AI Startup Funding: Q3 Hits $159B as October Mega-Rounds Roll On

Most of the money went to a small number of AI companies

Quarterly counts from the major data trackers make 2026 look like the most heavily funded year venture capital has ever had. Daily deal flow tells a slightly different story. A few companies collect most of the dollars, while a much longer list of startups raise ordinary Series A and B rounds and have to justify their valuations through revenue.

Crunchbase puts global venture funding for the third quarter at $159 billion, spread across nearly 6,000 startups. That makes it the weakest quarter of 2026 so far, but it is still larger than any quarter since Q2 2022.1 AI-related companies took $102 billion of that, or 64% of all venture dollars.1 The year-to-date total through September reached $679 billion, the most ever raised in the first nine months of a year.1

The record that stands out most is the count of very large rounds. Twenty-seven companies raised $1 billion or more in Q3, compared with 16 in Q2 and 14 in Q1. Those deals made up about a third of all venture capital raised during the quarter.3 Eight companies raised at least $3 billion each. Databricks and Safe Superintelligence led with $5 billion apiece, and Crusoe, Moonshot AI, Mistral AI, Nscale, The Boring Co. and Kling AI each raised more than $3 billion.2

The trackers disagree on the totals

The major data providers do not agree on how big the quarter was, and the gap is large enough to affect any conclusions.

KPMG's Venture Pulse counts $170.9 billion across 8,341 deals in Q3, with $728.3 billion for the first nine months.5 Fundraise Insider, working from a narrower dataset, counts only $107.5 billion raised by 1,611 companies and 22 billion-dollar financings rather than 27.8 The share going to AI also varies a lot. Fundraise Insider puts AI at 41.5% of capital, down from 49.2% in Q2. Its explanation is that several of the quarter's biggest deals went to towers, tunnels, rockets and fiber, not to AI companies.8

The gap mostly comes from definitions. Some trackers count debt facilities as venture funding and others do not. They also differ on what counts as an "AI company" and how much of the deal-reporting long tail they capture. The direction is the same in every dataset, though. Q3 was down from Q2, far above Q3 2025, and still dominated by large checks. Crunchbase says the total fell 25% from Q2's $212 billion but rose 53% from $104 billion a year earlier.3

The most useful idea in the coverage is that AI's share of funding is falling even as its reach widens. Fundraise Insider finds AI companies made up at least half of all deals at every stage from pre-seed through Series C, and seed crossed 50% for the first time.8 Its conclusion is that AI is now the default across the whole venture ladder, not something investors pay up for only at later stages.8 That fits the record-count finding. More companies are being funded as AI businesses, while the dollar share depends on whichever non-AI megadeals close in a given quarter.

Why the totals fell from Q2

The decline from Q2 does not look like investors losing interest. Crunchbase's North America analysis says the drop came mainly from the absence of new megarounds for OpenAI and Anthropic, not from any broad weakening in the market.7 Earlier in the year those two companies raised $110 billion and $65 billion, which pushed the first-half figures sharply higher.7 Attention is now turning to public markets. Anthropic is reportedly considering a listing as early as November, and OpenAI is expected to debut in 2027 after a confidential filing in June.9

The heavy spending has also moved down the stack, from model labs to the companies that build, rent and power compute. Fundraise Insider counts 16 compute, data center and AI chip companies that together raised $19.5 billion, about 18% of disclosed capital, with Crusoe's $3.9 billion Series F the largest.8 KPMG describes the market as bifurcated. It counts 14 rounds above $1.5 billion globally in Q3, nine of them in the US, and finds that AI accounted for eight of the ten largest rounds in the Americas.5

The boom is not limited to the US. KPMG says European venture investment rose for an eighth straight quarter, to $28.4 billion, its highest level in four years, helped by AI and defense deals such as Helsing's $1.8 billion and Quantum Systems' $1.2 billion.5 In Asia, Crunchbase reports $38.5 billion in Q3 funding, about 55% of it to AI companies, led by Moonshot AI's $3.5 billion round at a reported $35 billion valuation.10

October's biggest rounds

Deals announced in the first half of October show the same pattern. The largest is TypeSafe AI, the company behind the Jev model, which raised about $870 million at a $7.5 billion valuation in a round led by Andreessen Horowitz.27 Even the deal's label is inconsistent. One tracker calls it a Series A, with Sequoia Capital and DCVC participating.22 Dealroom calls it a late-stage round and places it in the 97th percentile of all US late-stage deep tech deals ever recorded.27 When a company less than a month past its product launch raises a "Series A" bigger than most growth rounds, the old stage names stop telling you much.

Manus is the most politically significant deal of the month. Its parent company, Butterfly Effect, raised more than $500 million in its first round since Beijing forced Meta to unwind its roughly $2 billion acquisition.23 Boyu Capital and IDG Capital led, and existing backers Tencent, HSG and ZhenFund participated.25 The reporting differs on valuation. Bloomberg reported in September that the round would double Manus's valuation to $4 billion, but the company has not disclosed a figure and did not answer questions about it.2325 Until Manus confirms a number, $4 billion should be treated as the figure from earlier negotiations, not the price at closing.22 Either way, investors were not put off by an unusual regulatory intervention.23

Arena is the clearest case of a valuation backed by revenue. The company runs the LMArena leaderboard and began as a UC Berkeley research project. It raised a $200 million Series B at a $3.1 billion valuation, led by Lightspeed and Khosla Ventures.24 That is close to double the $1.7 billion post-money valuation from its $150 million Series A in January. Annualized revenue grew over the same period from $30 million to $100 million.24 On those figures, the valuation rose more slowly than revenue did.

Other notable rounds:

• Nous Research raised a $90 million Series B at a $1.5 billion valuation, led by Robot Ventures, with Nvidia, Samsung Electronics and Y Combinator taking part.29
• FieldAI is reportedly raising $700 million at a $10 billion valuation, roughly five times its valuation of just over a year ago. That would put it close to Physical Intelligence at about $11 billion and Skild AI at more than $14 billion.26
• Mecka AI raised a $60 million Series B led by Sequoia, four months after a previous raise. It says it is on track for $300 million in run-rate revenue by year-end.20
• Skeleton Technologies, which makes power systems for AI data centers, said a Siemens-led strategic round lifted its valuation above $1 billion. The amount was not disclosed, and the CEO described it only as "9-digit."21
• Stand, a property insurance company, raised a $70 million Series C at a $1 billion valuation, led by Saga Ventures.21

A typical Series B

Most startups raise rounds that look more like RapidCanvas's. The Austin-based company raised $29 million in a Series B led by Vertex Ventures SEA & India. Hitachi Ventures joined, and existing investors Accel, Peak XV Partners, Valley Capital Partners and Titanium Ventures also participated.12 Its total funding is now $52.5 million, following a $16 million Series A in December 2024.13 The company says its annual recurring revenue grew eightfold over six quarters. It has not disclosed revenue figures or a valuation.13

Even this deal was reported in two versions. Some outlets described Vertex and Hitachi as co-leads.1117 The company's own announcement, as reported by several outlets, names Vertex SEA & India as the lead and says Hitachi joined.1318 The company's version should be treated as correct. One write-up also cited a claim that RapidCanvas automates up to 75% of a data scientist's work, and noted that the figure does not appear in the primary reporting.11

The RapidCanvas model is to pair agent software with forward-deployed engineers who sell to companies with $100 million to $10 billion in annual revenue.16 It is aimed at the gap between AI pilots and production, which is where enterprise buyers are now spending their budgets.18 The risk is that Google, OpenAI, Microsoft and Meta keep bundling agent features into products enterprises already pay for.11

What it means for founders

The best-supported conclusion is that the venture market is not cooling. It is splitting into tiers. At the top, the companies raising billion-dollar rounds, and the AI infrastructure firms behind them, set the quarterly totals. Below them, startups in the $25 million to $100 million range still raise money, but investors increasingly want revenue numbers like Arena's, Mecka's or RapidCanvas's ARR multiple before they will agree to a higher valuation.

It matters that the ten largest deals took a smaller share of capital than earlier in the year. Fundraise Insider says the top ten took 25.2% of Q3 disclosed capital, down from 44.2% in Q2 and 51.1% in Q1.8 That suggests the extreme concentration of the first half is easing slightly. Founders should still be careful about reading too much into the totals. One analysis warns that plenty of capital in the market does not prove customer demand, and points out that 43% of failed startups cite having no market need.6 Watch Q4 for the Anthropic and OpenAI listings. If they happen, they will show whether these private valuations hold up in public markets.