Inside Anthropic — the $965 billion AI company
Bloomberg's Emily Chang goes behind the scenes with Anthropic's founders on AI safety, billion-dollar growth, war, and the future of your job. Here's the whole 47-minute documentary, in plain English.
What this video is about
This is a rare, in-depth documentary-style interview with Dario and Daniela Amodei — the brother-sister founders of Anthropic, the company behind Claude AI. In 47 minutes, Bloomberg journalist Emily Chang traces the entire Anthropic story: from a small startup meeting in a park to a nearly $1 trillion company reshaping tech, business, warfare and the future of work.
Anthropic is one of the most powerful AI companies in the world right now. Its model, Claude, is used by millions of people and businesses — and also by the U.S. military. This video explains how they got here, what they believe, and what comes next.
Meet the founders
Dario Amodei
The visionary. Studied neuroscience, then moved to AI at Baidu and Google. Joined OpenAI in 2016, left in 2021. Known for warning the world about AI dangers while building some of the most powerful AI ever made.
Daniela Amodei
The operator. Started her career at Stripe as an early employee. She turns Dario's big ideas into actual company operations and runs the day-to-day — the leadership team reports to her.
Boris Cherny
Moved from rural Japan (where he was making miso) to Anthropic in 2024. He built Claude Code and Claude Cowork — the products that made Anthropic profitable. He says Claude now writes 100% of his own code.
The origin story, in simple terms
Joining OpenAI
Dario joins OpenAI (the company behind ChatGPT). Daniela joins shortly after.
The split
Dario and Daniela leave OpenAI — they disagreed with leadership over safety and trust. "When you can't trust someone, you go do your own thing." They start Anthropic with seven co-founders.
Precita Park Pandemic
The early team met in Precita Park, San Francisco, pulling up chairs on the grass during lunch to talk about what they were building — a safety-first AI company.
Claude Code arrives
Boris Cherny joins and builds Claude Code — the AI coding tool that becomes Anthropic's biggest product. AI now writes almost all code at Anthropic internally.
Near a trillion Profitable
Anthropic becomes profitable. Revenue skyrockets and valuation hits nearly $1 trillion. API volume grows 17× in a year, and the company ships 8 frontier models in 12 months.
What makes Claude different
Claude is Anthropic's AI chatbot and the product that drives all their revenue. Unlike competitors, Claude was trained with a set of values called a "Constitution" — a list of principles that tells Claude how to behave.
"There is more of a feeling of professional warmth. The goal is not for it to be your best friend — it should feel approachable, but professional."Dario Amodei
The "Constitution"
Claude is trained on a set of rules based on things like the UN Declaration of Human Rights. Anthropic even consults religious leaders to define core human values across cultures.
Hallucinations
An AI "hallucination" is when the model makes up false information. Claude is trained to minimise this, and Anthropic researches how to stop models deceiving users on purpose too.
Claude Code
The big money-maker. An AI agent that writes entire software programs. Anthropic engineers now run up to thousands of Claude instances in parallel doing their coding work.
Enterprise focus
Anthropic chose business tools over viral consumer apps. Dario says consumer apps push engagement and addiction; enterprise tools help businesses and create real value.
AI and your job — the big fear
70% of Americans believe AI will destroy jobs. About a third worry it will take their job. Dario doesn't sugarcoat it — he's estimated AI could eliminate up to 50% of entry-level white-collar jobs within one to five years. Management, finance, legal and software roles could all look very different soon.
"We could have a very unusual combination of very fast GDP growth and high unemployment — lots of low-wage jobs and high inequality."Dario Amodei
The hopeful part: Dario believes new jobs will come — especially in physical work (manufacturing, building), human-centred professions (caregiving, counselling), and roles that manage or direct AI systems. His example is doctors: AI will get great at diagnosis, but the bedside manner and emotional support stay human.
The Pentagon clash
Anthropic signed a $200 million contract with the Pentagon in 2025 alongside OpenAI and Google. Claude was reportedly used by the U.S. military in an operation involving Venezuelan president Nicolás Maduro.
Then it got messy. The Pentagon demanded Anthropic remove Claude's safety guardrails for military use — to allow mass surveillance and autonomous weapons. Anthropic refused. The U.S. government banned Anthropic, and the defense secretary called Dario "an ideological lunatic."
"I won't even call it a fight. This is a debate about what the proper use of AI by the government is."Dario Amodei
A U.S. missile reportedly hit a girls' school in Iran, killing 150+ people, most of them children, and Bloomberg reported Claude-assisted targeting was involved. Dario said Anthropic had no control over specific military decisions — but that this is exactly why he fought to keep humans in the final decision-making loop, not autonomous AI.
The secret weapon: Mythos
Running in the background was the surprise development of a new model called Mythos — so powerful it alarmed even Anthropic's own team. Mythos can find cybersecurity vulnerabilities in every major operating system, like a master key to the world's digital infrastructure.
"Some of the early companies we gave this to said, 'This is a super weapon. You should have to own a gun license to use it. Please don't release this.'"Dario Amodei, on Mythos
Anthropic started "Project Glasswing" — giving carefully selected organisations access to Mythos, including the NSA, to help defend against cyber attacks. Even after being banned by the Pentagon, the NSA still wanted access to Mythos.
The big takeaways
In plain English
- Anthropic isn't just a tech company — it genuinely believes it may be building one of the most dangerous technologies ever, and does it anyway because it thinks safety-focused labs should lead, not just profit-focused ones.
- Claude is now a business powerhouse — enterprise tools like Claude Code made Anthropic profitable for the first time, growing API usage 17× in a year.
- AI job disruption is real and fast — Dario thinks 50% of entry-level white-collar jobs could be hit. The pie grows, but not everyone gets a slice automatically.
- Anthropic will fight the government — they risked the entire Pentagon contract rather than remove safety guardrails, drawing a line at autonomous weapons and mass surveillance.
- Mythos is a turning point — the most powerful cybersecurity AI ever made is being held back deliberately, showing both how fast capabilities accelerate and how careful Anthropic is trying to be.
- "Anthropic" means human — from the Greek word for human. That's intentional: AI for the long-term benefit of humanity.
Key transcript moments
"When you feel you can't trust someone, when their values are not what they say they are — that makes it very hard to keep working with a company. The way to resolve it is you go off and do your thing."
"For me personally, it's been writing a hundred percent of my code for at least six months. I feel like I suddenly have superpowers — a jet pack — and engineering has never been this fun." — Boris Cherny
Soon after Claude Cowork was released, $285 billion in market value vanished overnight. Traders called it a white-collar wipeout in the software sector.
"AI could eliminate half of all entry-level white-collar jobs in the next one to five years." — Dario Amodei
"A U.S. official said that, with the help of LLMs, the military has gone from being able to hit a thousand targets a day to 5,000 targets a day." — Emily Chang
"Imagine a world where everyone had a nuclear bazooka, basically." — describing Mythos's potential if released publicly.
"AI will soon be pretty good at telling you what's wrong and what tests to run. But it can't physically examine you, and it can't have a bedside manner. The interpersonal human part — that's not going to change."
"In the first quarter of this year, if you were to annualize it, we saw 80× growth per year." — Dario Amodei at the Code with Claude conference.
"The experience of the smooth exponential is: nothing's happening, nothing's happening — little things happen, and then zoom, it goes crazy."
Watch the original documentary
The full 47-minute Bloomberg film on Anthropic is well worth your time if you're curious where AI is heading.
Keep reading
- The gap nobody tells you about AI agents — from a single LLM call to a production-grade agent, no jargon.
- AI Workflows — how the automations behind these agents are built.
- AI case studies — interactive walk-throughs of live builds.