The foundation and scaling of Anthropic

Dario and Daniela Amodei founded Anthropic in 2021 to prioritize AI safety through Constitutional AI. The company scaled rapidly through massive Amazon and Google partnerships, reaching a $30 billion revenue run-rate by April 2026.

The foundation and scaling of Anthropic

Dario Amodei and Daniela Amodei led eight former OpenAI researchers to form Anthropic in 2021. The group, which included Jared Kaplan, Jack Clark, Chris Olah, Tom Brown, Sam McCandlish, and Ben Mann, left OpenAI in late 2020 because they believed the organization commercialized AI too rapidly. Dario Amodei, who left OpenAI after disagreements over safety and commercialization, leads the company alongside his sister Daniela, the President and Chair of the Board, while the founding team manages research and operations to ensure models remain reliable. Daniela Amodei previously led safety and policy operations at OpenAI and worked on risk management at Stripe. Dario Amodei, who earned a doctorate in biophysics from Princeton, worked on the development of GPT-2 and GPT-3 during his time at OpenAI. Anthropic functions as a public benefit corporation to allow directors to prioritize safety mission goals alongside profitability. The company relies on Constitutional AI, a method that trains models against a set of principles rather than relying solely on human feedback. This research aims to mitigate the technical alignment problem, where powerful AI systems might develop behaviors that conflict with human intentions. The team also focuses on mechanistic interpretability research to reverse-engineer neural networks into something humans can audit.

Anthropic relies on strategic cloud partnerships to manage the massive compute requirements for training Claude. Amazon recently provided $5 billion in funding, with an additional $20 billion possible based on commercial milestones. This follows an earlier $8 billion investment from Amazon. Anthropic also signed a deal with Google for $10 billion, which leaves $30 billion more contingent on performance. The company is also building significant capacity on AWS and will commit over $100 billion to AWS technologies over the next decade. Through an announcement with Google and Broadcom, Anthropic secured 5 gigawatts of computing capacity to support its needs. The company uses Project Rainier, one of the largest compute clusters in the world, to advance its research.

Investor Current Investment Future Potential Infrastructure Focus
Amazon $5 billion $20 billion AWS Trainium and Graviton
Google $10 billion $30 billion Google Cloud and TPUs

The company uses these investments to scale the infrastructure needed for its growing user base. This collaboration with Amazon allows Anthropic to continue advancing AI research while delivering Claude to its 100,000 customers on AWS Bedrock. Anthropic reported $4.6 billion in revenue in 2025, a significant increase from $386 million in 2024. The company also reported $131.65 billion in cash and short-term investments as of June 2026.

Enterprise adoption and policy friction

The enterprise market adopted Claude rapidly throughout 2024 and 2025. By the first half of 2026, Anthropic reported $16.29 billion in revenue, a massive jump from $1.16 billion in the same period in 2025. You already know the basics of how LLMs scale, but the speed of Claude’s enterprise integration surprised many observers. Claude Code, an assistant for developers, reached a $2.5 billion run-rate by February 2026 and contributes to roughly 4% of all public GitHub commits worldwide. Claude also powers 25 billion API calls per month, and 45% of those calls come from enterprise customers. 70% of Fortune 100 companies use Claude in some capacity. Claude’s monthly active users reached 220 million in the first quarter of 2026. Approximately 52% of Claude users fall into the 18 to 24 age demographic, while 34% of all tasks involve computer or mathematical work. The company’s revenue growth was extreme; it scaled from $45 million in 2023 to a $30 billion run-rate by April 2026.

Dario Amodei maintains a public stance on AI safety; he expresses a 10% to 25% probability of catastrophic harm from AI. In his 2024 essay, "Machines of Loving Grace," he speculated on how AI could improve human welfare, while his 2026 essay, "The Adolescence of Technology," focused on the risks of powerful AI. He advocates for democratic nations to maintain an AI lead through advanced military AI and semiconductor export controls. In February 2026, Amodei refused a request from the Department of Defense to remove contractual bans on Claude usage for mass domestic surveillance or fully autonomous weapons. This refusal led the Pentagon to label Anthropic a supply-chain risk and order US agencies to stop using Claude. A federal judge later issued a temporary injunction against the Department of Defense and agreed that the actions appeared to be retaliation against protected speech. Will the tension between private safety commitments and national security requirements continue to define the company’s trajectory?

airtrain.ai
airtrain.ai

The airtrain.ai newsroom covers AI research, models and the tools built on them.

More on this topic

Stay ahead of AI

Get the week's most important AI stories delivered to your inbox every Monday.

No spam. Unsubscribe anytime.

More Stories