Skip to content

The Global AI Unicorn Panorama: 30 Ten-Billion-Dollar Companies, With Chinese Firms Breaking Into the Top 10

Nov 17, 07:55
The Global AI Unicorn Panorama: 30 Ten-Billion-Dollar Companies, With Chinese Firms Breaking Into the Top 10

By Xiaojing, Tencent Technology

On November 15, Deedy Das (@deedydas), a partner at veteran Silicon Valley venture firm Menlo Ventures, posted a ranking of global AI companies by valuation and revenue on social media.

Founded in 1976, Menlo Ventures is one of the oldest venture capital firms in Silicon Valley and has invested in well-known tech companies such as Uber and Chime. In recent years the firm has made AI a core investment theme, and at the end of 2024 it published the widely circulated "State of Generative AI in the Enterprise" report. Menlo Ventures is also an early investor in Anthropic, having joined its Series C round in 2023, and in 2024 it co-established the $100 million AI investment fund Anthology Fund with Anthropic.

The list covers private AI companies that, as of November 14, had a valuation above $10 billion and publicly disclosed annualized revenue of more than $100 million. Thirty companies already meet both conditions.

The "duopoly" at the top has widened its lead further: OpenAI continues to set the pace with a valuation of about $1.25 trillion, followed by Anthropic at about $965 billion, and the two have opened a trillion-dollar-scale valuation gulf with other competitors.

But when ranked by annualized revenue, the two have swapped places: Anthropic has surpassed OpenAI. The pair have pulled away from rivals on valuation, yet the revenue race has taken a new turn.

Among Chinese companies, DeepSeek and Moonshot AI entered the global top 10. DeepSeek ranks fifth with an $89 billion valuation, and Moonshot AI ranks seventh at $50 billion. According to the data disclosed in the list, both companies have annualized revenue of about $1 billion.

Still, the United States retains an absolute advantage across the list. Of the 30 companies, 23 are from the US, spanning foundation models, AI coding, compute cloud services and other fields. China has 2 entries, and the remaining 5 come from Canada, France, Germany and the UK.

Chart: Ranking of global private AI companies valued above $10 billion

The "duopoly" leads by a wide margin

Rewind a year, and OpenAI was still "far out in front."

At the end of 2025, OpenAI's annualized revenue had already reached about $20 billion, while Anthropic had only $9 billion. At that time there was still a more-than-twofold revenue gap between the two.

But entering 2026, Anthropic's growth began to clearly outpace OpenAI. As of the end of September this year, OpenAI's annualized revenue was about $50 billion; by the end of October, Anthropic's annualized revenue had reached $65 billion, overtaking OpenAI.

In terms of actual revenue, the shift is equally visible. In the third quarter of this year, Anthropic generated about $17 billion in revenue, versus about $10 billion for OpenAI. Although the two still differ in how they calculate cloud platform revenue — for example, whether Anthropic includes some hosted revenue on Amazon Bedrock and Google Cloud — its rapid penetration of the enterprise market is already directly reflected in its revenue growth curve.

The two companies' leading positions remain solid. Competition at the top of the global foundation model market has entered the trillion-dollar valuation stage.

More notable is that Anthropic is riding this growth momentum toward an IPO. On November 10, Anthropic announced an IPO price range of $70 to $78 per share and is expected to list on Nasdaq on November 19 under the ticker "ANT." Calculated at the top of the range, its fully diluted valuation would reach $965 billion, making it one of the largest tech IPOs in the world this year.

OpenAI has made a different choice. It has said it will not go public in 2026 and has pushed its IPO plan back to 2027. On November 6, OpenAI closed the first tranche of about $11 billion in a new funding round at a post-money valuation of $1.25 trillion, led by SoftBank with participation from Microsoft, Coatue and Thrive Capital. The funding not only set a record for a single round raised by a global startup, but also temporarily eased the financial pressure of OpenAI's compute spending.

A year ago, OpenAI still held a clear advantage in the model market, user scale and commercial revenue. Today, Anthropic has been first to complete the overtake on revenue growth, and is preparing to beat it to an IPO.

Chinese companies break through

Chinese AI companies have also made it onto this list of the world's highest-valued AI companies.

DeepSeek ranks fifth with an $89 billion valuation and about $1 billion in revenue; Moonshot AI ranks seventh at a $50 billion valuation, with revenue also about $1 billion. Both companies made the top 10, and they are the only two Chinese companies on the list.

Although DeepSeek and Moonshot AI both have annualized revenue of about $1 billion, their valuations imply dynamic price-to-earnings ratios of 89x and 50x respectively, meaning the capital market's pricing of the two does not depend on current revenue alone, but also takes into account their technology roadmaps, commercialization paths and future profit potential.

Beyond US and Chinese companies, Canada's Cohere is valued at $25.5 billion with annualized revenue of about $240 million; France's Mistral is valued at $24 billion with annualized revenue of more than $400 million.

Germany's defense AI company Helsing ranks 20th with an $18 billion valuation and annualized revenue of about $500 million; UK compute infrastructure provider Nscale ranks 28th with a $14.6 billion valuation and annualized revenue of about $280 million.

Super-sized M&A begins

Two companies highlighted in gray on the list draw particular attention: xAI and Cursor.

The two correspond to deal valuations of $250 billion and $60 billion respectively, and both are marked as being folded into the SpaceX system. Among them, xAI's $250 billion deal valuation includes the value of social platform X, but the $500 million in revenue listed does not include X's advertising revenue. For xAI, being folded into SpaceX means it can leverage the latter's vast satellite internet and energy infrastructure to build larger-scale data center clusters.

The other acquired company, Cursor, is a fast-growing AI coding tool company in recent years. Its deal valuation reaches $60 billion, with revenue of about $2.6 billion. Based on the list's data, Cursor's revenue scale already exceeds that of many compute service providers and vertical application companies. Leveraging the extremely high penetration of AI coding among developers, it has quickly converted users into high-frequency paying customers.

These two deals involve foundation models and AI application products respectively, showing that leading companies are expanding their business boundaries through capital consolidation. The links between model R&D, application products and commercialization channels are being further strengthened, and competition in the AI industry is beginning to evolve from a single technological edge toward ecosystem scale.

Compute and data infrastructure flourish

Although foundation model companies occupy the top of the valuation list, another group of companies deserves attention: the infrastructure companies that provide compute, data and inference services to the entire AI industry. As demand for model training and application deployment grows, these companies have been the first to prove out a closed commercial loop.

Companies that directly provide compute support

Crusoe focuses on AI data center construction and cloud computing services, with a valuation of $30.9 billion and revenue of about $2 billion. Fluidstack provides computing capacity to AI companies through GPU compute resources and cloud services, with a valuation of $18 billion and revenue also about $2 billion. CoreWeave ranks 14th with a $23 billion valuation and annualized revenue of $3.2 billion.

The business of such companies has similarities with traditional cloud computing vendors, but they build infrastructure around AI training and inference needs, requiring sustained heavy investment in buying GPUs, building data centers and securing long-term customer contracts.

Data infrastructure service providers

VAST Data is valued at $30 billion with revenue of about $2 billion, mainly providing large-scale data storage and management services to enterprises. Scale AI is valued at $29 billion with 2025 revenue of about $1 billion, focusing on AI data labeling, model evaluation and high-quality synthetic data generation.

As AI models keep advancing toward complex reasoning, multimodality and agents, the quality of training data, the efficiency of data processing and the importance of model evaluation are all rising. VAST Data and Scale AI have entered from data storage and data processing respectively, becoming indispensable service providers for large model training and enterprise deployment.

Fast-growing model inference service platforms

Baseten is valued at $30 billion with revenue of about $600 million; Fireworks AI is valued at $17.5 billion with revenue of about $1 billion; Modal is valued at $15.75 billion with revenue of about $1 billion; Together AI is valued at $15 billion with revenue of about $840 million; fal is valued at $14 billion with revenue of about $1 billion.

These companies mainly help developers and enterprises deploy and run AI models, reducing inference costs and improving operating efficiency. Among them, fal places particular emphasis on inference services for generative AI models such as image and video.

Compared with large model companies, these infrastructure companies can serve multiple model and application vendors at the same time and can also support different technology roadmaps. As more AI applications enter actual operations, model call volumes, compute consumption and data processing needs keep expanding, allowing infrastructure companies' revenue to scale rapidly.

That said, this does not mean infrastructure is necessarily a more robust business. Compute service providers face high capital expenditure, while inference service platforms must cope with continuously falling model call prices and competition from large cloud vendors. Whether demand growth can ultimately translate into healthy profit margins remains a question these companies must answer.

Divergence in the application layer intensifies

Compared with foundation models and compute infrastructure, the AI application layer presents a different competitive landscape: although valuations are generally lower than those of leading model companies, the revenue growth and valuation multiples of some application companies are already quite striking.

High-frequency scenarios scale first

Among them, the first to reach scale are high-frequency applications such as AI coding, voice and search.

Cognition, through its AI coding agent Devin, has reached a $48 billion valuation with annualized revenue of about $1 billion; ElevenLabs has won a $22 billion valuation on the strength of its voice generation and voice agent businesses, with revenue of about $330 million; AI search engine Perplexity is valued at $18 billion with revenue of about $200 million.

Founded in 2024, lovable focuses on letting ordinary people with no coding experience develop software through natural language. Its valuation has already reached $13.3 billion, with annualized revenue of about $600 million.

Vertical professional fields build deep moats

The more noteworthy changes are happening in vertical industries such as legal, healthcare and enterprise services.

Legal AI company Harvey is valued at $15.5 billion with annualized revenue of more than $400 million. It has already been adopted by 80% of the top 100 large law firms in the United States, helping lawyers with professional work such as contract review, legal research and due diligence.

Enterprise customer service agent company Sierra has reached a valuation of $15.8 billion. Its business model is also highly representative: customers can pay according to the problems the AI actually solves — for example, successfully handling a single after-sales inquiry — rather than simply paying for software seats.

Healthcare AI platform OpenEvidence provides clinical research and medical information retrieval services free to certified physicians, earning revenue through advertising aimed at the pharmaceutical industry. According to Sacra's estimate, its annualized revenue has reached $130 million, with a valuation of about $10.5 billion.

What these companies have in common is that they are beginning to enter work steps previously performed by professionals such as lawyers, doctors and customer service staff. Rather than simply offering an AI chat tool, they need to understand industry rules, connect to enterprises' existing systems and even redesign business processes. This not only gives them clear willingness to pay, but also makes it easier for them to build competitive moats in vertical scenarios.

This also makes the valuation logic of the application layer more interesting. On a simple calculation based on the list, some vertical AI application companies are valued at dozens of times their annualized revenue, even higher than many compute infrastructure companies. What investors value is no longer just the sales scale of the software itself, but also its long-term room to replace traditional human services in a specific industry.

This trend also appears in other research. On November 11, investment firm Leonis Capital's "2026 AI 100" report showed that vertical industry AI companies on the list increased from 28 last year to 43, and the focus of enterprise AI services is accelerating its penetration from general-purpose tools into vertical industries.

The AI application market is diverging at an accelerating pace. As foundation model capabilities keep improving, the functional advantages of general-purpose AI applications are increasingly easy to replicate, and products that merely wrap model capabilities will find it hard to sustain high valuations over the long term.

By contrast, the capital market is starting to favor companies rooted in vertical industries with unique data and professional moats. They often need long-term accumulation of industry experience, customer trust and business processes, and once they penetrate core operations, customers' switching costs are also higher.

Amid the nightmare of models devouring everything, this may be exactly where the opportunity lies for AI startups at the application layer.

Join the official Coincamps community:

X: https://x.com/coincamps

Telegram: https://t.me/coin_camps

Recommended