SemiAnalysis: Anthropic's Q3 profit will exceed $1 billion

Author: Xu Chao

A new analysis by research firm SemiAnalysis reveals that Anthropic is reshaping the AI commercialization landscape with profitability and growth rates far exceeding its competitors. Leveraging a high-margin, API-centric business model, Anthropic has become the leader in the B2B AI market.

According to the in-depth report published by SemiAnalysis, Anthropic is expected to achieve $1 billion in GAAP earnings before interest and taxes (EBIT) by the third quarter of 2026, corresponding to a margin of approximately 6%. Meanwhile, its annual recurring revenue (ARR) has surged from $9 billion at the end of 2025 to over $60 billion currently. The firm projects that if Anthropic maintains a pace of roughly $15 billion in net new ARR (NNARR) per month, its ARR could reach $300 billion by the end of 2027, implying an enterprise value of $6 trillion and making it the most valuable company in the world.

Anthropic confidentially filed for an IPO on June 1. SemiAnalysis believes that going public at this time carries strategic urgency — Alphabet has completed an $84.75 billion equity financing, and Meta has also reportedly raised tens of billions of dollars in funding, meaning the capital market window is narrowing. The report notes that Anthropic's superior financial data and business model mean it should list before OpenAI to seize the initiative in capital competition.

Claude Code Ignites B2B Market, ARR More Than Triples in a Single Quarter

The inflection point in Anthropic's performance stems from the explosive adoption of Claude Code. SemiAnalysis data shows that Claude Code now accounts for over 7% of all code commits on GitHub, directly driving the company's new monthly ARR to jump from $3 billion in January to $11 billion in March.

In terms of revenue structure, Anthropic and OpenAI show significant divergence. Approximately 75% to 85% of Anthropic's ARR comes from usage-based API billing, while consumer subscriptions account for only 5% of total ARR. In contrast, OpenAI still derived over 65% of its revenue from subscriptions in Q1 2026, with consumer ARR accounting for about 40%.

SemiAnalysis points out that the core advantage of the API model is the absence of a per-user revenue cap — as the same customer adopts more agentic workflows, their token consumption and corresponding revenue will continue to grow without the need to acquire new customers. Anthropic CFO Krishna Rao disclosed on a podcast in May that the company's net revenue retention (NRR) is as high as 500% — meaning that among the customers contributing $30 billion in ARR in Q1, that same cohort of customers contributed only $2 billion a year earlier.

Gross Margin Advantage Creates a Compounding Flywheel, OpenAI Lags Significantly

The difference in business model is directly reflected in gross margins. SemiAnalysis estimates that Anthropic's current blended gross margins have risen to the mid-60% range, compared to a negative 94% in 2024. The API business alone has gross margins exceeding 80%.

The core driver of the dramatic gross margin improvement is inference efficiency gains. Measured by ARR per megawatt of compute, Anthropic's metric will reach $60 million later this year, compared to just $16 million nine months ago. Since inference compute costs are largely fixed, when the number of tokens processed per unit of compute or the token pricing increases, the incremental profit margin approaches 100%.

The report calculates that if both Anthropic and OpenAI reach $100 billion in ARR, OpenAI — which must support over 900 million free users (SemiAnalysis estimates a monthly service cost of roughly $0.70 per user) — would have approximately $25 billion less in gross profit than Anthropic. This gap will directly affect each company's ability to reinvest in training next-generation models.

SemiAnalysis introduces "Earnings Before Training Interest and Taxes" (EBTIT) as a core metric for measuring a lab's reinvestment capacity. Anthropic's EBTIT margin reached 36% in Q2 2026. The report predicts that by 2028, Anthropic's cumulative EBTIT will be $250 billion higher than OpenAI's.

Beyond Coding, Cybersecurity Could Be the Next Growth Engine

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

The report believes that cybersecurity will be the next explosive vertical after coding, and expects that the launch of the new Fable model will further increase token pricing and expand application scenarios, pushing monthly NNARR beyond the current $10 billion per month level in the second half of 2026. Healthcare, finance, biotechnology, and other verticals are also listed as potential major TAM expansion directions.

On the distribution front, the "Token-as-a-Service" (TaaS) model — indirect sales through hyperscale cloud platforms like AWS Bedrock and Azure Foundry — is growing rapidly, now accounting for 15% to 20% of Anthropic's ARR, compared to just 5% to 10% a quarter ago. SemiAnalysis believes that paying hyperscalers a 20% to 30% revenue share remains economically reasonable given the efficiency of enterprise reach and compliance convenience.

Compute Bottleneck Is the Biggest Variable, IPO Provides Funding Channel

The core constraint facing Anthropic's growth prospects comes from compute supply.

SemiAnalysis predicts that by 2030, the combined unconstrained compute demand of Anthropic and OpenAI will exceed 100 GW, while the net new compute added in 2025 and 2026 is only 2.5 GW and 5 GW respectively. Currently, the two companies' total available compute is just over 6 GW.

It is this supply-demand gap that gives the IPO its clear strategic significance. The report states that the funds raised from the listing will primarily be used to fill the growing gap in compute demand between inference operations and new model training, and to lock in compute resources in advance at more favorable financing costs. The report also mentions that Meta is reportedly considering leasing compute to external parties (based on market rumors from July 1, 2026), and expects Anthropic to purchase incremental compute from such trusted suppliers.

SemiAnalysis also lists key risk factors, including: OpenAI's rumored price cuts, competitive pressure from Google DeepMind and Meta in coding models, potential government regulatory restrictions on frontier model releases, and the dilutive effect on blended gross margins from the rising share of TaaS revenue. The report explicitly states that if the regulatory regime hinders model releases and narrows the capability gap between open-source models and frontier proprietary models, it would fundamentally undermine Anthropic's commercial moat.