Rhodium测算中国AI模型总营收为OpenAI、Anthropic总和10%

This Rhodium AI industry report helps general readers quickly grasp core updates on the development of the AI sectors in China and the U.S., and clarify the real state of the industry.

1. Core revenue comparison: Calculated on an annual recurring revenue (ARR) basis, the total revenue of all Chinese AI model providers equals only 10% of the combined revenue of the two leading U.S. AI firms, OpenAI and Anthropic. OpenAI posts an ARR of $40 billion, while Anthropic reaches $65 billion. Among top Chinese AI players, ByteDance reports $4 billion in ARR and Alibaba $2.4 billion, with other domestic AI startups generating far lower revenue.

2. Industry dynamics and risk alerts: Domestic AI startups generally have higher valuation-to-revenue multiples than leading U.S. peers. A number of these firms are reportedly preparing for public listings, and Z.ai has raised its revenue target for the end of 2026. At present, Chinese model companies have limited revenue scale and rely on the equity market, which lacks stable funding support. State capital is mainly directed to computing hardware, and recently listed AI firms have seen sharp share price volatility.

This report on China-U.S. AI industry revenue provides clear decision-making references for brands planning AI-related business initiatives and selecting technology partners.

1. Tool selection and cost benchmarking: Leading U.S. AI models are mostly closed-source, with per-task inference costs far higher than those of Chinese models. Most domestic AI models are open-source, allowing third parties to download and run models independently with sufficient hardware. Brands can select cost-effective domestic model services based on their own business needs to lower the implementation cost of AI applications.

2. Industry trends and partnership risk alerts: Usage of domestic AI models has already grown sharply. Going forward, Chinese AI labs will explore higher revenue share models for third-party model invocation scenarios, so brands need to anticipate cost changes for related services in advance. At this stage, domestic AI model firms are generally small in revenue, carry high valuations, and depend on equity financing for growth. When selecting partners, brands should prioritize evaluating operational stability to avoid business disruption risks.

This AI industry revenue report helps sellers identify AI-related business opportunities and reasonably mitigate market and partnership risks.

1. Clear efficiency improvement opportunities: Usage of domestic AI models is growing rapidly. Per-task inference costs for Chinese models are far lower than for U.S. closed-source models, and most support open-source deployment. Sellers can integrate AI capabilities at low cost to improve operational efficiency.

2. Operational considerations and risk alerts: Domestic AI labs are exploring higher revenue share structures for third-party model invocation. If sellers build AI-derived businesses, they should monitor changes to revenue sharing rules early and adjust their profit models accordingly. At present, leading domestic AI model firms have limited overall revenue, high valuations, and rely on the less stable equity market for development. When selecting AI tools, sellers should assess service provider operational stability, and guard against investment losses caused by capital market volatility in the AI sector.

This China-U.S. AI industry revenue measurement report provides clear directional references for factories advancing digital transformation and identifying AI-related supporting production opportunities.

1. Digital transformation pathway guidance: Most domestic AI models are open-source, with per-task inference costs far lower than U.S. closed-source models. Factories can download and run models independently with sufficient hardware, accessing AI capabilities at low cost to advance digital upgrades on the production side without heavy spending.

2. Business opportunities and risk alerts: Over 60% of state-backed equity investment in the AI sector flows to computing hardware such as chips and servers. Demand for hardware manufacturing and supporting processing related to computing infrastructure is backed by ample funding. Combined with the sharp growth in domestic AI model usage, demand for related hardware components will continue to rise. When providing production supporting services to AI firms, factories should evaluate the operational stability of model company clients to prevent payment collection risks.

This AI industry revenue report clearly outlines current development trends and core pain points in the AI sector, providing a decision-making basis for AI-related service providers to adjust business direction and optimize service offerings.

1. Clear secular industry opportunities: Domestic AI model usage has entered a phase of rapid growth. Models are generally open-source with low invocation costs, and Chinese AI labs are exploring higher revenue share models for third-party invocation scenarios. Service providers can build offerings around open-source model deployment and implementation of invocation revenue sharing rules to match market demand.

2. Clear client pain points and potential service directions: Leading domestic AI labs currently have insufficient revenue to support sustainable scaled expansion, rely heavily on the relatively unstable equity market, and receive limited direct support from state capital, which is mainly directed to computing hardware. Service providers can design targeted services addressing AI labs' financing needs and commercial monetization requirements. In addition, leading U.S. firms have already warned about risks from overly rapid AI technology development, making related risk mitigation supporting services a common industry need going forward.

This AI industry revenue measurement report provides references for platform companies to understand AI sector development patterns, optimize merchant acquisition and operation strategies, and proactively mitigate operational risks.

1. Ecosystem building and merchant acquisition guidance: Domestic AI model usage is growing rapidly. Most models are open-source with low invocation costs, and model firms are exploring revenue sharing pathways for third-party invocation. Platforms can target AI model companies and AI application developers for onboarding, build supporting systems for model invocation, transaction processing and revenue splitting, and improve their AI service ecosystems. Platforms can also integrate cost-effective domestic model capabilities to provide AI services to business operators within their ecosystems, expanding platform service revenue.

2. Operational risk mitigation references: Domestic AI startup model companies generally have high price-to-sales ratios, listed AI firms have seen significant share price volatility, and model companies rely heavily on the unstable equity market for development. When onboarding and partnering with relevant entities, platforms should conduct thorough assessments of business qualifications and operational stability, monitor industry knock-on effects from capital market volatility and AI technology risk warnings, and prepare risk response plans in advance.

This AI industry revenue measurement report released by Rhodium Group outlines multiple new trends and practical issues in the current development of China and U.S. AI sectors, with high value for industry research.

1. Core empirical industry data: Calculated on an ARR basis, total revenue of all Chinese AI model providers equals only 10% of the combined revenue of two leading U.S. firms, OpenAI and Anthropic. Domestic AI startups have significantly higher price-to-sales ratios than leading U.S. peers, and over 60% of state-backed equity investment in the AI sector goes to computing hardware including chips and servers. These data can serve as foundational material for research on China-U.S. AI competitiveness comparisons and industrial structure.

2. Research direction implications: Leading Chinese AI labs face the practical constraint of insufficient revenue to support sustainable scaled expansion, rely heavily on the relatively unstable equity market, and do not receive direct government funding for model R&D. Meanwhile, domestic models are predominantly open-source, and the industry is exploring new business models around revenue sharing for third-party invocation scenarios, a development pathway with strong research value. Differences between Chinese and U.S. firms in AI technology risk response and capital market performance also represent viable directions for future segmented research.