Anthropic tells investors it will post second straight quarterly profit

Anthropic expects a second consecutive quarter of positive adjusted operating income, following a reported Q2 adjusted operating profit as revenue rose 14-fold year over year to $11.5 billion. Annualized revenue reportedly reached $65 billion at end-July, versus $9 billion at end-2025, while gross margins exceeded 80% before partner revenue sharing and AI-model training costs. The company has selected Nasdaq for a potential IPO that could value it at $2 trillion or more, although investor scrutiny of AI development costs and model-security risks remains elevated.

Analysis

The key investable signal is not reported adjusted profitability but the implied monetization efficiency of enterprise AI workloads. If Anthropic can sustain high gross margins after inference costs, pricing pressure on Microsoft/OpenAI and Google (GOOGL) may be less severe than feared; conversely, the exclusion of partner revenue-sharing, model-training expense and SBC means the metric says little about durable free-cash-flow conversion. Investors should treat any IPO valuation anchored to adjusted operating income as vulnerable to a sharp reconciliation discount once audited public-company disclosures separate recurring inference economics from development spend.

AMZN has asymmetric strategic upside: Claude demand can raise AWS utilization and reinforce AWS as a neutral AI distribution layer, while Amazon's commercial terms capture part of Anthropic's economics without requiring AMZN to own all model risk. The second-order negative is that a highly successful Anthropic broadens customers' ability to avoid Azure/OpenAI and Google cloud stacks, potentially intensifying hyperscaler AI price competition over the next 6-18 months. A credible industry-led safety slowdown would also favor scaled incumbents with capital, compliance teams and installed enterprise channels, while reducing the near-term hardware order velocity supporting SMCI.

An eventual Nasdaq listing is a sentiment and market-structure catalyst, not a material near-term earnings driver for NDAQ. The IPO could reset public AI valuation benchmarks: a premium valuation would support AI infrastructure multiples initially, but audited disclosures showing large training and partner costs could trigger multiple compression across high-expectation AI beneficiaries. The contrarian view is that the market may capitalize a revenue run-rate before establishing revenue durability, customer concentration, net retention, and the cash cost of maintaining frontier-model parity.

Key Decisions for Investors

• Maintain or initiate a 6-12 month long AMZN / short GOOGL relative-value position in modest size: Anthropic distribution and AWS consumption are incremental upside to AMZN, while model choice commoditization is a greater risk to GOOGL's integrated-stack narrative. Falsify if AWS AI growth fails to accelerate in the next two earnings reports or if Anthropic meaningfully shifts workload deployment away from AWS.
• Do not underwrite a direct Anthropic valuation proxy or chase private-market read-through into SMCI. Set a watch trigger for audited IPO filings: training cash spend, partner revenue share, customer concentration and net revenue retention must be disclosed before treating the profitability claim as evidence of sustainable FCF.
• Use NDAQ as a limited-risk 3-6 month event exposure rather than a core AI trade; add only on weakness ahead of a formal filing. Upside comes from IPO pipeline/multiple sentiment, but direct listing economics are unlikely to move consolidated earnings materially; exit if filing timing slips or broader IPO issuance weakens.
• Reduce tactical SMCI beta into any AI-safety-driven policy or voluntary-capex slowdown headlines over the next 1-3 months. The thesis reverses if hyperscaler capex guidance remains intact and server lead times re-expand, which would demonstrate that model-safety discussions are not altering hardware procurement.