All funding news

AI funding news

100 recent AI rounds across our tracked sources.

Instinct logo
🇺🇸InstinctAI Assistant

Instinct builds an AI agent that organizes your life by connecting to your apps and devices, accessible via text and call.

$250MSeries B
Index Ventures logo
Index Ventures
Benchmark logo
A $250M Series B for a conversational life-management agent signals that multi-app orchestration via natural language is now table stakes for AI assistants—the market is past proof-of-concept and into distribution/retention bets. Instinct is likely burning this on: (1) expanding integrations across the app ecosystem, (2) improving reliability of multi-step workflows (the hard part), and (3) user acquisition to hit retention thresholds that justify the burn. If you're building in vertical SaaS or workflow automation, watch whether Instinct's traction forces you to add a conversational layer or risk commoditization.
General Intuition logo
🇺🇸General IntuitionPhysical AI / Robotics

General Intuition builds foundation models for AI agents that learn to move and act through physical space, enabling generalized robotic control.

UndisclosedSeries B
Valor Equity Partners logo
Valor Equity Partners
PSKhosla Ventures logo
+1
Point72 and Valor both backing this suggests institutional capital is comfortable with the risk profile at Series B scale right now—likely a signal the company has proven unit economics or a defensible moat. If you're in fintech, enterprise software, or anything with a clear path to institutional adoption, this round composition (hedge fund + traditional VC + founder-led) means LPs are actively deploying into proven-traction companies, not just hype.
Gatik logo
🇺🇸GatikAutonomous Vehicles

Gatik builds autonomous trucks for middle-mile freight logistics using AI.

$200M
Investor undisclosed
A $200M round for middle-mile autonomous trucking signals that investors are finally betting on the unglamorous logistics layer—not robotaxis. This capital likely funds fleet expansion and ops scaling rather than R&D, which means Gatik thinks their tech is production-ready. If you're building in logistics software or last-mile delivery, watch whether Gatik's unit economics actually work; if they do, you've got a wedge to sell into their expanding fleet.
芯光界 logo
🇨🇳

芯光界

Chip Design

芯光界 designs AI chips optimized for edge computing and inference workloads.

$14MAngel
启明创投 logo
启明创投
A $14M angel for edge AI inference chips from a top-tier Chinese VC signals the inference-at-edge category is moving past hype into real deployment phase—this isn't pre-product money. If you're building AI applications or infrastructure, watch whether Chinese edge chips start undercutting Western alternatives on cost; if they do, your margin story changes fast.
Moonshot AI logo
🇨🇳Moonshot AILLM Infra

Moonshot AI builds Kimi Operator, an AI assistant platform powered by large language models for enterprise and consumer applications.

$3.5BSeries F
Investor undisclosed
A $3.5B Series F for a Chinese LLM platform signals that post-frontier model commoditization, the real capital is flowing to agent/operator layers—the stuff that actually does work. If you're building in automation, workflow, or enterprise AI, this validates that investors are betting hard on agentic systems over raw model capability, which means your GTM should emphasize task completion, not just inference quality.
Rillet logo
🇺🇸RilletAI Accounting/ERP

Rillet automates accounting workflows for finance teams by pulling data from business tools like Salesforce and Brex.

$100MSeries C
I
Iconiq
Sequoia logoAndreessen Horowitz logo
A $100M Series C for accounting automation signals that enterprise finance teams are finally willing to pay for AI that actually reduces headcount—not just speeds up existing work. The Iconiq + Sequoia + a16z combo suggests this is being positioned as a platform play, not a point solution, which means Rillet is likely burning cash on sales/GTM and building out integrations to become sticky across the entire finance stack. If you're building any workflow automation in enterprise, watch how they handle the integration moat—that's where most of these companies either become indispensable or get commoditized.
幺正量子 logo
🇨🇳

幺正量子

Quantum Computing

幺正量子 develops quantum computing solutions for enterprise applications.

UndisclosedSeries A
Investor undisclosed
线控科技 logo
🇨🇳

线控科技

AI

线控科技 builds AI-powered control systems for industrial automation and smart manufacturing.

UndisclosedSeries A
Investor undisclosed
Hengwan Technology logo
🇨🇳Hengwan TechnologyEdge Computing Infrastructure

Hengwan Technology converts cellular base stations into distributed computing nodes for edge AI inference.

UndisclosedSeries B+
Investor undisclosed
深海智人 logo
🇨🇳

深海智人

AI

深海智人 builds AI-powered solutions for enterprise customers in China using deep learning and computer vision.

$70MSeries A
Investor undisclosed
A $70M Series A for a China-based computer vision shop signals that enterprise AI adoption in China has moved past pilot phase—customers are writing big checks for production deployments. If you're building vision or deep learning tools for any regulated vertical (manufacturing, logistics, finance), watch how 深海智人 navigates compliance and data residency; that playbook will matter when you eventually sell into Asia or to Asian companies operating globally.
Atoms logo
🇺🇸AtomsRobotics

Atoms builds a universal robotics platform for heavy industry, manufacturing, and logistics automation.

$1.7B
Andreessen Horowitz logo
Andreessen Horowitz
A16z writing a $1.7B check to a robotics platform company signals they're betting hard that the unit economics of industrial automation finally work—likely because vision models + cheaper hardware have crossed a threshold. Atoms is probably using this to scale manufacturing of their own robots and build out the software layer that makes them work across different factory setups. If you're building any kind of automation tooling (software, hardware, or integration), this validates that customers will pay for solutions that reduce labor costs, not just optimize existing processes.
Fractile logo
🇺🇰FractileChip Design

Fractile designs AI inference chips optimized for edge deployment and efficient computation.

$600MSeries B
Lightspeed Venture Partners logo
Lightspeed Venture Partners
Redpoint logoFounders Fund logoAccel logo
+3
A $600M Series B for edge AI chips signals serious conviction that inference is moving off-cloud—this isn't speculative, it's capital-intensive infrastructure betting. The NATO Innovation Fund co-leading suggests geopolitical urgency around sovereign compute, which means regulatory tailwinds and defense/industrial customers are real revenue drivers, not just nice-to-haves. If you're building any application layer that depends on model deployment (robotics, autonomous systems, embedded ML), you should care because the chip economics just shifted—edge inference is about to get cheaper and faster, which changes what's actually viable to build.
Velatir logo
🇩🇰VelatirAI Infrastructure

Velatir helps enterprises monitor, control, and safely scale employee AI tool usage across their workforce.

$5.5MSeed
S
Spintop Ventures
UNE
A $5.5M seed for employee AI governance signals enterprises are past the 'let people experiment' phase and now need guardrails—this is the compliance layer that follows adoption. If you're building any internal tool or workflow product, watch how Velatir positions around data leakage and audit trails; that's becoming table stakes for enterprise sales, not a nice-to-have.
Domyn logo
🇮🇹DomynLLM Infra

Domyn builds large-language models and LLM infrastructure for enterprises.

$1.1B
Investor undisclosed
A $1.1B round for an Italian LLM infrastructure play signals that enterprise AI infrastructure is still consolidating around non-US players—likely because Domyn has locked in specific verticals (finance, manufacturing) where data sovereignty or regulatory moats matter more than raw model quality. If you're building B2B AI tooling, this validates that enterprises will pay for localized inference and fine-tuning stacks, not just API access to frontier models.
Callosum logo
🇺🇰CallosumAI InfrastructureVerified

Callosum helps developers route AI workloads across multiple models and hardware providers to optimize compute efficiency.

$100MSeed
Atomico logo
Atomico
PU
A $100M seed for an AI routing layer signals that model fragmentation is now a real operational problem—not theoretical. Callosum is likely building the plumbing to let teams avoid vendor lock-in and arbitrage cost/latency across Claude, GPT, open models, and custom hardware, which means they're betting on a future where no single provider dominates inference. If you're building any AI product with meaningful compute spend, this matters: the economics of your stack just became negotiable.
Gestalt Technology logo
🇨🇳Gestalt TechnologyAI

Gestalt Technology builds AI-powered solutions for enterprise clients in China.

$58.8MAngel+
Investor undisclosed
A $58.8M angel-plus round for an enterprise AI play in China signals investors are betting on rapid deployment cycles and willingness to pay in that market—this isn't a Series A, so the capital likely funds sales/implementation teams and custom model work rather than R&D. If you're building B2B AI tools, watch whether Gestalt's go-to-market (direct sales vs. platform) becomes the template for China-first companies, since the playbook there often diverges from US enterprise.
Unitree logo
🇨🇳UnitreeRobotics

Unitree builds humanoid robots powered by physical AI for industrial and commercial applications.

$21IPO
Investor undisclosed
A Chinese robotics company going public signals that physical AI has moved from R&D theater to revenue-generating reality—investors are betting on unit economics, not just demos. If you're building in adjacent hardware spaces (autonomous systems, industrial automation, supply chain), watch Unitree's IPO filing for gross margins and customer concentration; that'll tell you whether the market is actually paying for humanoid labor or if it's still subsidized by hype.
BookMyShow logo
🇮🇳BookMyShowEntertainment TicketingVerified

BookMyShow is an online ticketing platform for movies, events, and shows across India.

$50MStrategic
K
KKR
KKR's $50M strategic check into BookMyShow signals that India's entertainment ticketing is now mature enough for PE-style consolidation plays—they're likely betting on margin expansion and cross-category bundling rather than user growth. If you're building in adjacent Indian consumer verticals (travel, dining, events), this validates that large cheques flow to platforms that can aggregate fragmented supply and lock in repeat transactions.
Reach Capital logo
🇺🇸Reach CapitalAI-focused VC Fund

Reach Capital is a venture fund backing AI founders building applications for learning, health, and work.

$265MFund V
Investor undisclosed
A $265M Fund V close in Aug 2026 signals LP conviction that AI-in-education/health/work is past the hype phase and into deployment—especially with institutional LPs like pensions and foundations committing. If you're building in adjacent verticals (compliance, HR tech, credentialing), this validates that boring-but-essential use cases are now fundable at scale, not just consumer AI.
魔芯科技 logo
🇨🇳

魔芯科技

4D World Model

魔芯科技 builds 4D world model technology for AI applications using advanced spatial-temporal understanding.

$140M
Investor undisclosed
A $140M round for 4D world models signals serious capital flowing into embodied AI infrastructure—this isn't just vision, it's spatial reasoning at scale. Chinese investors are betting hard that 4D understanding (not just 2D perception) becomes table stakes for robotics, autonomous systems, and simulation. If you're building anything that needs to reason about physical space or time-series spatial data, watch whether 魔芯's tech becomes a commodity layer or stays proprietary—that determines your build-vs-buy calculus in 18 months.
OceanBase logo
🇨🇳OceanBaseAI Data Platform

OceanBase builds a distributed database and AI data platform for enterprises to manage real-time data and AI workloads at scale.

$35MSeries A
Investor undisclosed
A $35M Series A for a distributed database in 2026 signals that enterprises are finally willing to pay for infrastructure that handles real-time + AI workloads together—not as separate systems. If you're building any data-adjacent product (analytics, observability, feature stores), this validates that the TAM for "unified data platforms" is real enough to attract serious capital, even from China-based teams competing against established players.
Rillet logo
🇺🇸RilletAI Accounting/ERPVerified

Rillet automates accounting workflows for finance teams by pulling data from business tools like Salesforce and Brex.

$100MSeries C
ICONIQ logo
ICONIQ
Andreessen Horowitz logoSequoia logo
A $100M Series C for accounting automation signals that enterprise finance teams are finally willing to pay for AI that actually reduces headcount—not just speeds up existing work. If you're building workflow automation in any back-office function (HR, legal, ops), watch how Rillet uses this capital: likely going deep on vertical-specific connectors and land-and-expand motion, which is the playbook that actually works at this stage.
Hengwan Technology logo
🇨🇳Hengwan TechnologyEdge Computing Infrastructure

Hengwan Technology converts cellular base stations into distributed computing nodes for edge AI inference.

UndisclosedSeries B
Investor undisclosed
Etched logo
🇺🇸EtchedChip Design

Etched designs custom AI chips manufactured by TSMC to compete with Nvidia's offerings.

$700MSeries D
J
Jane Street
A $700M Series D for a chip design startup signals that the market believes there's real room to compete with Nvidia on inference workloads—this isn't vaporware money, it's scale-up capital. Jane Street's involvement (a quant shop with deep hardware conviction) suggests the unit economics work at volume. If you're building AI infrastructure or applications, this matters because custom silicon is becoming table stakes for cost-sensitive deployments; you should be stress-testing whether your roadmap assumes commodity GPUs or planning for alternatives.
独到科技 logo
🇨🇳

独到科技

AI Agent

独到科技 builds an AI Agent compiler platform that transforms business workflows into autonomous agents for enterprises.

$14MSeries A+
赛
赛富
A $14M Series A+ for an AI Agent compiler in China signals that workflow automation is moving past chatbots into actual execution—enterprises want agents that *do* work, not just talk about it. At this stage and size, they're likely building out verticalized templates (finance, HR, ops) and enterprise sales motion. If you're building any kind of business process software, watch how they position 'compilation' vs. orchestration—that framing might matter for how you pitch automation to risk-averse buyers.
中科驭数 logo
🇨🇳

中科驭数

AI

中科驭数builds AI-powered data processing and analytics infrastructure for enterprises using GPU acceleration.

UndisclosedSeries C+
Investor undisclosed
思昇科技 logo
🇨🇳

思昇科技

Brain-Computer Interface

思昇科技 develops CNS drugs and ultrasonic brain-computer interface technology for neurological applications.

UndisclosedSeed+
Investor undisclosed
智合AI logo
🇨🇳智合AIAI

智合AI provides AI-powered solutions for Chinese businesses, likely focusing on enterprise automation and intelligent services.

UndisclosedSeries A
Investor undisclosed
Velaura AI logo
Velaura AIChip Design

Velaura AI designs specialized hardware chips optimized for AI workloads.

Undisclosed
Investor undisclosed
Wispr logo
🇺🇸WisprSpeech RecognitionVerified

Wispr builds AI-powered speech-to-text and meeting transcription for professionals seeking accurate, hands-free documentation.

$280MSeries B
Menlo Ventures logo
Menlo Ventures
NNEA logoN
+8
A $280M Series B for speech-to-text signals the category has moved past novelty—investors are betting on winner-take-most consolidation in pro transcription, likely driven by enterprise adoption at scale. Wispr's probably using this to build out vertical-specific models (legal, medical, finance) and lock in distribution through integrations rather than just chasing accuracy gains. If you're building any workflow tool that touches meetings, calls, or documentation, watch whether Wispr becomes table stakes or remains a point solution—that determines your integration strategy.
PandaAI logo
🇨🇳PandaAIAI

PandaAI builds AI tools for Chinese users and businesses.

Undisclosed
Investor undisclosed
思昇科技 logo
🇨🇳

思昇科技

Brain-Computer Interface

思昇科技 develops CNS drugs and ultrasonic brain-computer interface technology for neurological applications.

Undisclosed
Investor undisclosed
猿声科技 logo
🇨🇳

猿声科技

AI

猿声科技 builds AI-powered voice and audio synthesis technology for content creators and entertainment.

$14M
Investor undisclosed
A $14M round for Chinese voice synthesis in mid-2026 signals that real-time audio generation has moved past the hype phase—this is now infrastructure money, not moonshot betting. If you're building creator tools or platform features that need audio (dubbing, personalization, live content), watch how 猿声科技 positions around latency and voice fidelity; that's where the actual moat lives, not just model size.
拨动一下 logo
🇨🇳

拨动一下

AI Gaming

拨动一下 builds AI-powered casual games for mobile players using generative AI.

UndisclosedAngel
Investor undisclosed
Pragmatik Labs logo
🇨🇳Pragmatik LabsAgents

Pragmatik Labs builds AI agents for knowledge work and physical environments that reason, use tools, and execute long-horizon tasks.

$2B
Investor undisclosed
A $2B Series round for an agent company in August 2026 signals that reasoning + tool-use is now table stakes for enterprise AI—the market has moved past single-task automation. If you're building in adjacent verticals (supply chain, field ops, customer support), expect your investors to ask why your product *isn't* multi-step reasoning; this round just reset the baseline for what "serious" agent infrastructure looks like.
Groq logo
🇺🇸GroqAI ChipsVerified

Groq builds specialized AI inference chips and cloud services for developers and enterprises to run large language models faster and cheaper.

$350MSeries A
D
Disruptive
A $350M Series A for inference chips signals the market is past the 'will anyone need this?' phase and into competitive unit economics—Groq's betting inference becomes a commodity bottleneck worth solving at scale. At this stage and size, they're likely burning cash on manufacturing partnerships and sales/GTM to prove their chips actually win on latency/cost in production workloads, not just benchmarks. If you're building any LLM application layer (RAG, agents, fine-tuning platforms), watch whether Groq's inference becomes a real alternative to NVIDIA—it changes your deployment assumptions and margin math.
Higgsfield logo
HiggsfieldVideo Generation

Higgsfield builds AI video generation software for content creators and enterprises.

$400M
Investor undisclosed
A $400M round for video generation in mid-2026 signals the category has moved past novelty—this is enterprise-scale deployment money, likely split between infrastructure (compute, model training) and go-to-market. If you're building in adjacent generative media (3D, audio, interactive), watch whether Higgsfield's customer concentration skews creator vs. enterprise; that'll tell you which distribution moat actually holds in this space.
微灵医疗 logo
🇨🇳

微灵医疗

Brain-Computer Interface

微灵医疗 develops brain-computer interface technology to restore neural function and communication for patients with neurological conditions.

$14M
Investor undisclosed
A $14M Series A for a Chinese BCI company signals that neural interface hardware is moving past pure research into clinical validation—this is the stage where you need regulatory pathway clarity and animal/early human data. If you're building in adjacent neurotech (neural recording, signal processing, or clinical software), watch whether they're pursuing NMPA approval first or hedging with international trials; that choice telegraphs how fast the Chinese regulatory environment is actually moving for invasive devices.
Databricks logo
🇺🇸DatabricksData & Analytics Infrastructure

Databricks builds a unified data and AI platform for enterprises to run analytics and machine learning workloads on Apache Spark.

$50B
Investor undisclosed
A $50B round for Databricks in 2026 signals that enterprise data infrastructure is consolidating around unified platforms—the market is betting that fragmented stacks (warehouse + lakehouse + ML ops) lose to integrated alternatives. At this stage and size, the money is almost certainly going to sales/go-to-market and R&D to defend against cloud-native competitors (Snowflake, cloud vendors' native offerings). If you're building any data-adjacent tool, this is a warning: enterprises increasingly want one vendor to own the full stack, so point solutions need either a defensible moat or a clear acquisition path into a platform.
Graas logo
🇸🇬GraasRetail Commerce AI

Graas builds AI agents for retailers that unify product, customer, and inventory data to power smarter recommendations across online and offline channels.

$17MSeries B
L
LemmaTree
ITT
+2
A $17M Series B for retail AI agents signals that unified commerce (online + offline) is finally moving from nice-to-have to must-have for mid-market retailers—the data integration problem is real enough to fund. If you're building in adjacent commerce verticals (logistics, pricing, supply chain), this validates that retailers will pay for AI that actually connects their fragmented systems rather than bolting on another siloed tool.
原力无限 logo
🇨🇳

原力无限

Robotics

原力无限 builds ego-native world models for robots to accelerate their autonomous intelligence and decision-making.

$140M
Investor undisclosed
A $140M raise for a Chinese robotics foundation model company signals serious capital is flowing into embodied AI—but the undisclosed investors and vague "ego-native world models" framing make it hard to read whether this is genuine technical progress or hype. If you're building in autonomous systems or sim-to-real, watch whether 原力无限's models actually reduce training time or data requirements; that's the real moat, not the funding size.
Thrive Holdings logo
🇺🇸Thrive HoldingsEnterprise AI

Thrive Holdings acquires traditional businesses and integrates AI solutions into their operations across accounting, IT, and regulatory services.

$2B
SoftBank logo
SoftBank
DA
A $2B round for a roll-up that bolts AI onto legacy service businesses signals SoftBank's conviction that the real money isn't in AI tools—it's in owning the customer relationships and recurring revenue streams that *use* them. If you're building horizontal AI infrastructure, this is a warning: the acquirers have capital to consolidate fragmented markets faster than you can land enterprise deals. Watch how Thrive actually deploys this capital—if it's mostly M&A rather than product, you're in a race against consolidation, not competition.
Laihua logo
🇨🇳LaihuaAI Animation & Content Creation

Laihua builds AI-powered animation and 3D content creation tools for manga and drama producers targeting overseas markets.

$9.5MSeries D
Investor undisclosed
A $9.5M Series D for an AI animation tool suggests the manga/drama export market is consolidating around tooling rather than content—Chinese studios are betting on efficiency gains to compete globally. If you're building for creators (video, games, design), watch how Laihua monetizes: per-project licensing, subscription tiers, or API access will signal which creator workflows are actually bottlenecked enough to pay.
Databricks logo
🇺🇸DatabricksData & Analytics Infrastructure

Databricks builds a unified data and AI platform for enterprises to run analytics and machine learning workloads on Apache Spark.

$5BSeries G
Investor undisclosed
A $5B Series G in 2026 signals that enterprise data infrastructure is consolidating around unified platforms—the days of point solutions are over. Databricks is likely using this to accelerate their AI/ML moat (think: native LLM fine-tuning, governance layers) and international expansion, not just to extend runway. If you're building any data-adjacent tool (observability, quality, lineage), you're now competing against a $50B+ valuation company with distribution—consider whether you're a feature or a standalone business.
Uforce logo
🇺🇰UforceRobotics

Uforce builds autonomous drones for land, air and sea operations serving defence forces.

$4B
Investor undisclosed
A $4B round for autonomous drones signals defence spending is moving from procurement cycles to venture-scale bets—this isn't traditional mil-tech, it's VC-backed autonomy. If you're building in robotics or autonomous systems, watch how Uforce structures its go-to-market with government customers; the playbook for selling to defence at scale is still being written and will likely leak into adjacent sectors.
CodeRabbit logo
CodeRabbitDeveloper Tools

CodeRabbit automates code review for development teams using AI-powered analysis.

$143MSeries C
Investor undisclosed
A $143M Series C for code review automation signals that AI-native developer tools can command serious capital even in a crowded space—the bet is on workflow lock-in and land-and-expand motion, not just point-solution novelty. CodeRabbit is likely using this to build out enterprise sales infrastructure and expand into adjacent CI/CD workflows (testing, deployment gates) where they can own more of the developer's day. If you're building any tool that touches the dev workflow, watch whether they're bundling features or staying focused—that'll tell you if the market rewards specialization or consolidation.
CodeRabbit logo
CodeRabbitDeveloper Tools

CodeRabbit automates code review for development teams using AI-powered analysis.

Undisclosed
Investor undisclosed
希奥端计算 logo
🇨🇳

希奥端计算

Edge Computing

希奥端计算 builds edge computing solutions powered by AI for distributed inference and on-device processing.

UndisclosedSeries A
Investor undisclosed
语用科技 logo
🇨🇳

语用科技

AI

语用科技 builds AI solutions for enterprise applications, founded by Lin Junyang.

UndisclosedAngel
Investor undisclosed
Lovable logo
🇸🇪LovableDeveloper Tools

Lovable builds an AI coding assistant that lets developers build software through natural language and visual feedback.

$400M
Investor undisclosed
A $400M round for an AI coding assistant signals that the market is betting hard on LLMs actually shipping production code—not just drafting it. At this stage and size, Lovable is likely burning cash on compute, hiring senior engineers to improve code quality/safety, and building enterprise sales infrastructure. If you're building any developer tool, watch whether they can actually reduce churn; the graveyard of AI coding assistants is full of tools with great demos but users who bounce after the novelty wears off.
Lovable logo
🇸🇪LovableDeveloper ToolsVerified

Lovable builds an AI coding assistant that lets developers build software through natural language and visual feedback.

$400MSeries C
Menlo Ventures logo
Menlo Ventures
SEC
+12
A $400M Series C for an AI coding assistant signals that the market has moved past 'can AI write code?' to 'who owns the developer workflow?' — Lovable's visual feedback loop suggests they're betting on the IDE as the real moat, not just the model. If you're building any developer tool, this validates that founders are willing to pay for AI that reduces friction in their actual work, not just autocomplete; the question is whether you're solving a step in their flow or trying to replace it entirely.
AOE Tech Labs logo
🇨🇳AOE Tech LabsAgent Infrastructure

AOE Tech Labs builds agent infrastructure that optimizes AI models for long-horizon task execution through their Floatboat desktop product.

UndisclosedSeed
HongShan logo
HongShan
V
HongShan leading a seed in August 2026 signals they're still hunting early, but the investor mix (HongShan + VLight) suggests this is either a geographic play or a space where traditional tier-1 VCs are sitting out. If you're building in infrastructure or deep tech, watch whether HongShan's participation means they're rotating capital away from consumer—that's a real signal about where dry powder is actually flowing.
Pragmatik Labs logo
🇨🇳Pragmatik LabsAgents

Pragmatik Labs builds AI agents for knowledge work and physical environments that reason, use tools, and execute long-horizon tasks.

$220MAngel
G
Gaorong
HongShan logoTencent logoS
A $220M angel round from top-tier China VCs signals serious conviction that agentic AI for knowledge work is past the hype phase—this isn't seed-stage exploration anymore. The size suggests Pragmatik is either solving a specific, high-ROI problem (likely enterprise automation) or the bar for "angel" has shifted dramatically in China's AI arms race. If you're building in adjacent automation spaces (workflow, RPA, or domain-specific agents), watch whether they go horizontal or vertical—that'll tell you if the market is consolidating around general-purpose reasoning or staying fragmented by use case.
MemoraX AI logo
🇨🇳MemoraX AIAI

MemoraX AI builds AI-powered memory and knowledge management tools for individuals and teams.

UndisclosedSeed++
Investor undisclosed
郝建业 logo
🇨🇳

郝建业

LLM Infra

郝建业 builds memory technology infrastructure for large language models.

Undisclosed
Investor undisclosed
Westlake Robotics logo
🇨🇳Westlake RoboticsRobotics

Westlake Robotics builds embodied AI humanoid robots with unified full-body models for industrial and commercial applications.

UndisclosedSeries A
Investor undisclosed
Kling AI logo
🇨🇳Kling AIVideo Generation

Kling AI generates high-quality videos from text and images using advanced AI models.

$2.8B
Investor undisclosed
A $2.8B round for a text-to-video model signals that video generation has moved past the hype phase into serious infrastructure competition—this is OpenAI/Google money, not early-stage VC. If you're building in multimodal AI or content creation tools, watch whether Kling's model becomes the backbone layer (like Stable Diffusion for images) or stays locked behind a proprietary API; that determines whether you build on top or around it.
River AI logo
🇺🇸River AICustom AI ToolsVerified

River AI builds custom AI tools for enterprises, founded by a former XAI co-founder.

$1.1B
Investor undisclosed
A $1.1B raise for enterprise custom AI tooling signals that the market has moved past generic LLM APIs—enterprises are willing to pay for differentiated, purpose-built systems and the founder pedigree (XAI) suggests this isn't just another wrapper. The chip co-investors (NVIDIA, AMD) indicate they're betting on inference-heavy workloads that need hardware optimization, which means River likely burns capital on compute and model fine-tuning, not just headcount. If you're building any vertical AI product, watch how River positions itself against both OpenAI's enterprise offerings and in-house model teams—that positioning will tell you whether custom tooling remains a defensible category or collapses into commoditization.
Discovered Materials logo
🇺🇸Discovered MaterialsMaterials Discovery, SemiconductorVerified

Discovered Materials uses AI agents to automate semiconductor materials discovery, from simulation through synthesis to experimental validation.

$9MSeed
L
Lightspeed India Partners
Y Combinator logoPP
+2
A $9M seed for autonomous materials discovery signals that VCs believe AI agents can compress the 5-10 year semiconductor R&D cycle—this is less about hype and more about the physics being tractable now. If you're building in biotech, drug discovery, or battery materials, watch how Discovered Materials structures their agent workflows; the bottleneck isn't usually the simulation layer, it's integrating real experimental feedback loops without human friction.
S
🇨🇳SeedMitAI Tools

SeedMit combines AI with human expertise to help startups validate real market needs and avoid building for fake problems.

UndisclosedAngel
Investor undisclosed
桥介数物 logo
🇨🇳

桥介数物

Robotics

桥介数物 builds a universal robot operating system enabling seamless control across diverse robotic platforms.

$14M
Investor undisclosed
A $14M Series A for a cross-platform robot OS suggests China's robotics stack is maturing past single-vendor lock-in—this is the infrastructure play that usually comes after hardware proliferation. If you're building robotics applications or middleware, this signals that abstraction layers are now fundable, meaning the market believes fragmentation is expensive enough to solve. Watch whether they go horizontal (all robot types) or vertical (specific verticals like manufacturing)—that'll tell you if the bet is on standardization or specialization.
River AI logo
🇺🇸River AICustom AI Tools

River AI builds custom AI tools for enterprises, founded by a former XAI co-founder.

$1.1BSeed
General Catalyst logo
General Catalyst
A $1.1B seed round for enterprise custom AI tooling signals that generalist LLM APIs are now table stakes—the real money is in verticalized workflows and integration complexity. General Catalyst betting this big on a former XAI founder suggests they're betting on execution + credibility over novel tech, which means the bar for differentiation in this space just got higher (you need either deep domain expertise or a distribution moat). If you're building B2B AI tools, this validates that enterprises will pay for customization, but it also means you're now competing against well-funded teams with strong pedigrees.
Unitree logo
🇨🇳UnitreeRobotics

Unitree builds humanoid robots powered by physical AI for industrial and commercial applications.

UndisclosedIPO
Investor undisclosed
L
🇨🇳LatentVerseEmbodied AI

LatentVerse builds embodied AI systems with world models for autonomous agents and robotics.

Undisclosed
高瓴 logo
高瓴
Cambridge Aerospace logo
🇺🇰Cambridge AerospaceAir Defence

Cambridge Aerospace builds drone and missile interceptor systems for air defence using advanced aerospace engineering.

$300MSeries C
D
DFJ Growth
LAL
+1
A $300M Series C for a UK air defence startup signals serious geopolitical appetite for autonomous interception tech—this isn't venture capital, it's strategic capital betting on near-term deployment. If you're building in robotics, autonomy, or sensor fusion, watch how Cambridge structures their go-to-market with defense primes; that playbook (government sales cycles, integration requirements, export controls) will define what's actually fundable in adjacent hard-tech spaces over the next 18 months.
Deployment Inc logo
🇮🇳Deployment IncEnterprise AI

Deployment Inc embeds AI engineers within enterprises to design, implement, and operate AI systems across critical business workflows.

$5MSeed
C
Cars24
A $5M seed for embedded AI engineering services signals enterprises still can't hire fast enough to build AI in-house—this is a staffing arbitrage play, not a software play. If you're building AI tooling or platforms, this validates that implementation bottlenecks remain real even in 2026, which means there's still room for tools that make those embedded engineers more productive. Watch whether Deployment's unit economics work; if they do, you'll see a wave of similar services companies, which could either validate your market or cannibalize it depending on what you're selling.
CoreWeave logo
🇺🇸CoreWeaveAI Infrastructure

CoreWeave provides GPU cloud infrastructure for AI workloads, enabling developers and enterprises to train and deploy large language models at scale.

$2.6BStrategic
Investor undisclosed
A $2.6B strategic round for GPU cloud infrastructure signals that compute scarcity for AI training/inference is real enough to justify massive capital—this isn't hype, it's buyers voting with money. CoreWeave likely uses this to expand capacity and lock in long-term supply agreements before the next wave of model scaling. If you're building any AI application layer, watch whether CoreWeave's pricing/availability becomes a constraint on your unit economics; infrastructure costs are about to be a real competitive moat.
国奥科技 logo
🇨🇳

国奥科技

Robotics

国奥科技 builds micrometer-precision embodied robots for manufacturing using self-developed technology.

$14MSeries A+
Investor undisclosed
A $14M Series A+ for precision manufacturing robotics out of China signals the category is moving past research into deployment—this isn't VC betting on breakthroughs, it's capital flowing to companies with working systems. If you're building in adjacent hardware (vision systems, motion control, factory software), watch whether 国奥科技 gets design wins with Tier 1 manufacturers; that's the real signal for whether precision robotics becomes a defensible moat or a race-to-the-bottom commodity.
蚂蚁灵波 logo
🇨🇳

蚂蚁灵波

AI

Ant Lingbo builds AI-powered financial intelligence and risk management solutions for enterprises and financial institutions.

$210MSeries A
Investor undisclosed
A $210M Series A in China for enterprise financial AI signals that LLM-powered risk/compliance tooling has moved past proof-of-concept—banks are now willing to deploy at scale. If you're building B2B AI in regulated verticals (healthcare, insurance, supply chain), watch how Lingbo handles the compliance-to-product tension; that playbook will matter more than the funding size itself.
Lancium logo
🇺🇸LanciumData Center Infrastructure

Lancium builds AI-optimized data center infrastructure for compute-intensive workloads.

$3BStrategic
Nvidia logo
Nvidia
Nvidia writing a $3B strategic check to Lancium signals they're betting hard on dedicated inference infrastructure outside their own walls—likely because hyperscalers are building private capacity and Nvidia needs distribution partners who can lock in long-term GPU commitments. If you're building any AI application that needs predictable, high-throughput compute (not just training), this validates that specialized data center operators are becoming the new moat, not cloud generalists.
Ommo Technologies logo
Ommo TechnologiesRobotics

Ommo Technologies builds spatial intuition software for robots to navigate and understand their environment.

UndisclosedSeries A
Investor undisclosed
WindBorne Systems logo
🇺🇸WindBorne SystemsWeather ForecastingVerified

WindBorne collects weather data via long-flying balloons and sensors, then uses AI to generate accurate forecasts.

$37MSeries B
Khosla Ventures logo
Khosla Ventures
GTL
A $37M Series B for weather data collection signals that enterprise customers (energy, agriculture, logistics) are willing to pay for hyperlocal forecasts—the market's moved past 'nice to have' to 'operationally critical.' If you're building anything that depends on weather inputs or real-time environmental data, this validates that there's budget to integrate your product into their workflows, not just sell them a dashboard.
Omilia logo
OmiliaConversational AI

Omilia builds conversational AI platforms for enterprise customer service teams to automate and enhance voice and chat interactions.

$67MSeries B
Investor undisclosed
A $67M Series B for conversational AI in customer service signals that enterprises are finally moving past chatbot pilots—they're committing real budget to replace human-heavy support tiers. At this stage and size, Omilia is likely burning cash on sales infrastructure and model fine-tuning for vertical-specific use cases (banking, telecom, etc.), not just product. If you're building any workflow automation for support teams, watch how they're positioning ROI to CFOs—that's your playbook for selling to the same buyer.
PokeBot logo
🇨🇳PokeBotRobotics

PokeBot builds embodied AI robots for household tasks like clothes folding and cooking.

UndisclosedPre-A
S
Shunwei Capital
Matrix Partners logo
AI Infrastructure Capital logo
AI Infrastructure CapitalAI Infrastructure

AI Infrastructure Capital provides funding and support for AI infrastructure companies building the foundational tools and systems powering AI development.

$17.4M
Investor undisclosed
A $17.4M fund for AI infrastructure in mid-2026 signals that LP capital is still flowing into foundational tooling, but the bar has shifted—this isn't about generic compute anymore, it's about solving specific bottlenecks (inference optimization, data pipelines, observability). If you're building in adjacent infrastructure (monitoring, data ops, or ML platforms), watch what portfolio companies they back; that's your early signal for which problems are actually getting solved vs. which ones are still unsolved.
Unitree logo
🇨🇳UnitreeRobotics

Unitree builds humanoid robots powered by physical AI for industrial and commercial applications.

$20.8MIPO
Investor undisclosed
A Chinese robotics company going public at $20.8M signals the market still sees humanoid robots as a near-term commercial play, not just research—but the modest valuation suggests investors are pricing in real execution risk and regulatory uncertainty. If you're building in embodied AI, logistics automation, or hardware-software stacks, watch whether Unitree's post-IPO burn rate and customer concentration reveal whether the unit economics actually work at scale, or if this is still a capital-intensive bet on future demand.
Firmus logo
🇦🇺FirmusAI Infrastructure

Firmus builds AI infrastructure for enterprises to deploy and manage large language models efficiently.

$20B
Investor undisclosed
A $20B round with NVIDIA and Blackstone signals that enterprise LLM deployment—not just model training—is now a capital-intensive, infrastructure-grade problem worth betting on at scale. Firmus is likely burning this on GPU procurement, data center partnerships, and sales motion to lock in Fortune 500 customers before the market consolidates. If you're building any layer of the AI stack (observability, fine-tuning, RAG), watch how Firmus prices compute access—it'll set the floor for what enterprises expect to pay, which directly impacts your unit economics.
Awomo logo
🇨🇳AwomoRobotics, World Models, Autonomous Driving

Awomo builds world models for physical AI systems using composable latent-space concepts instead of pixel rendering.

$140KSeed
Investor undisclosed
A $100K seed from mostly China-based funds suggests world models for robotics are still pre-product in most geographies—this is pre-PMF capital, not validation of a hot category. The latent-space approach (avoiding pixel rendering) is technically interesting but the tiny round size signals either very early stage or that investors see this as a research bet rather than near-term revenue play. If you're building embodied AI or sim-to-real systems, watch whether Awomo can actually reduce compute vs. pixel-based baselines in practice—that's the real moat, not the concept.
幺正量子 logo
🇨🇳

幺正量子

Quantum Computing

幺正量子 develops quantum computing solutions for enterprise applications.

UndisclosedSeries A
Investor undisclosed
真觉万象 logo
🇨🇳

真觉万象

Embodied AI

真觉万象 builds data infrastructure for embodied AI systems to enable robots and autonomous agents.

Undisclosed
Investor undisclosed
DeepSeek logo
🇨🇳DeepSeekLLM

DeepSeek builds large language models for developers and enterprises using advanced AI research.

Undisclosed
Investor undisclosed
Wordsmith logo
🇺🇰WordsmithLegal AI

Wordsmith automates legal workflows for in-house teams using AI agents to handle routine requests across multiple channels.

$14MSeries B Extension
Investor undisclosed
Legal AI is moving past document review into workflow automation—this extension signals investors still see room to scale before the category consolidates. At $14M for a Series B extension, Wordsmith is likely doubling down on multi-channel integration and compliance depth rather than land-grab expansion, which means the real moat is operational stickiness, not feature breadth. If you're building in regulated verticals (healthcare, finance, insurance ops), watch how they're solving the channel fragmentation problem—that playbook transfers directly.
Moss logo
🇪🇺MossAI Finance

Moss builds an AI-powered finance platform delivering automated financial services and solutions for businesses.

$32.7MSeries C
Investor undisclosed
A $32.7M Series C for an EU fintech AI platform signals that automated financial operations (likely accounting, reconciliation, expense management) are moving past the "nice-to-have" phase into must-have infrastructure—especially in markets with fragmented compliance. If you're building B2B SaaS that touches financial workflows, this validates that buyers will pay for AI that reduces manual finance work, not just reports on it.
Kaiwang Data logo
🇨🇳Kaiwang DataData Infrastructure

Kaiwang Data provides data infrastructure for embodied AI, collecting and labeling multimodal datasets for autonomous driving and robotics.

$140KStrategic
B
Beijing E-Town Industrial Upgrade Fund
HSL
+3
A $100K strategic round from a mix of industrial funds and robotics companies signals China is quietly consolidating embodied AI infrastructure—this isn't Series A momentum, it's strategic players locking in data supply chains before the category gets expensive. If you're building robotics, autonomous systems, or world models anywhere, watch who's investing in data labeling infrastructure in your region; that's where the real bottleneck will be in 18 months.
Moonshot AI logo
🇨🇳Moonshot AILLM Infra

Moonshot AI builds Kimi Operator, an AI assistant platform powered by large language models for enterprise and consumer applications.

UndisclosedPre-IPO
Investor undisclosed
Zhigu Tianchu logo
🇨🇳Zhigu TianchuRobotics

Zhigu Tianchu builds AI-powered cooking robots for commercial kitchens and food service operations.

$140KStrategic
C
China Merchants Venture Capital
A $100K strategic check from a corporate VC suggests Zhigu Tianchu is still pre-product or very early revenue—this isn't validation money, it's relationship capital from someone who might become a customer or acquirer. The cooking robot space in China is crowded and capital-light relative to Western robotics, so watch whether they're solving a real labor shortage problem or just riding hype; if they crack unit economics in food service, that playbook ports to other high-turnover operations.
Zanskar logo
🇮🇳ZanskarPain Management

Zanskar delivers personalized pain management through an app-guided at-home care protocol paired with branded relief products.

$6MSeries A
Bessemer Venture Partners logo
Bessemer Venture Partners
India's consumer health market is now attracting tier-1 US VCs for non-acute categories—Bessemer backing a pain management app signals that personalized wellness (not just diagnostics) can scale in price-sensitive markets. If you're building habit-forming health tools, the playbook here is bundling digital + physical products to justify repeat engagement and defensible unit economics.
Sarvam logo
🇮🇳SarvamLLM Infra

Sarvam builds AI models and infrastructure for Indic language processing across text, speech, vision and documents.

$74MSeries B
NVIDIA logo
NVIDIA
GGI
+2
NVIDIA leading a $74M Series B for Indic language AI signals they're serious about non-English LLM infrastructure as a defensible moat—this isn't charity, it's supply-chain control. Sarvam likely uses this to scale model training, build proprietary datasets across Indian languages, and lock in enterprise customers before the obvious players (Meta, Google) ship their own. If you're building any vertical SaaS for India or Southeast Asia, watch whether Sarvam becomes the de facto inference layer; if so, your moat just got thinner.
Wordsmith logo
🇺🇰WordsmithLegal AI

Wordsmith automates legal workflows for in-house teams using AI agents to handle routine requests across multiple channels.

$14MSeries B
Index Ventures logo
Index Ventures
F
Legal AI is moving past document review into workflow automation—this $14M Series B signals that in-house teams (not just law firms) are now the beachhead for AI agents handling repetitive work. If you're building enterprise automation in any regulated vertical, watch how Wordsmith handles multi-channel request routing and compliance; that's the hard part VCs are now funding.
June logo
🇺🇸JuneEnterprise AI / AI Deployment

June scans enterprise systems to identify bottlenecks and deploys AI agents to automate business processes.

$20MPre-Seed
T
Time Ventures
MAG
A $20M pre-seed from Dell/Box/CrowdStrike's angels signals that enterprise AI deployment—not just model building—is now fundable at scale. June's bottleneck-scanning + agent-execution play suggests investors believe the real money is in the unglamorous work of actually *running* AI in legacy systems, not selling models. If you're building workflow automation or observability tools, this validates that enterprises will pay for anything that reduces the friction between 'we have AI' and 'it actually works in production.'
Kily logo
🇮🇳KilyAgentic AI

Kily builds autonomous AI agents for consumer brands to automate ecommerce and quick commerce operations like advertising, pricing, and inventory.

$0Series A
S
Sorin Investments
RW
A $0M Series A is either a data error or a pre-announcement—either way, not enough signal to read. If this is real and sub-$5M, it suggests Indian quick commerce is still hunting for unit economics wins rather than throwing capital at agentic AI; if it's a larger round that hasn't been disclosed yet, watch whether Razorpay's involvement means payment infrastructure is becoming table stakes for commerce automation. Either way, if you're building ops automation for any vertical, the fact that someone's betting on autonomous agents for ecommerce pricing/inventory means the margin-compression playbook is moving from logistics to decision-making.
X Mile logo
🇯🇵X MileWorkforce Management, Logistics & Operations AI

X Mile builds AI-powered workforce management and recruitment software for logistics, construction, and manufacturing companies in Japan.

$21.4MSeries C
V
Vertex Growth
ASS
+5
A $21.4M Series C for Japanese workforce logistics software signals that labor scarcity in manufacturing/construction is now a venture-scale problem in Asia—not just a compliance headache. X Mile is likely using this to expand beyond Japan and build out AI matching/scheduling (the hard part), which means if you're building ops software in tight labor markets anywhere, the playbook of "AI + local regulatory expertise" just got validated with real capital. Watch how they handle cross-border expansion; that's the real test of whether this model scales beyond Japan's specific constraints.
术也科技 logo
🇨🇳

术也科技

Physical AI

术也科技 builds physical AI systems for laboratory automation to streamline experimental workflows and equipment operation.

UndisclosedPre-A
Investor undisclosed
Obsidian Security logo
🇺🇸Obsidian SecurityAI Security

Obsidian Security builds AI-powered threat detection and response tools for enterprise security teams.

Undisclosed
Investor undisclosed
Fangqing Technology logo
🇨🇳Fangqing TechnologyAI Chip

Fangqing Technology designs AI chips for accelerated computing workloads.

UndisclosedSeries A1
Investor undisclosed
HappyRobot logo
🇪🇸HappyRobotEnterprise AI AgentsVerified

HappyRobot enables enterprises to build, deploy, and manage AI agents that automate complex workflows.

$150MSeries C
P
Prysm Capital
EA16z logoBase10 logo
+4
A $150M Series C for enterprise AI agents from a Spanish startup signals that workflow automation is moving past proof-of-concept—enterprises are ready to deploy at scale, and investors believe the defensibility is in orchestration, not just LLMs. If you're building in adjacent automation spaces (RPA, no-code platforms, vertical SaaS), this validates that buyers will pay for agents that actually integrate with their existing systems rather than point solutions.
风火轮萤图 logo
🇨🇳

风火轮萤图

Embodied AI

风火轮萤图 builds embodied AI systems with active interaction capabilities for robotics and autonomous applications.

$4.2MSeed+
Investor undisclosed
A $4.2M seed+ for embodied AI in China signals the market is moving past simulation—investors want teams tackling real-world robot control, not just foundation models. This round size suggests they're funding hardware integration and real-world data collection, which is the actual moat in embodied AI. If you're building in robotics, autonomous systems, or even sim-to-real tooling, watch whether Chinese teams start shipping faster than US competitors on the same problem.
PaXini logo
🇨🇳PaXiniRobotics

PaXini builds tactile-sensing chips and technology for robots to perceive and interact with their environment.

$14MStrategic
B
BOC International Investment
KHD
+4
Chinese VCs are betting hard on tactile sensing as table stakes for robot dexterity—this $14M strategic round signals the hardware perception layer is moving from research to commercialization. If you're building robot software, manipulation stacks, or sim-to-real tools, watch whether PaXini's chips become the de facto standard; if they do, you're building on top of their sensor data, not around it.
昉擎科技 logo
🇨🇳

昉擎科技

Chip Design

昉擎科技 designs custom semiconductors and systems for AI and computing applications.

UndisclosedSeries A1
徐
徐汇资本
珠中国
+9
A Series A1 with this many Chinese institutional co-investors (徐汇资本 leading, plus state-backed 珠海科技产业集团) signals serious domestic capital backing for what's likely a deep-tech or industrial play—not a consumer app. If you're building B2B infrastructure or hardware in China, this round size and investor mix tells you the bar for follow-on funding is now higher; you'll need to show unit economics or defensible IP, not just user growth.