The $40 Billion Investment and Enterprise Adoption
Anthropic reports more than 1,000 customers that spend over $1 million annually. This count doubled in two months after the company announced its Series G funding round in February 2026. While some industry observers call the news of a potential $40 billion investment from Google an unverified rumor, Anthropic’s annual revenue run rate exceeded $65 billion by the end of July 2026 according to Reuters reports. The company’s valuation reached $380 billion in February 2026. It sees massive demand for its models, including Claude Opus 4.1 and Claude Code. TELUS uses Claude Opus 4.1 to build 13,000 AI tools, which saved 500,000 hours and produced $90 million in benefits. Brex uses Claude Opus 4 on AWS Bedrock to automate 75% of transactions and save 169,000 hours monthly. Snowflake uses Claude Sonnet 3.5 and Opus 4.1 in Cortex AI to achieve over 90% accuracy on text-to-SQL queries. TELUS and Brex attribute tens of millions of dollars in realized benefits directly to Claude-powered automations. Snowflake and DoorDash adopted Claude through Bedrock to simplify compliance and unlock integrated monitoring. CRED uses Claude Sonnet 4 and Claude Code to double developer velocity and increase test coverage by 10%. A federal judge in San Francisco signed off on a $1.5 billion settlement in July 2026 regarding a class-action lawsuit from authors. Claude Team provides access to Opus 5, Sonnet 5, Haiku 4.5, and Claude Code. The plan costs $20 per user per month on an annual plan or $25 on a monthly plan, with a limit of 150 seats. Premium seats cost $100 per user per month for an annual plan or $125 monthly.
| Plan Tier | Annual Cost per User |
|---|---|
| Claude Team | $20 |
| Claude Team (Premium) | $100 |
Long-Context Accuracy in Medical Reasoning
Claude Opus 5 (Xhigh) scored 80.3% on the AA-LCR v1.1 long context benchmark in September 2026. It also achieved 58.3% on a private medical-record synthesis and reasoning benchmark. This test evaluates performance across long, fragmented case documents using a private held-out set of data where higher values indicate better performance. The task tests reasoning by analyzing fragmented case documents. Claude Fable 5.1 (Max, Default Fallback) scored 71.1% in the medical reasoning test. Claude Fable 5 (Max, Opus 4.8 Fallback) reached 64.4% while GPT-6 Astra (Max) achieved 35.0%. The Opus 5 model family also reached 166.4 on the Epoch Software ECI and 68.1% on the AA Coding Agent Index.
| Model | Benchmark Result |
|---|---|
| Claude Fable 5.1 (Max, Default Fallback) | 71.1% |
| Claude Fable 5 (Max, Opus 4.8 Fallback) | 64.4% |
| Claude Opus 5 (Xhigh) | 58.3% |
| GPT-6 Astra (Max) | 35.0% |
Reliability and the Hyperscaler Strategy
The company experienced eight outage events in April 2026. These outages included a ten-hour downtime on April 6 and authentication failures on April 28 that generated 12,000 user reports. Independent monitoring shows the claimed 99% uptime for Claude Max hit 84% in practice. Patrick Moorhead notes that Anthropic’s acceptable use policy overrides user requirements, which creates procurement issues for businesses with defense-adjacent revenue. This follows a February 27 order from the Trump administration to halt federal agency use of Claude. Defense Secretary Pete Hegseth designated Anthropic a supply-chain risk due to policy clauses regarding mass domestic surveillance and fully autonomous weapons. How will Anthropic resolve the supply-chain risk designation from the Pentagon? You should probably reserve direct Anthropic access for R&D and the newest features. Direct your production workloads toward hyperscalers like AWS Bedrock or Google Vertex AI to ensure better SLAs and operational maturity. Anthropic announced a 99.99% SLA in March 2026, but these terms are negotiated case-by-case. AWS provides a 99.9% uptime SLA for its Bedrock offering. Anthropic announced a $100 billion deal with AWS for 5 gigawatts of Trainium 3 capacity and a deal with Google and Broadcom for 3.5 gigawatts of TPU capacity in 2027.
