SemiAnalysis: Anthropic's third-quarter profit will exceed 1 billion dollars

Author: Xu Chao

The latest analysis from research firm SemiAnalysis reveals that Anthropic is reshaping the AI commercialization landscape with profitability and growth rates far exceeding its competitors. With a high-margin business model centered around APIs, Anthropic has become a leader in the B2B AI market.

According to a deep report released by SemiAnalysis, Anthropic is expected to achieve $1 billion in GAAP EBITDA by the third quarter of 2026, corresponding to a profit margin of about 6%. Meanwhile, its annual recurring revenue (ARR) has skyrocketed from $9 billion at the end of 2025 to over $60 billion currently. The firm predicts that if Anthropic maintains a net new ARR (NNARR) pace of about $15 billion per month, its ARR by the end of 2027 could reach $300 billion, corresponding to a $6 trillion enterprise value, making it the highest-valued company in the world.

Anthropic secretly submitted its IPO application on June 1. SemiAnalysis believes that going public at this time is of strategic urgency—Alphabet has completed $84.75 billion in equity financing, and Meta has also been rumored to have financing plans worth tens of billions, indicating that the capital market window is narrowing. The report points out that Anthropic's superior financial data and business model mean it should go public ahead of OpenAI to seize the initiative in capital competition.

Claude Code ignites the B2B market, with quarterly ARR growth exceeding twofold

The turning point for Anthropic's performance stems from the explosive popularity of Claude Code. Data from SemiAnalysis shows that Claude Code currently accounts for over 7% of all code submissions on GitHub, directly driving the company's ARR from a monthly addition of $3 billion in January to $11 billion in March.

In terms of revenue structure, Anthropic shows significant differentiation from OpenAI. Approximately 75% to 85% of Anthropic's ARR comes from usage-based API business, while consumer subscriptions account for only 5% of total ARR. In contrast, OpenAI still derives over 65% of its revenue from subscription models in the first quarter of 2026, with consumer ARR accounting for about 40%.

SemiAnalysis points out that the core advantage of the API model lies in the absence of a revenue cap per user— as the same customer adopts more Agentic Workflows, their token consumption and corresponding revenue will continue to grow, allowing for expansion without needing new customers. Anthropic's Chief Financial Officer Krishna Rao disclosed in a podcast this May that the company's net revenue retention rate (NRR) is as high as 500%, meaning that among the customers contributing $30 billion in ARR in the first quarter, they had only contributed $2 billion a year ago.

Gross margin advantage forms a compounding flywheel, with a clear gap from OpenAI

The differences in business models are directly reflected in gross margins. SemiAnalysis estimates that Anthropic's current overall gross margin has risen to the mid-60% range, while this figure was negative 94% in 2024. The gross margin for the API business exceeds 80%.

The core driver of the significant improvement in gross margin is the enhancement of inference efficiency. Measured by ARR per megawatt of computing power, Anthropic's figure is expected to reach $60 million later this year, up from just $16 million nine months ago. Since the cost of inference computing power is largely fixed, when the amount of tokens processed per unit of computing power or the token pricing increases, the marginal profit margin approaches 100%.

The report estimates that if both Anthropic and OpenAI reach $100 billion in ARR, OpenAI will have about $25 billion less in gross profit due to the need to support over 900 million free users (SemiAnalysis estimates the monthly service cost at about $0.70 per person), which will directly affect both parties' reinvestment capabilities in training the next generation of models.

SemiAnalysis introduces "Earnings Before Tax and Interest on Training" (EBTIT) as a core metric for measuring laboratory reinvestment capability, with Anthropic's EBTIT profit margin reaching 36% in the second quarter of 2026. The report predicts that by 2028, Anthropic's cumulative EBTIT will exceed OpenAI's by $250 billion.

Beyond programming, cybersecurity may become the next growth engine

SemiAnalysis estimates that currently over 65% of the laboratory's ARR comes from programming-related use cases, with programming tool startups like Cursor, Cognition, Loveable, and Replit collectively contributing about $6 billion in ARR. Meta is Anthropic's largest single customer, but its share remains between 3% and 5%.

The report suggests that cybersecurity will be the next explosive vertical field following programming, and it is expected that the release of the new Fable model will further enhance token pricing and expand application scenarios, driving monthly NNARR to exceed the current $10 billion level in the second half of 2026. Vertical fields such as healthcare, finance, and biotechnology are also listed as potential significant TAM expansion directions.

In terms of distribution channels, the "Token as a Service" (TaaS) model, sold indirectly through large-scale cloud platforms like AWS Bedrock and Azure Foundry, is growing rapidly, currently accounting for 15% to 20% of Anthropic's ARR, up from just 5% to 10% a quarter ago. SemiAnalysis believes that paying 20% to 30% of revenue to large-scale cloud platforms remains economically reasonable from the perspective of enterprise customer reach efficiency and compliance convenience.

Computing power bottlenecks are the biggest variable, IPO provides a financing channel

The growth prospects for Anthropic face core constraints from computing power supply.

SemiAnalysis predicts that by 2030, the combined unconstrained computing power demand of Anthropic and OpenAI will exceed 100 gigawatts (GW), while the net new computing power for 2025 and 2026 will only be 2.5 GW and 5 GW, respectively, and currently, the two companies have a combined available computing power of just over 6 GW.

This supply-demand gap gives the IPO clear strategic significance. The report points out that the funds raised from the IPO will mainly be used to fill the continuously expanding computing power demand gap between inference operations and new model training, and to lock in computing power resources at more favorable financing costs in advance. The report also mentions that Meta is considering leasing computing power to external parties (according to market rumors from July 1, 2026), and it is expected that Anthropic will procure incremental computing power from such trusted suppliers.

SemiAnalysis also lists major risk factors, including: OpenAI's rumored price reduction plans, competition pressure from Google DeepMind and Meta in programming models, potential regulatory restrictions from the government on the release of cutting-edge models, and the dilution effect on overall gross margin from the rising share of TaaS revenue. The report clearly states that if regulatory frameworks hinder model releases and narrow the capability gap between open-source models and cutting-edge proprietary models, it will fundamentally weaken Anthropic's business moat.