patterns suggest Alibaba more frequently anchors a financing in the recorded rounds, while Tencent commonly appears alongside other backers.
Moonshot AI and Baichuan AI appear in both portfolios. Alibaba and Tencent also joined 6 rounds together involving 3 startups, including Zhipu AI. This repeated overlap shows that China's major tech platforms help finance rival model developers even while competing for cloud and consumer customers. We cannot infer either company's individual check size from the round total.
70+pages
Preview only. Some numbers may have changed since we made this page, but your report will be up to date.
Do Amazon and Salesforce Ventures back rival AI labs?
Amazon and Salesforce Ventures back rival frontier AI model labs instead of relying on one provider.
Amazon joined 7 rounds across 6 startups and was not named lead in any of the 5 rounds for which a lead was published. Its tracked relationships include OpenAI, Luma AI, Upstage, Odyssey and Twelve Labs. Salesforce Ventures joined 8 rounds across 7 startups and was named lead in 1, with Anthropic, Cohere, Mistral AI, Hark and Black Forest Labs among its portfolio companies.
The operating motives differ. Amazon sells cloud infrastructure and distributes model access, while Salesforce sells enterprise software that can use several model families. Amazon can benefit when more models run on cloud infrastructure, and Salesforce can offer customers more model choices. A minority investment does not prove an exclusive customer contract.
Why does SoftBank Vision Fund focus on so few AI labs?
SoftBank Vision Fund makes concentrated bets on frontier AI labs, especially companies that need huge growth rounds.
SoftBank Vision Fund joined 7 rounds involving 2 startups, Skild AI and OpenAI, and was named lead in 5 of its 6 rounds with a known lead. It joined 4 Growth Equity rounds, the highest participation we measured at that stage. It is a very narrow portfolio by company count.
SoftBank's commercial relationship with OpenAI reinforces the distinction. In February 2025 the companies announced a joint enterprise AI initiative in Japan; they subsequently launched SB OAI Japan to market customized AI products. SoftBank therefore participates in a funding market where large investments and distribution arrangements can reinforce one another. Skild AI gives it exposure to general-purpose intelligence for robots, a different frontier from language-only assistants. That focus leaves SoftBank more exposed to the success of a few model developers while linking its financing to operating partnerships.
Which frontier AI labs investors back seed rounds, and which back growth rounds?
Andreessen Horowitz backs the most frontier AI lab seed rounds, with 7, while NVIDIA leads participation at Series D and later with 8.
Seed has 61 of the 209 rounds with a known stage. SoftBank Vision Fund stands out in Growth Equity with 4 participations.
Where investors enter makes a big difference. Andreessen Horowitz can lead founding research bets such as Mirendil; NVIDIA can participate in model builders after the first proof of capability; SoftBank Vision Fund can focus on businesses with large infrastructure and distribution requirements. At seed, investors judge the research team. In growth rounds, they also have to judge whether the model can reach customers at scale.
The table below shows, for 12 of the most active frontier AI labs investors, how many rounds they joined at each stage, from seed to growth.
If you want to go more into detail and see where each investor places its bets, from seed to growth, please check our frontier AI labs report.
| Investor | SeedSeed | Series AA | Series BB | Series CC | Series D+D+ | GrowthGrowth |
|---|---|---|---|---|---|---|
| NVIDIA | 6 | 8 | 10 | 4 | 8 | 3 |
| Andreessen Horowitz | 7 | 5 | 2 | 1 | · | 2 |
| AMD Ventures | 1 | 2 | 4 | 2 | 3 | 1 |
| Sequoia Capital | 4 | 5 | 2 | · | 2 | 1 |
| Alibaba | · | 1 | 5 | 3 | · | · |
| Lightspeed Venture Partners | 2 | 4 | 1 | · | 2 | 1 |
| Tencent | · | 1 | 3 | 1 | · | · |
| Salesforce Ventures | 1 | 1 | 2 | 1 | 3 | · |
| Khosla Ventures | 2 | 5 | · | · | · | · |
| SoftBank Vision Fund | · | 1 | 1 | 1 | · | 4 |
| GV | 3 | 2 | 1 | · | · | 1 |
| Amazon | 1 | 1 | 3 | · | · | 1 |
Most active at each stage, among all frontier AI labs investors: Seed, Andreessen Horowitz · Series A, NVIDIA · Series B, NVIDIA · Series C, NVIDIA · Series D+, NVIDIA · Growth, SoftBank Vision Fund. Bigger dots mean more rounds. Data from January 2024 through 13 November 2026.
Will traditional VCs still fund AI labs before they have customers?
Traditional VCs still fund frontier AI labs before customer demand is well established.
Andreessen Horowitz joined 7 seed rounds, Khosla Ventures joined 2 and Lux Capital joined 2. These are stage participations, while a lead position depends on the investors disclosed for the specific financing. Andreessen Horowitz was named lead in every round for which its lead status is published; Khosla Ventures was named lead in all 7 of its recorded rounds with a published lead.
Look at the teams receiving those early bets. Mirendil promises AI-assisted research; Thinking Machines Lab was backed around the experience of its research team; Physical Intelligence's published pi-zero work centers on general-purpose robot manipulation. None can be evaluated by asking only how many seats of conventional SaaS software they have sold. Backers need credible technical milestones, access to researchers and a feasible sequence of training experiments. For an early investor, backing the right researchers before large funds show up can matter more than joining the next giant round.
70+pages
This is just a preview page. Don’t worry if a number looks old: the report gets refreshed, and you’ll always get the latest version.
Can VCs really lead a billion-dollar Series A for an AI lab?
Venture firms really can lead a billion-dollar Series A for a frontier AI lab built around a major research bet.
Flapping Airplanes completed a $1.0B Series A in October 2026 led by Index Ventures and Lux Capital. Mirendil completed a $1.0B Series A in November 2026 led by Kleiner Perkins and Clearlake Capital. The two companies pursue distinct ideas: Flapping Airplanes emphasizes more data-efficient model training, while Mirendil works on models and systems that automate parts of AI research. These are full round sizes, not the amounts personally contributed by any named backer.
Index Ventures' October 2026 founder interview described Flapping Airplanes' work on training efficiency and its own role leading the financing. Mirendil's company announcement outlined early research outputs and the intent to expand that system. These are far bigger than ordinary software Series A rounds. Labs need an exceptional team and a credible technical approach to raise at that scale.
Is GV becoming a bigger frontier AI lab investor?
GV is backing more frontier AI labs this year, even though it trails the most active investors overall.
GV joined 7 rounds across 4 startups and was named lead in 3. Its portfolio includes Flapping Airplanes, Recursive, Flourish and Odyssey. It joined 5 rounds this year, compared with no round last year in the tracked set. GV has clearly become more active lately.
The portfolio is technically varied. Flapping Airplanes develops data-efficient foundational approaches; Odyssey and other world-model developers explore representations and generation beyond text; Recursive pursues advanced model research. Such choices fit a view that future frontier capability may come from model architecture and training methods, not only additional compute deployed behind established interfaces. GV is looking beyond the existing biggest labs and backing different approaches to model research. Its individual checks remain undisclosed.
Can specialist funds like Radical Ventures lead big AI rounds?
Specialist funds such as Radical Ventures lead frontier AI lab rounds without needing NVIDIA-sized portfolios.
Radical Ventures joined 6 rounds involving 6 startups and was named lead in 5, with a 100% lead rate among the 5 rounds with a published lead. Its companies include Cohere, World Labs, Decart, Generalist AI and Writer. The portfolio spans enterprise language models, world models and embodied-model research, reflecting a thesis about reusable intelligence across different tasks.
Radical's September 2026 announcement of its Breakouts Fund emphasized its prior work with Cohere and World Labs and investments in Generalist and Prime Intellect. That history helps Radical evaluate new model teams before standard sales metrics exist. Fewer rounds do not prevent a specialized fund from leading difficult research bets.
Which frontier AI labs investors most often invest together?
AMD Ventures and NVIDIA are the most frequent co-investors in frontier AI labs, joining 7 rounds across 6 startups.
Alibaba and Tencent joined 6 rounds together across 3 startups, while NVIDIA and Salesforce Ventures joined 5 rounds across 4 startups.
The pairings tell us who keeps backing the same teams. AMD Ventures and NVIDIA both benefit from a larger model-training ecosystem, even while competing as chip suppliers. Alibaba and Tencent overlap in Chinese laboratories including Moonshot AI and Baichuan AI, and Salesforce Ventures shares exposure with NVIDIA to enterprise-oriented model developers such as Cohere. Sharing a deal does not mean they committed equal checks or formed a formal alliance.
The table below lists the pairs of frontier AI labs investors named together in at least two equity rounds, with the startups they backed together.
If you want to go more into detail and see which investor networks back the same startups, please check our frontier AI labs report.
| Investor pair | Rounds together | Startups backed together |
|---|---|---|
| AMD Ventures + NVIDIAxAI, Cohere, Thinking Machines Lab | 7 | xAI, Cohere, Thinking Machines Lab |
| Alibaba + TencentMoonshot AI, Baichuan AI, Zhipu AI | 6 | Moonshot AI, Baichuan AI, Zhipu AI |
| NVIDIA + Salesforce VenturesCohere, Mistral AI, Hark | 5 | Cohere, Mistral AI, Hark |
| Andreessen Horowitz + NVIDIAOpenAI, Thinking Machines Lab, Luma AI | 4 | OpenAI, Thinking Machines Lab, Luma AI |
| Lightspeed Venture Partners + NVIDIAReflection AI, Ineffable Intelligence, doubleAI | 4 | Reflection AI, Ineffable Intelligence, doubleAI |
| NVIDIA + Radical VenturesCohere, Decart, Generalist AI | 4 | Cohere, Decart, Generalist AI |
| Greycroft + NVIDIAHark, Recursive, AMI Labs | 3 | Hark, Recursive, AMI Labs |
| Andreessen Horowitz + Sequoia CapitalOpenAI, Safe Superintelligence, TypeSafe AI | 3 | OpenAI, Safe Superintelligence, TypeSafe AI |
Pairs of investors named in the same equity round. Data from January 2024 through 13 November 2026.
70+pages
Just a preview. If a number here looks old, no worries: your report comes with the latest data.
Do Andreessen Horowitz and Sequoia Capital invest together in AI labs?
Andreessen Horowitz and Sequoia Capital sometimes co-invest in frontier AI labs despite competing for access to leading research teams.
They joined 3 rounds together across 3 startups, including Safe Superintelligence, OpenAI and TypeSafe AI. Separately, the firms accumulated substantial participation records, with different patterns of lead responsibility. An overlap in a syndicate does not mean the funds have identical strategies or that each contributed an equal check.
The rationale becomes clearer in enormous research financings. Frontier laboratories must secure long research runways and potentially very large compute commitments; a financing syndicate may therefore include several investors whose individual risk appetite or strategic exposure differs. Andreessen Horowitz's public thesis for Mirendil emphasizes research tooling and a broader model-building ecosystem, while Sequoia has highlighted Reflection AI's independent open-frontier approach. Both see opportunities in proprietary scientific capability, but they need not agree about the ultimate winning distribution model. The co-investment record suggests shared conviction in selected teams, not a merger of investment strategies.
Are European VCs being left out of frontier AI investing?
European VCs still help finance frontier AI labs, though the busiest investors are based elsewhere.
Index Ventures joined 5 rounds across 3 startups and was named lead in 4; its portfolio includes Flapping Airplanes, Simile and Cartesia. Index Ventures helped lead Flapping Airplanes' $1.0B Series A in October 2026, showing that European funds can win major US research deals. The company's founders described a strategy focused on making model training more efficient, which directly targets a central constraint in frontier AI.
European model builders can also attract global strategic financing. Mistral's September 2026 Series D+ involved Samsung and other lead investors with industrial and infrastructure interests. Investor nationality and startup headquarters are therefore not interchangeable: a European VC may back a US laboratory, and a French model builder may receive financing from outside Europe. Index Ventures shows that a European VC can lead a major research deal even without appearing near the top by volume.
Are big AI investors crowding smaller VCs out of seed rounds?
Smaller VCs are not yet being pushed out of frontier AI seed rounds, even as giant growth checks attract larger investors.
Andreessen Horowitz is strongly represented at seed, while NVIDIA's busiest named stages were Series B at 10 rounds and Series D+ at 8. Khosla Ventures joined 7 rounds in total and was named lead in every one with a published lead; Lux Capital joined 5 and was named lead in all 5. Specialist firms can still lead young research teams before the bigger growth investors participate.
Technical specialists can help a founding team assess whether a new training strategy has genuine leverage. Khosla Ventures' portfolio includes Sakana AI, General Intuition and Recraft; Lux Capital's includes Physical Intelligence, Sakana AI and Flapping Airplanes. These are not identical products, and lead participation alone does not predict returns. A young AI lab still needs investors willing to back an unproven research direction.
Is leading an AI lab round better than just joining it?
Leading a frontier AI lab round shows more responsibility for the financing, while joining can bring useful capital and industry relationships.
Andreessen Horowitz frequently appears as a named lead. Sequoia Capital was named lead on 9 rounds, and Lightspeed Venture Partners on 8 among 10 participations. NVIDIA joined many more financings but was less frequently named lead. The busiest participant is not necessarily the most active lead.
There are also disclosure limits. Press announcements tend to identify lead investors even when not every smaller participant is named, so an apparent high lead rate can partly reflect the reporting pattern. We use published names consistently and refrain from assigning undisclosed checks to investors. A founder seeking a lead VC should make a different shortlist from one seeking a strategic backer for a later round.
The table below shows selected traditional VC investors ordered by rounds joined, distinguishing participations from named lead rounds.
| Investor | Joined | Led | Startups |
|---|---|---|---|
| Andreessen Horowitz | 17 | 14 | 15 |
| Sequoia Capital | 14 | 9 | 12 |
| Lightspeed Venture Partners | 10 | 8 | 8 |
| Khosla Ventures | 7 | 7 | 6 |
| SoftBank Vision Fund | 7 | 5 | 2 |
| GV | 7 | 3 | 4 |
| Radical Ventures | 6 | 5 | 6 |
70+pages
This is a preview page. Spotted an outdated number? That’s normal here. The report you get is always the freshest version.
Can smaller funds still win access to elite frontier AI teams?
Smaller funds can still lead important frontier AI lab financings without appearing at the top of investor rankings.
Kleiner Perkins joined 4 rounds, all with a published lead position, and was among the investors leading Mirendil's $1.0B Series A in November 2026. Index Ventures joined 5 rounds and was named lead in 4, including Flapping Airplanes' $1.0B Series A in October 2026. New Enterprise Associates joined 5 rounds and was named lead in all 5, backing companies including World Labs and Sakana AI.
These examples show why an investor's relationship with a founder may matter more than volume in a technically narrow category. Kleiner Perkins' Mirendil relationship built on an early financing and a model-research thesis. Index Ventures' founder interview emphasized Flapping Airplanes' data-efficiency approach, rather than an existing mass-market software product. Such specialization can give a comparatively less frequent investor credibility with exceptional teams. The caveat is that fewer participations also mean a narrower sample of realized company outcomes; the ranking measures activity and leadership, not investment returns.
So, who are the biggest investors in frontier AI labs today?
NVIDIA is the busiest frontier AI lab investor by participation, while Andreessen Horowitz leads the most rounds.
We tracked 337 named investors across the market, with at least one identified investor on 230 of 244 qualifying equity rounds. NVIDIA's frequent participation shows how closely model funding is tied to computing infrastructure. Yet Andreessen Horowitz, Sequoia Capital, Lightspeed Venture Partners and specialist funds have distinct roles in originating and leading transactions. Corporate investors such as AMD Ventures, Alibaba and Tencent repeatedly join as well, sometimes alongside direct competitors.
Frontier AI labs rely on lead VCs, corporate investors and groups that repeatedly fund the same companies. A founder seeking a seed lead, a lab negotiating a growth financing and a model builder looking for infrastructure partnerships may rationally approach different investors, even when those firms appear in the same round. We rank NVIDIA first for participation, not for checks invested or rounds led.
The table below shows what we found.
| Question | Answer |
|---|---|
| Are traditional VCs still the busiest backers? | Not overall, because a strategic chipmaker dominates frontier AI participations. NVIDIA joined 43 rounds across the covered financing window. Traditional venture firms nevertheless lead many individual deals and negotiations. |
| Does Andreessen Horowitz mostly follow other investors? | No, Andreessen Horowitz frequently appears as a named financing lead. It led 14 rounds across our covered frontier lab market. Its approach emphasizes backing research teams before standardized metrics emerge. |
| Has Sequoia become a real frontier investor? | Yes, Sequoia Capital consistently participates in independent model-building company financings. It joined 14 rounds across 12 frontier AI startups. That breadth includes open-model research alongside major commercial platforms. |
| Is AMD too far behind NVIDIA strategically? | AMD Ventures is active, but its financing participation remains narrower. It joined 16 rounds involving 13 frontier AI startups. Hardware partnerships can extend its influence beyond minority equity ownership. |
| Are Chinese tech investors financing competing labs? | Yes, Chinese platforms repeatedly appear in overlapping frontier-model investment syndicates. Alibaba joined 12 rounds, while Tencent joined 10 rounds. Their shared portfolios coexist with competition in adjacent technology markets. |
| Is a seed investor necessarily a small investor? | No, lead investors can back ambitious original-model research from inception. Andreessen Horowitz joined 7 seed rounds during the covered window. Early research conviction matters more than conventional software-stage financing labels. |
| Are large growth rounds controlled by NVIDIA? | No, a different investor appears most often in Growth Equity. SoftBank Vision Fund participated in 4 Growth Equity rounds. That concentration reflects a strategy unlike broad early-stage portfolio building. |
| Can specialized venture funds still set terms? | Yes, specialist investors can lead meaningful deals without maximum participation. Radical Ventures was named lead in 5 recorded rounds. Technical judgment can matter as much as a large portfolio. |
| Do the biggest chip suppliers finance rivals? | Yes, semiconductor competitors repeatedly joined the same independent model laboratories. AMD Ventures and NVIDIA co-invested in 7 rounds. Shared exposure does not erase rivalry over training infrastructure sales. |
| Are global syndicates more than isolated partnerships? | Yes, recurring pairs show stable relationships across multiple frontier companies. Alibaba and Tencent joined 6 rounds together in our coverage. Repeat syndication can broaden access without requiring strategic exclusivity. |
| Do all active investors lead most deals? | No, participation and lead responsibility differ sharply across the market. NVIDIA was named lead in just 5 rounds it joined. Count both measures before choosing a financing partner for a lab. |
70+pages
Preview only. Some numbers may have changed since we made this page, but your report will be up to date.
Our methodology
Our data runs through November 13, 2026. We systematically tracked equity rounds in the frontier AI labs market from January 2024 through that date. We define the market as independent startups primarily training original, reusable language, reasoning, multimodal, image, video, speech, world or embodied-model families at or toward the capability frontier. Every company meets our startup definition: independent, founded in 2000 or later, designed for rapid growth, privately financed and unlisted or listed within the last 5 years; a company acquired within the last 3 years stays in. We include credible borderline laboratories instead of dropping them from the market.
We count the equity rounds each investor joined and the rounds where it was named lead, not the undisclosed amount of its individual check. IPOs and post-listing offerings are excluded. We normalize investor names, omit vague attributions such as "existing investors," and merge reports of the same financing. Our current-year activity column begins in March once sufficient months are available for a useful comparison.
Announcements name lead investors more reliably than smaller participants, so rounds that an investor joined without leading are undercounted and lead rates are an upper bound. We therefore describe an investor as named lead in the published rounds, not as having led every financing it may have participated in outside public reporting.
Rounds confirmed by a company announcement, a press article or a regulatory filing are counted in full. Where no article exists, a round is counted only when at least two recognised databases agree on its date (within one month) and amount (within 5%), and it is marked as low confidence; a round found in a single database stays in the company's history but is left out of totals. For companies that have raised more than $100M, every round counted is backed by an article.
We do not use commercial investor lists from Crunchbase, PitchBook or similar databases on their own. Those lists can mistakenly attach investors to rounds or mix equity with grants and debt. We identified at least one named investor on 230 of 244 qualifying equity rounds; those without a named investor stay in the deal count but cannot add a participation to any named investor, so published investor totals are conservative floors.
A round counts once in the funding tracker but once for each confirmed participant in an investor ranking, explaining why these totals are not interchangeable. A large syndicate also does not reveal individual check sizes. We rank participation by joined rounds, identify lead responsibilities separately and use company headquarters, investor identity and disclosed stage only for the comparison they actually support. A specialist that joined fewer rounds may still have led highly consequential research financings; this is an activity ranking, not a ranking of realized returns.
Key sources include NVIDIA's venture investment program and strategic rationale, NVIDIA's model and infrastructure investment portfolio, NVIDIA and OpenAI hardware and research infrastructure agreement, OpenAI strategic relationships with Amazon and NVIDIA, Andreessen Horowitz's research-platform investment thesis, Andreessen Horowitz's Thinking Machines Lab investment thesis, Sequoia Capital's Reflection AI investment case, Anthropic's May 2026 investor and enterprise-demand announcement, Cohere's financing and AMD, NVIDIA and Salesforce participation, AMD and Cohere technical partnership, SoftBank's enterprise partnership with OpenAI, Launch of the SoftBank and OpenAI Japan joint venture, Radical Ventures' model-lab strategy and fund announcement, Index Ventures and Flapping Airplanes' training-efficiency thesis, Mirendil's later financing and research progress, Mirendil's early financing and founding team, World Labs' investor list and research strategy, Mistral's financing and independent model strategy, and Samsung's industrial AI agreement with Mistral.
Written by the New Market Pitch teamWe track new markets so founders and investors can move faster. About us









