Global VC Hits $750B💰, AI Raises $519B📈, a $200 Claude Code Team🤖
- 01Theme: Venture capital is hitting record levels, but capital is hyper-concentrated in AI and mega-rounds
- 02AI is absorbing the majority of venture dollars
- 03The money is concentrated in a handful of US companies
- 04Large funds are being raised to match
- 05Theme: AI progress and revenue are outrunning expert forecasts
- 06Even superforecasters underestimated AI progress
1. Key Themes
Theme: Venture capital is hitting record levels, but capital is hyper-concentrated in AI and mega-rounds
AI is absorbing the majority of venture dollars
AI now takes more than half of all funding, and the AI-specific total is already more than double last year's.
"Venture investment reached roughly $750 billion with a quarter still to go in 2026, above the previous record set in 2021. AI took 54% of 2026 funding, and 88 rounds of $1 billion or more make up over 60% of disclosed deal value."
"AI companies raised $519B in 2026 through mid-September, 2.3 times the 2025 total"
The money is concentrated in a handful of US companies
Geographic and deal-level concentration is extreme, so the "record" is driven by a very small number of players.
"US-headquartered companies took $393B, or 76% of the total, with China second on $59B and Europe third on $40B. The 15 largest rounds accounted for $312B, led by a $122B OpenAI round and two Anthropic rounds of $65B and $30B."
Large funds are being raised to match
Fund formation is mirroring the concentration, with a $10B growth/late-stage vehicle leading the list of launches.
"Disruptive announced its $10 billion flagship fund to back growth and late-stage private technology companies."
Theme: AI progress and revenue are outrunning expert forecasts
Even superforecasters underestimated AI progress
Forecasters were too conservative on both capabilities and commercial traction, and the people closest to the technology were the most accurate.
"Most forecaster groups predicted slower progress than occurred, and AI company leaders came closest. FrontierMath hit 94% by September 2026 against a 2030 forecast of 97%, and OpenAI and Anthropic passed $100B in annualized revenue by July against a superforecaster year-end figure of $90B."
Closed models still lead open-weight models
Open-weight models trail most in science-oriented evaluations, which may matter for founders choosing between open and closed stacks.
"The Intelligence Index v5 gives knowledge work, coding, science and general skills 25% of the score each, which leaves eight closed models ahead of open-weight Kimi K3.1 Pro. Open models trail most on science evals such as Terminal-Bench-Science and CritPt v2."
Theme: Agentic AI is moving from demo to production, in finance and in the CRM/workflow stack
Autonomous agents are taking over finance operations
Reported results show dramatic cycle-time improvements on top of existing ERPs.
"Hyphenate learns a company's finance policies and workflows, runs order-to-cash and procure-to-pay, and brings in a person when a judgment call is needed. Dura Software reports a month-end close cut from 3 weeks to 5 days, and Kiteworks reports 98% straight-through processing on cash transactions."
Production AI needs real-time data and narrow agents
Learnings from financial-services teams suggest general-purpose agents rarely survive production.
"Teams at Databricks and Brex explain why behavioral data beats point-in-time snapshots and why one agent for everything rarely survives production. The whitepaper also shows how AI governance helped one team ship 12x faster."
Software is being rebuilt for agents as users
"Your team now runs on people and agents, and both need the controls of business software at the scale of infrastructure."
(Note: this quote comes from a sponsored placement for Attio.)
Theme: Early-stage and secondary markets are shifting
Pre-seed is growing as a share of activity while deal sizes shrink
In crypto, early-stage share is rising and the median deal size is falling, even as valuations hit records.
"Pre-seed grew to 25% of crypto and blockchain venture deals in Q3 2026, while later-stage deals slipped to 24%. Median valuations hit an all-time high of roughly $90M as the median deal size fell from Q2's record $5M to roughly $3M."
Pre-seed dominates the current founder fundraising pipeline
"Pre-seed makes up 74% of the list, and the top round comes from a bootstrapped US fintech with $12.13M in GMV raising a $10M late seed."
The venture secondary market may have peaked
"Early SpaceX buyers are already cashing out, and PitchBook expects a larger pullback once Anthropic and OpenAI exit. Demand is broadening from top AI names into fintech and defense as AI pricing grows steeper."
2. Contrarian Perspectives
Small teams plus agents can replace a salaried team's repeat work for a fraction of the cost
Against the consensus that scaling output means scaling headcount, the article claims a five-person team's recurring work can be covered by a single $200/month subscription, with the founder retaining judgment calls.
"Scheduled tasks and subagents take over repeat work from inbox triage to post-deploy checks, while the founder keeps approvals and customer calls. All eight run on Claude Max 20x at $200 a month, about 0.35% of the $679,900 a year five US software developers earn at the median wage."
Caveat: this compares automating "repeat work" against full developer salaries, so the 0.35% figure is a framing device rather than a like-for-like replacement.
"One agent for everything" doesn't survive production
Counter to the narrative of a single general-purpose agent, practitioners at Databricks and Brex reportedly found that narrow, governed systems built on behavioral data work better.
"Teams at Databricks and Brex explain why behavioral data beats point-in-time snapshots and why one agent for everything rarely survives production."
Investor debates fail because investors argue from different mindsets, not because of disagreement on facts
Rather than treating deal debates as disputes over the merits, the essay claims they turn incoherent because participants weigh different kinds of evidence. The prescribed fix is consistency of philosophy.
"Financial, productive, frontier and speculative investors each weigh different evidence, and debates break down when one deal gets argued from several mindsets at once. Stage often tracks the mindset, with first checks closest to frontier bets and growth rounds closest to financial ones, so the essay argues for one core philosophy applied consistently."
3. Companies Identified
- Description: AI lab behind Claude
- Why mentioned: Raised two mega-rounds, passed $100B annualized revenue alongside OpenAI, and is making costly safety commitments
- Quotes:
- "two Anthropic rounds of $65B and $30B"
- "Anthropic's CEO points to costly choices such as holding Mythos-class models until June and restricting sales to PRC-controlled companies"
- "Anthropic is putting 5% of the company into a public benefit fund."
OpenAI
- Description: Frontier AI lab
- Why mentioned: Largest single AI round of the year; cited in the revenue forecasting miss
- Quotes:
- "led by a $122B OpenAI round"
- "OpenAI and Anthropic passed $100B in annualized revenue by July against a superforecaster year-end figure of $90B."
Attio
- Description: CRM positioned for the agentic era (sponsor)
- Why mentioned: Sponsored placement claiming agent-scale architecture
- Quotes:
- "1000x the agent traffic of a legacy CRM"
- "Your whole CRM queryable in SQL"
- Description: Autonomous agents for the finance close, layered on existing ERPs
- Why mentioned: Case study of agentic AI reaching production in finance
- Quotes:
- "Dura Software reports a month-end close cut from 3 weeks to 5 days, and Kiteworks reports 98% straight-through processing on cash transactions."
- Description: Private aerospace company
- Why mentioned: Its shares are starting to release into the secondary market
- Quotes:
- "Early SpaceX buyers are already cashing out"
- Description: AI inference chip maker
- Why mentioned: Largest deal of the week
- Quotes:
- "raised $620M in Series D funding to scale production of its AI inference chips."
- Description: AI-powered enterprise workflow platform
- Why mentioned: Series C
- Quotes:
- "raised $150M in Series C funding to expand its AI-powered enterprise workflow platform."
- Description: Builder of autonomous factories
- Why mentioned: $100M at Series A
- Quotes:
- "raised $100M in Series A funding to build autonomous factories."
Collective
- Description: AI-native financial operating system
- Why mentioned: $100M raise including a $40M Series B
- Quotes:
- "raised $100M, including a $40M Series B, to expand its AI-native financial operating system."
Casap
- Description: AI platform for financial services
- Why mentioned: Series B
- Quotes:
- "raised $56M in Series B funding to expand its AI platform for financial services."
Wabi
- Description: AI assistant and collaborative mini-app platform
- Why mentioned: $55M raise
- Quotes:
- "raised $55M in funding to develop its AI assistant and collaborative mini-app platform."
Opentrons Labworks
- Description: Laboratory robotics platform
- Why mentioned: Series E
- Quotes:
- "raised $40M in Series E funding to scale its laboratory robotics platform."
Braid
- Description: Business connectivity platform
- Why mentioned: Series A
- Quotes:
- "raised $38M in Series A funding to expand its business connectivity platform."
StimCardio
- Description: Wearable therapy for atrial fibrillation
- Why mentioned: Series A
- Quotes:
- "raised $30M in Series A funding to advance its wearable therapy for atrial fibrillation."
Easol
- Description: Experience-commerce platform
- Why mentioned: Series A extension
- Quotes:
- "raised $30M in Series A extension funding, including $20M in equity, to expand its experience-commerce platform."
Midstream Health
- Description: Healthcare financial operations platform
- Why mentioned: Series A
- Quotes:
- "raised $20M in Series A funding to expand its healthcare financial operations platform."
Databricks and Brex
- Description: Data and fintech companies
- Why mentioned: Practitioner teams cited for production AI lessons in financial services
- Quotes:
- "Teams at Databricks and Brex explain why behavioral data beats point-in-time snapshots"
Plaid, PitchBook, Galaxy, Bloomberg, Voices
- Description: Sources of the reports linked in the newsletter
- Why mentioned: Publishers of the underlying research (data on financial AI, the secondary market, crypto VC, global funding, and voice AI data)
- Quotes:
- "Venture investment reached roughly $750 billion..." (Bloomberg)
- "PitchBook expects a larger pullback once Anthropic and OpenAI exit."
- "Pre-seed grew to 25% of crypto and blockchain venture deals in Q3 2026" (Galaxy)
METR
- Description: AI evaluation organization
- Why mentioned: Serves as Anthropic's embedded evaluator
- Quotes:
- "METR, the first embedded evaluator, affirms the risk is low and questions the risk report's methods"
Artificial Analysis / Kimi K3.1 Pro
- Description: AI benchmarking firm / leading open-weight model
- Why mentioned: Benchmark results show closed models leading
- Quotes:
- "eight closed models ahead of open-weight Kimi K3.1 Pro"
Newly launched funds: Disruptive ($10B flagship), Inspired Capital ($450M raised, Fund IV), Trust Ventures ($400M target, Fund V), Interlagos Capital ($400M target, Fund II), AP Ventures ($300M, Fund III, industrial and energy tech), DPI Venture Capital ($112M, Africa/Middle East), Entourage (€100M, Fund II), Auriga Partners (€80M interim close, cybersecurity), Phaze Ventures (over $75M combined), BY Venture Partners ($60M, AI-native)
- Why mentioned: Fund formation tracking
- Quotes:
- "Disruptive announced its $10 billion flagship fund to back growth and late-stage private technology companies."
- "AP Ventures closed its third venture fund at $300 million to back early-stage industrial and energy technology companies."
- "BY Venture Partners closed its third venture fund at $60 million to back early-stage AI-native companies."
4. People Identified
Ruben Dominguez
- Description: Author of The VC Corner
- Why mentioned: Curator of the newsletter, resources, and the startup list
- Quotes:
- "Every week, founders send me decks asking for the honest read. And every week I see the same pattern: five friendly people said 'looks great, maybe tighten slide 12,' then forty investors said nothing at all."
Nathan Benaich
- Description: Author associated with the State of AI Report
- Why mentioned: Source of the $519B AI funding analysis
- Quotes:
- "AI companies raised $519B in 2026 through mid-September, 2.3 times the 2025 total"
Benjamin Todd
- Description: Author of the forecasting accuracy analysis
- Why mentioned: Showed forecasters underestimated AI progress
- Quotes:
- "Most forecaster groups predicted slower progress than occurred, and AI company leaders came closest."
Martin Casado
- Description: Investor, credited with the four-mindsets essay
- Why mentioned: Framework for why VC deal debates become incoherent
- Quotes:
- "Financial, productive, frontier and speculative investors each weigh different evidence"
Zvi Mowshowitz
- Description: Commentator on AI policy
- Why mentioned: Analysis of Anthropic's safety stances and the scrutiny they face
- Quotes:
- "critics read its federal bill and revised scaling policy as resistance to binding limits."
Nathan Lambert
- Description: AI researcher and writer focused on open models
- Why mentioned: Analysis of the Artificial Analysis v5 index
- Quotes:
- "Open models trail most on science evals such as Terminal-Bench-Science and CritPt v2."
Eli
- Description: Author of the chip-level verification piece
- Why mentioned: Proposes how a US-China AI slowdown could be enforced
- Quotes:
- "compute is the easiest input to watch because it is energy-intensive and comes from a few suppliers."
Luis Llorens
- Description: Collaborator on the startup list
- Why mentioned: Co-built the October list of startups raising
- Quotes:
- "46 founders submitted their round and deck for the October list, built with Luis Llorens"
5. Operating Insights
Systematize repeat work with scheduled tasks and subagents, keeping humans on approvals and customer relationships
A small team can offload inbox triage and post-deploy checks to agents while the founder retains judgment, approvals, and customer calls.
"Scheduled tasks and subagents take over repeat work from inbox triage to post-deploy checks, while the founder keeps approvals and customer calls."
Design AI systems with human escalation and narrow scope
Production-grade agents succeed by learning an organization's policies, automating the routine, and escalating judgment calls rather than aiming for total autonomy.
"Hyphenate learns a company's finance policies and workflows, runs order-to-cash and procure-to-pay, and brings in a person when a judgment call is needed."
Build governance in early to ship faster
Governance is framed as a speed enabler, not a brake.
"The whitepaper also shows how AI governance helped one team ship 12x faster."
6. Overlooked Insights
Deal-size inflation may be a mix effect, not a broad-based boom
With 88 rounds of $1B+ making up over 60% of disclosed value, the "record" headline masks a very different picture for typical rounds. The founder list shows a median ask of just $750K, with 74% pre-seed.
"42 of them state an exact amount, with a median ask of $750K."
Corporate venture arms are a large, under-appreciated source of seed capital for AI
Outside China, hundreds of corporates write early checks, and a meaningful minority of active funds are doing AI deals right now, which is useful for founders mapping who to pitch.
"Outside China, 610 corporate venture arms write seed checks into AI, and 274 of the 2,594 venture funds with a dated deal announced an AI deal in the last 30 days."