More Money, Narrower Channel: Two US Labs Took 43% of Global Venture in H1
FutureX Research · AI Lab · 2026.06.20 · 14 pp · preview 3 pp · Revised 2026.11.02
Listen · Audio Summary
5-8 min · AI narration in English · abstract + all key findings
Bottom Line
Capital is pooling in two US labs and spreading from fundraising to exits; China runs on a different clock.
- Why now
- Anthropic reportedly plans to start its IPO roadshow on November 9, so the $1.8T to $2T valuation some prospective investors assign it will face public markets for the first time.
- What it means
- For investors, structure now matters more than totals; for founders, US funding outside OpenAI and Anthropic has tracked the same trend since late 2024, so the surge has not reached everyone else.
Abstract
Money for AI has grown while the channel has narrowed, and in the US concentration is moving from fundraising to exits. US venture invested a record $412.7B in H1 2026, 86% of it in AI. Global venture reached $510B, already past all of 2025, and OpenAI and Anthropic took 43%. In Q3, US venture fell about 40% and AI's share slipped to 65.9%, so concentration now rests on two names. Anthropic reportedly plans to start its IPO roadshow on November 9. In China, DeepSeek is reportedly raising at least RMB 80B, at a valuation about 5% of OpenAI's pre-money value. (Data as of 2026-10-11)
Key Points
- 01PitchBook/Crunchbase: US venture hit a record $412.7B in H1 2026, with 86% going to AI; OpenAI and Anthropic together took 43% of global venture funding (Reported)
- 02PitchBook-NVCA (reported October 8, 2026): US venture deal value fell about 40% in Q3 to $98.4B, and AI's share dropped from 82.7% year to date to 65.9% (Reported)
- 03Anthropic plans to start its IPO roadshow on November 9, with some prospective investors at $1.8T to $2T, per Bloomberg on October 1; a draft prospectus obtained by Reuters shows a $42B net loss for 2025 (Reported)
- 04OpenAI is in talks to raise at least $30B at about $1.4T, per Bloomberg on September 29, and Altman has ruled out a 2026 listing; an end-September run rate of about $50B puts that at about 28x (Reported)
- 05DeepSeek is raising at least RMB 80B (about $12B) at a valuation of at least about $75B, about 5% of OpenAI's pre-money value, per Bloomberg on October 6; Moonshot was reported the same day at about $50B (Reported)
I. In H1, 86% of US Venture Went to AI and Two Companies Took 43% of Global Venture
In H1 2026, venture capital poured into AI, and more than 40% of global venture went to just two companies.
| Metric | H1 2026 | Comparison |
|---|---|---|
| US venture total (PitchBook) | $412.7B | Nearly 30% above all of 2025 |
| Of which AI (PitchBook) | $355.9B, or 86% | 86 cents of every venture dollar |
| Global venture total (Crunchbase) | $510B, over 70% to AI | Above 2025's full-year $440B; a record for any half-year |
| OpenAI + Anthropic | $122B + $95.6B | 43% of global venture combined |
H1 was, in effect, a two-company market. The change in late July: concentration did not ease. It kept spilling down the chain from models to compute, energy and distribution. In China, new funding talks at DeepSeek and Moonshot pushed the "narrative premium" higher.
For primary-market investors, structure now matters more than totals. When one sector takes 86%, conventional portfolio diversification stops working. The sections that follow start from here.
II. United States: Anthropic Races Toward an IPO While OpenAI Leans Toward 2027
OpenAI raised $122B in H1 at an $852B post-money valuation. Amazon ($50B committed in February), Nvidia and SoftBank ($30B each) led the round.
Its IPO timing has shifted. Per multiple late-June reports, management leans toward listing in 2027. CFO Sarah Friar favors waiting, citing cash burn, compute commitments and the burden of public reporting; Altman reportedly refuses to list early at a price below a trillion dollars.
Anthropic is ahead. It filed a confidential S-1 on June 1, targeting a Nasdaq listing as early as October, with Goldman Sachs, JPMorgan and Morgan Stanley underwriting a raise of over $60B. Its Series H private valuation is $965B, secondary markets imply $1.05-1.15T, and annualized revenue was about $30B as of April. On July 27 it expanded its enterprise partnership with Cognizant.
For the first time, regulation shaped release timing. GPT-5.6 (the Sol/Terra/Luna lineup) rolled out in stages under the pre-release review created by the June 2 executive order and opened broadly only on July 9: in effect, Washington now decides when frontier models reach the public. On July 14 Nvidia imposed a compliance whitelist on Asian customers; reportedly more than half of existing customers failed it.
On hardware, Apple sued OpenAI over trade secrets on July 10 (the risk section breaks this down); OpenAI still shipped the $230 Codex Micro coding device on July 15.
III. China: From Valuation Leaps to Proof of Delivery, With Moonshot Priced at Over 100x ARR
In China, the story is moving from valuation leaps to proof of delivery.
DeepSeek closed its first external round in mid-June: about RMB 50B (about $7B) at roughly $52B post-money. Founder Liang Wenfeng put in RMB 20B, 40% of the round, as the largest single backer; Tencent (RMB 10B), CATL-linked investors (RMB 5B) and JD, NetEase and IDG (RMB 3B each) followed. On July 15, multiple outlets reported talks with new investors at roughly $71-74B (figures vary), with an IPO anchored to 2027 and possibly earlier. The harder evidence is product: DeepSeek V4 went live in stages from July 31 (V4-Flash) to August 13 (V4-Pro), a 1.6T-parameter MoE with 1M-token context across the lineup and first-of-its-kind peak/off-peak API pricing.
Moonshot AI opened its round at $31.5B pre-money and closed it in late July at $35B post-money, raising $3.5B. With June ARR around $300M, that is over 100x ARR, which is where the "narrative premium" debate becomes concrete. It reportedly planned a pre-IPO round at up to $50B and a possible Hong Kong listing this year. Its 2.8T-parameter Kimi K3, released July 16, topped Frontend Code Arena at 1,679 points, ahead of Claude Fable 5 (1,631) and GPT-5.6 Sol (1,618); the weights were open-sourced on July 27 as promised, the largest open-weights model ever.
Zhipu released and open-sourced GLM-5.2 on June 17. On July 9 it announced a ~HK$31.4B H-share placement at HK$1,588 per share, the year's largest single tech placement in Hong Kong. On July 11, Tang Jie's internal letter launched the "Touch High" plan, aimed at four peaks: long-horizon tasks, autonomous agents, fully self-training systems and safety governance.
WAIC closed in Shanghai on July 20 with the launch of the World AI Cooperation Organization, the first intergovernmental AI body, headquartered in Shanghai. Separately, about two-thirds of the world's new robotics unicorns in 2026 are reportedly Chinese.
FutureX Position · Open Source Breaks the Deadlock: The US–China Mismatch Splits Concentration in Two (Xiamen keynote, 2026-09-03)
This keynote reinforces this report's conclusion on the US, corrects it on China, and supplies the mechanism behind concentration at the funding end.
FutureX reads the two markets as a mismatch table:
| Dimension | United States | China |
|---|---|---|
| Cycle | Late in the mania phase | Early to mid phase |
| Funding concentration | Three companies took 67.3% of global AI funding in Q1 2026 | 64.6% of AI funding went to early-stage rounds |
| Exits | No AI company listed in Q1 | 155 Chinese companies went public in H1 |
| Compute | Constrained by power and advanced packaging | Domestic AI accelerators at a 41% share |
| Open vs. closed | Mostly closed | Leads on open source; over 45% of OpenRouter traffic |
So the view this report set out in early September, that concentration spreads from funding to exits, holds only in the US. In China, funding sits in early rounds and the exit window opened first. The two sides cannot run on one clock.
The gap has not closed. AI venture investment in 2025 was $194 billion in the US against $14 billion in China, a roughly 14x difference (PitchBook/NVCA/CAICT). ByteDance's secondary trade in April 2026 valued it above $600 billion, against a Meta market cap of about $1.47 trillion: a 55% discount on price-to-sales. That is our read.
Why does funding concentrate? The keynote's mechanism: AI and non-AI companies now live in two different eras. Median revenue growth for listed legacy SaaS is below 20%. The 2021 VC vintage sits at a median TVPI of about 1.0x (Cambridge Associates). Chinese fund-of-funds investment fell 21.84% year on year in H1 2026 (China FOF Research Center). FutureX's judgment: capital markets now price "uses AI" and "is AI" companies on two different durations, and duration risk on non-AI assets is rising. Concentration follows.
Full deck: /reports/open-source-breakthrough (first 5 pages public).
Latest developments, newest first (data current as of 2026-10-11)
- 2026-10-09 · OpenAI's annualized revenue was about $50B at the end of September, below the nearly $70B reported earlier; the gap comes mainly from converting to Anthropic's method, which counts sales through cloud partners while OpenAI's does not. Bloomberg said OpenAI expects a run rate at or above $70B by year-end, figures shared with investors as it negotiates at a $1.4T pre-money valuation · Reported (first by the FT; Reuters on Oct 8 and Bloomberg on Oct 9, citing people familiar)
- 2026-10-08 · US venture deal value reached $515.8B in the first nine months, about 44% above the previous full-year record; Q3 fell about 40% to $98.4B, and AI's share dropped from 82.7% year to date to 65.9% in the quarter; PitchBook's exit model gives Anthropic an 86% chance of an IPO within a year and OpenAI 12% · Reported (PitchBook-NVCA Venture Monitor, via SiliconANGLE)
- 2026-10-06 · DeepSeek is close to raising at least RMB 80B (about $12B), up from an initial RMB 50B target, and nearly $15B if the total approaches RMB 100B; Tencent and CATL committed some of the largest amounts, the round values it at a minimum of about $75B, and it plans to restructure ahead of an IPO in early 2027 · Reported (Bloomberg)
- 2026-10-06 · Moonshot AI closed its final private round before listing at about $50B and is considering a Hong Kong IPO in Q1 2027 raising up to $5B; that is about 43% above the $35B post-money value in July, though Bloomberg did not say whether $50B is pre- or post-money · Reported (Bloomberg)
- 2026-10-05 · Global venture funding was $159B in Q3, down 25% from Q2 and up 53% year over year; a record 27 companies raised rounds of $1B or more, together about a third of the quarter's total · Reported (Crunchbase data)
- 2026-10-01 · Anthropic plans to start its IPO roadshow on November 9 and, per people familiar, still expects to list by year-end at the latest; it expects to meet prospective investors at its San Francisco headquarters on October 14; some prospective investors put its fair value at $1.8T to $2T · Reported (Bloomberg, via The Standard)
- 2026-09-29 · Anthropic's unpublished draft prospectus shows 2025 revenue of $4.6B, an operating loss above $8B, a net loss of $42B, and $20.28B in cash, cash equivalents and short-term investments at year-end; Q1 2026 revenue was $4.73B; it plans to spend $518B on cloud services and data centers; two clients supply a quarter of revenue; per the FT, more than a third of the draft covers business risks, including "existential risks to humanity" · Reported (obtained by Reuters, also reported by the FT, summarized by Fortune)
- 2026-09-29 · OpenAI is in talks to raise at least $30B in a pre-IPO round at about $1.4T, roughly 64% above the $852B set in March; the round would serve as a bridge to an IPO, and CEO Sam Altman has ruled out a 2026 listing · Reported (Bloomberg, via TechCrunch)
- 2026-09-24 · DeepSeek is targeting RMB 50B (about $7.5B) for its second round at an RMB 500B pre-money valuation, closing by the end of October, about two months later than reported in late August; Liang Wenfeng told investors annualized revenue has reached $1B, up from $400M-500M in July, with an 82.9% gross margin on the model API business in the first seven months; it has retained CITIC Securities for a STAR Market listing, plans to file within the year and targets a 2027 debut · Reported (Sina Finance and Tencent News citing The Information)
- 2026-09-15 · OpenAI is weighing a pre-IPO round; Bloomberg puts the valuation above $1.2T and Forbes, on Sept 16, at $1.5T; both are headline-level, and the round size and investors could not be verified; that is a 41% to 76% step-up from the $852B March round · Reported (Bloomberg, Forbes)
- 2026-09-13 · Anthropic has picked Nasdaq; the IPO could raise up to $100B at a discussed $2T valuation, with Morgan Stanley, Goldman Sachs and JPMorgan leading; Q2 revenue more than doubled to $11.6B; Nvidia is reportedly weighing about $10B in the offering; the raise ceiling is about two-thirds above the "more than $60B" recorded in June · Reported (Bloomberg; Yahoo Finance on Sept 15)
- 2026-09-08 · Cognition announced a $2B Series E at a $48B post-money valuation, twice the size first reported; The Information put it at more than $2B · Verified (Bloomberg, Reuters)
- 2026-09-08 · Cognition's round was led by a16z and Accel among others, with outlets differing on the full list of leads; annualized revenue rose from $492M at its May round to nearly $900M · Reported (Investing.com and others)
- 2026-09-08 · Mistral AI closed a €3B Series D (about $3.58B) led by Samsung Electronics, with the EQT-managed Scaleup Europe Fund and existing investor PSG Equity as co-leads, at a post-money valuation above €21B (about $24B by CNBC's conversion), nearly double the €11.7B set a year earlier in the ASML-led round; new investors include Advent, BlackRock-managed funds and the Grand Duchy of Luxembourg, and a16z, ASML, Nvidia and Salesforce Ventures returned; the company calls it the largest equity round ever by a European tech company · Verified (TechCrunch, CNBC)
- 2026-09-08 · Mistral says it operates in 20 countries and serves more than 125 enterprises, including Airbus, ASML and HSBC; proceeds go to compute, infrastructure, commercial growth and international expansion; CEO Arthur Mensch told CNBC he expects to beat his earlier forecast of more than $1B in annual recurring revenue in 2026 · Per company disclosure (Mistral announcement, CNBC interview)
- 2026-09-02 · Cognition was in talks to raise about $1B at about $47B · Reported (Bloomberg)
- 2026-08-25 · Anthropic is pitching IPO investors on a so-called $30T market opportunity · Reported (Quartz)
- 2026-08-17 · DeepSeek moved to peak/off-peak API pricing from this date, with peak rates up as much as 1,100% and off-peak at half the peak rate · Verified (Quartz, Engadget, reported Aug 13-16)
- 2026-08-13 · DeepSeek released V4-Pro · Verified (Quartz, Engadget)
- 2026-08-13 · Anthropic's investors expect an October listing at a valuation above $2T, which would be the largest IPO ever; the figure traces to a single sourcing chain, is unconfirmed by the company and has no independent verification · Reported (Quartz, Fortune)
- 2026-08-11 · Moonshot AI set August 27 as the final close of its pre-IPO round at about $50B pre-money, with a Hong Kong filing planned by September 30; that is about 40% above the $35B post-money in July, and single-sourced · Reported (KrASIA citing IPO Zaozhidao)
What This Means for Founders
The surge has not spread: PitchBook notes that without the OpenAI and Anthropic rounds, US funding has tracked the same trend since late 2024. Large rounds go to companies that already have real revenue: Cognition, at nearly $900M in annualized revenue, lifted its raise from $1B to $2B in six days. A third pool, beyond the US and China, is forming in Europe: Mistral's €3B round drew Samsung, the Luxembourg state and an EU-backed fund, though it is still a fraction of a single US frontier-lab round. Signals to watch: Anthropic's pricing after its November 9 roadshow, and the early-2027 listing windows for DeepSeek and Moonshot.
What This Means for Investors
The concentration call holds, but two companies carry it: AI's share of US venture fell to 65.9% in Q3. Capital needs are likely to keep pooling at the top: Anthropic's draft prospectus reportedly shows plans to spend $518B. The valuation anchor is highly sensitive to revenue definitions: OpenAI is reportedly in talks to raise at about $1.4T pre-money, about 28 times a $50B run rate and about 20 times $70B (Reported). The US-China question has shifted from who is valued higher to who prices first: Anthropic is set for Nasdaq, DeepSeek is preparing for the STAR Market. This is not investment advice.
Where We Could Be Wrong
- 01If Anthropic ultimately prices below $1T, or its listing slips into 2027, the view that concentration has spread to exits should be downgraded.
- 02If AI's share of US venture stays below 70% in Q4, or OpenAI's current round closes below $1T, the concentration call should be downgraded to plateauing at a high level.
- 03If neither DeepSeek nor Moonshot has filed or listed by the end of Q1 2027, the view that US and China exits run on separate clocks does not hold.
Key Questions
How large was US and global AI venture funding in H1 2026?
US venture funding hit a record $412.7B in H1 2026, nearly 30% above all of 2025, with AI taking $355.9B (86%). Global venture reached $510B, already exceeding 2025's full-year $440B. OpenAI ($122B) and Anthropic ($95.6B) together absorbed 43% of global venture funding (data as of 2026-08-01, per PitchBook/Crunchbase).
When will Anthropic and OpenAI IPO, and at what valuation?
Anthropic filed a confidential S-1 on June 1, 2026, reportedly now targeting a Nasdaq listing as early as mid-November (before Thanksgiving) to raise up to $100B, with prospective investors discussing a valuation of about $1.8-2T. OpenAI reportedly leans toward delaying its IPO to 2027; its latest raise was $122B at an $852B post-money valuation.
What funding and model moves did DeepSeek and Moonshot AI (Kimi) make in July 2026?
DeepSeek V4 went GA on July 20 (1.6T-parameter MoE, 1M-token context) with peak/off-peak API pricing, and DeepSeek was reportedly in talks for a new round at roughly $71-74B. Moonshot released the 2.8T-parameter Kimi K3 on July 16, open-sourced its weights on July 26, closed a $3.5B round at a $35B post-money valuation, and reportedly seeks a pre-IPO round at up to $50B.
Watch & Listen
▶ CNBC Interview | Where Does China's Real AI Advantage Lie?YouTube · needs VPN in China
▶ Half a Year of AI: Where Are the Second-Half Opportunities? | Jul 22 LivestreamYouTube · needs VPN in China
▶ US-China AI Investing 2026 | LivestreamYouTube · needs VPN in China
▶ The Next Non-Consensus OpportunityYouTube · needs VPN in ChinaIn China: search WeChat Channels for 「倩姐投AI」; full library → Qian on AI
Sourcing and standards
Compiled from public sources; data current as of 2026.10.11. The text separates four evidence grades: verified, reported, per company disclosure and public market data, and states the derivation behind every estimate. When we get something wrong, the correction is written into the report body with the original call left visible, and logged publicly.
Research standards & corrections →📄 Full Report
Full report: 14 pages · provided to professional investors & partners only
This is the public preview. The full report includes the sections below. For compliance reasons it isn't posted publicly or offered as a free download. To request a copy, contact the FutureX team.
- 🔒IV. Narrative Reversal: Kimi K3's Coding Crown and Open-Weights Landing
- 🔒V. Compute and Energy: xAI's APR Energy Acquisition and the Vertical Integration Race
- 🔒VI. Channels and Going Global: SoftBank x Sierra and Enterprise AI Agent Deployment
- 🔒VII. Value Chain Breakdown: Beneficiaries Across Training, Inference, Energy, Hardware and Applications
- 🔒VIII. Risk Notes: Valuation Scissors, Litigation, Regulation and Exit Uncertainty
- 🔒IX. H2 Outlook: The Anthropic IPO, Open-Source Monetization and US-China Capital Rhythms
Where we stand on this
US and China in AI: one cycle, two positions
The US and China sit at two positions in one AI cycle: the US late in its mania phase, China early to mid. The gap is about 14x in capital, about 2.7% in model capability and about one third in valuation. Founders who bill in dollars are closing that discount themselves.
Full argument and falsification tests →The bubble sits in the middle layer that must keep refinancing: the FutureX seven-dimension bubble scorecard
The AI bubble sits in the middle layer that must keep refinancing: compute leasing on borrowed money, circular financing, GPU-backed debt. FutureX Capital has scored seven dimensions daily since March 2026; the August reading lit six, and liquidity triggered on August 17. The scorecard answers one question: which layer cannot be bought at the consensus price.
Full argument and falsification tests →Questions people ask next
Related Research
Private Markets / Exits
AI Exits Now Stand on Three Legs, but Mega-IPO Pricing Is Tightening
Foundation Models
Open Models Near the Frontier at 1/40th the Price, Making AI Sovereignty Institution-Scale
Embodied Intelligence / Physical AI
Physical AI's Year One: Shipments Climb, Unitree Sits About 61% Below Its Debut Open
Industry research from FutureX Capital's AI Lab, compiled from public information; not investment advice; contains no fund performance, AUM, or offer to raise capital.