Market map · as of 2026-10-13

AI Agent Market Map

159 AI agent players from the US, Europe, China, Korea and Japan in a single table. Filter by country, type, scale tier, customer, pricing model and ownership; sort by users, ARR or country. Every figure carries the verification status from the source data (✔ primary source · filing / △ press report · estimate / ✘ not the case). Most are △.

Research as of 2026-10-13Last updated 2026-10-15Marks: ✔ primary source or filing / △ press report or estimate / ✘ not the caseReport an error

The industry in six numbers

$7.6B ↔ $201.9BThe “AI agent market” — standalone products (2025 · three research firms △) ↔ spending embedded in enterprise software (2026 · Gartner △). About 27x apart on definition alone

1.2BChatGPT weekly users (2026-09-29 · △ single source). China’s No. 1 app, 豆包 Doubao, has 382M MAU (QuestMobile 2026-06 △)

$24.2BAgentic AI venture investment in 2025 · 1,311 deals (PitchBook △)

$0.10 · $0.99One Agentforce action (✔ · formerly $2 per conversation) · one Fin resolution (△) — pricing shifts to actions and outcomes

$60BSpaceX’s acquisition of Cursor (agreed 2026-06-16 · △) — the largest startup acquisition ever

40%+Share of agent projects expected to be canceled by end of 2027 (Gartner 2025-06 · △)

Top standalone agent companies by ARR · revenue USD millions · excludes model companies OpenAI (about 70,000) and Anthropic (65,000+) · timing and definitions vary by company

Splitting “AI agents” into eight types

One label, but different buyers and different ways of paying. A company that spans several types (e.g. Mistral = model company + general-purpose + coding) carries every applicable type number, and the type filter shows it if any one matches.

Extra filters and why · Scale tier: metrics differ by company (weekly or monthly users · app MAU · ARR), so the numbers can’t be compared directly. Each headline metric is grouped into S (100M+ users or ARR $10B+) · A (10M+ / $1B+) · B (1M+ / $100M+) · C (100K+ / $10M+) · D (below that) · ? (unconfirmed). Where both users and ARR exist, the higher tier applies. · Pricing model: added to track money moving from seats → usage and credits → outcome-based pricing (classified automatically from the pricing column of the source data · approximate). · Customers · ownership: B2C · B2B · B2G · developers, and listed companies · conglomerate affiliates · startups. Korea and Japan are dominated by conglomerate affiliates and SI firms; among newcomers in the US, Europe and China, startups lead. · Flags: “Local usage (foreign service)” — the most-used agents in Korea and Japan are US services, so Korean and Japanese figures for ChatGPT, Gemini and Perplexity are listed separately. “Acquisition · merger”, “Shut down · discontinued” and “Started 2023 or later” show how the industry is consolidating.

Players

Country

Type

Scale tier

Metric type

Customers

Pricing model

Ownership

Flags

Sort

#	Name	Country	Type	Tier	Headline metric	Pricing	Check

Click a row to expand the company, people, pricing as listed, revenue, funding and recent developments. The 89 entries marked “? Unconfirmed” are those for which this 15-minute research pass found no comparable figure (conglomerate affiliates and deployment vendors do not report agent revenue separately).

Market size — read each definition on its own

Scope	Value	Definition · source	Check

Business models — 10 ways the money comes in

Model	Actual prices · examples	Trend

Fastest to ARR $100M

Company	Region	Time to $100M	Since then

All △ press reports. Whether the clock starts “from launch” or “from founding” varies by company. For reference, Bessemer’s “Supernova” benchmark is $40M in year one.

People moving the industry

Name	Country	Company · title	Why they matter

12 industry trends (2025–2026)

Five standouts, one per region

Chosen so that their business models don’t overlap: outcome-based · credit-based app builder · general-purpose credits + overseas revenue · free + ads + characters · AI + human outsourced operations.