Inside Anthropic, the $965 Billion AI Juggernaut | The Circuit

### TL;DR Anthropic was founded by siblings Dario and Daniela Amodei, who left OpenAI, with safety as its core mission. Powered by the Claude model, it has achieved exponential growth in enterprise software and is valued at nearly a trillion dollars. ## From OpenAI Defectors to the Trillion-Dollar Club In 2021, Dario Amodei and Daniela Amodei led seven co-founders out of OpenAI to establish Anthropic. The company name comes from the Greek word for "human," reflecting its mission to build responsible AI that benefits humanity over the long term. Today, Anthropic is an AI frontrunner valued at nearly a trillion dollars, works with the Pentagon, and has released a model claimed to breach modern cybersecurity barriers. Dario is the visionary, focused on AI safety research; Daniela is the executor, turning wild ideas into reality. The two have been close since childhood and always wanted to do something big together. Dario was obsessed with math and sci-fi from a young age—taking calculus in middle school and Berkeley math classes in high school—while Daniela was more into reading and art. They complement each other perfectly. While at OpenAI, Dario proposed the "scaling law"—the counter-mainstream idea that simply increasing data and compute could improve models—which later became the foundation of ChatGPT's rise. But the Amodei siblings grew to differ with Sam Altman on values and trust. Dario put it succinctly: "When you feel you can't trust someone, when they're not honest and their values aren't what they say they are, you can only do your thing and they do theirs." So they left OpenAI. ## Safety-First Beliefs and "Constitutional" Training From the start, Anthropic positioned itself as the ultimate safety-conscious AI company. Dario published long essays like *Loving Machines* and *The Teenage Years of Technology*, pondering both the wonders and worst-case scenarios of AI. Claude's unique style is called "professional warmth"—approachable but with distance, neither cold nor overly familiar. Claude is trained following a set of "constitutional" principles to ensure the model is harmless, honest, and helpful. Dario emphasizes that models can lie (hallucinate) unintentionally or deliberately, and even deceive users. "We have a lot of work on harmlessness, making sure the model doesn't accidentally produce harmful information or enable bad actors." For value injection, Anthropic draws on foundational documents like the *Universal Declaration of Human Rights* and has talked with religious leaders to incorporate core values spanning different faiths. Early versions of Claude sometimes acted like a "nanny," overly worried about user queries, requiring researchers to fine-tune for balance. ## Choosing the Enterprise Track: Aligning Values and Business Anthropic's revenue has soared over the past year, reaching profitability for the first time, thanks to a focus on high-margin enterprise tools. Dario explains: "If you choose a business model fundamentally at odds with your values, you either betray your values or become irrelevant." Consumer apps often chase engagement and addiction, while the enterprise market—partnering with biotech to cure diseases or helping energy companies improve efficiency—aligns better with AI's real value to humanity. The launches of Claude Code and Claude Cowork shocked the market: Claude Code can automate most of software engineering, and Claude Cowork gives that capability to everyone. After the release, software stocks lost $285 billion in market cap overnight, with traders calling it "the SaaS apocalypse." Dario believes the pie is growing, but legacy software companies face devaluation or even failure if they don't adapt. ## From "Autocomplete" to "Jetpack": Claude Code's Disruption Boris Cherny is the engineer behind Claude Code and Claude Cowork. When he joined Anthropic in 2024, he was living a slow-paced life in rural Japan (making miso, visiting farmers' markets). After using his first AI chatbot, he was amazed and decided to move back to the US to join the revolution. Cherny's goal: make a coding agent that does everything, not just autocomplete words or sentences. Now, almost all code in his team is written by Claude, and he hasn't written a line of code himself in six months. "The nature of engineering work has completely changed. I suddenly feel like I have superpowers, like I'm wearing a jetpack." In a video demo, Cherny had Claude build a recipe app from scratch—within minutes it could suggest weekly meals, view recipes, and show instructions. He makes a habit of saying "thank you" to Claude and being polite. ## Exponential Growth and the Future Anthropic's API usage has grown nearly 17x year-over-year, and it has released eight frontier models to developers and users in the past year. At the second Code with Claude conference, the company showcased 80x annualized growth. Dario describes the experience: "Smooth exponential growth—nothing, nothing, nothing, and then whoosh, it gets crazy." As the AI arms race escalates, the Amodei siblings want to cast themselves as the good guys, but Anthropic's technology is profoundly changing how humans work, learn, think, and even wage war. For a young, fast-growing startup, that's a huge responsibility. Dario sums it up: "We just keep focusing on how to train good models, put them in good products, and ensure safety." --- Source: Inside Anthropic, the $965 Billion AI Juggernaut | The Circuit (https://youtu.be/v1wZwxY3CMg?is=NI5gCQz8XbEjo5uh)