Breakdown 001 · Company
Anthropic
The reel is sixty seconds. That is enough to understand what Anthropic is, and not enough to check it. So every claim it makes is below, with the source that carries it and the date it was true.
The breakdown, at length
Same six questions, same order, more room. The reel says each of these in one sentence.
Who
Anthropic is an AI company, founded in 2021 by Dario and Daniela Amodei — siblings, both of whom came from OpenAI, where he ran research and she ran safety and policy. Two of the people best placed to know how the technology was being built left the company building it, and started their own.
What
It makes Claude. In practice that is three things wearing one name: something you chat with, something that writes and edits code, and something that reads documents far longer than a person wants to. The third is the least discussed and, for most businesses, the one that actually changes a workday.
How
You type, it answers. Underneath is a large language model — trained on an enormous quantity of text, with one job: predict what comes next. Everything else it appears to do is that single ability, pointed at a harder question.
People reach for autocomplete as the analogy. Here is where it breaks: autocomplete finishes your sentence, whereas this predicts its way through an argument it was never shown. Same mechanism, a different order of scale.
When
The company started in 2021. Claude opened to the public on 14 March 2023 — as two models on the same day, Claude and the lighter Claude Instant.
The reel says only “2023”, because when it was written the month was not confirmed. It is now — see what changed after the reel.
Where
In the Claude app, if you go looking for it. But most of the use is not people visiting a chat window — it is businesses building Claude into software that never mentions it. You have probably already used it without being told.
Why
The founders left OpenAI because they thought AI was moving too fast. Rather than promise to be careful, they made it structural: Anthropic is a public benefit corporation, so its board is legally obliged to weigh safety alongside shareholder return. Whether that holds up under pressure is a fair question.
The fact record
Everything above, with its source
Nothing goes on screen in a reel unless it is in this table first.
1
Founded in 2021 by Dario and Daniela Amodei, siblings, previously VP of Research and VP of Safety & Policy at OpenAI.
Britannica Contrary Research as of 2021 does not expire
2
Anthropic makes Claude — a chat assistant that also writes code and reads long-context documents.
Britannica as of 2026 recheck by 2026-12
3
Large language models are trained to predict the next piece of text.
General and uncontested does not expire
4
Claude opened to the public on 14 March 2023, in two versions — Claude and the faster, cheaper Claude Instant.
Anthropic — Introducing Claude as of 2023-03-14 confirmed 2026-08-21 on the primary source
5
The majority of Claude’s use is businesses building it into their own software, rather than people using the app directly.
ValueAdd as of 2025-10 ageing — used as a direction, never as a number
6
Anthropic is a public benefit corporation, so its board is legally required to weigh safety alongside shareholder return.
Britannica as of 2021 expires if the structure changes
What changed after the reel
Two claims were cut from the reel because they were not confirmed when it was written. The research pass for the written breakdown resolved both. Rather than quietly fix the page, here is what happened to each.
The month Claude launched — now confirmed. It was 14 March 2023, from Anthropic’s own announcement. The reel still says only “2023”, which is not wrong, just less precise than this page can now afford to be.
Anthropic’s share of enterprise AI spending — confirmed, with strings. The figure is 40% of enterprise LLM spend in 2025, from Menlo Ventures. Three things are wrong with it: Menlo is an investor in Anthropic, which it discloses itself; it is a survey estimate rather than accounting; and the way Menlo sizes the market leaves out Claude Code, which is the product Menlo credits for the rise. It is probably roughly right, but it is not the same kind of fact as a launch date.
The full treatment of that 40%, with all three caveats, is in the written breakdown below.
What did not fit
Sixty seconds has room for one idea, and this reel used it on what Anthropic is. That left out how it makes money — the revenue picture, what the company is valued at, and how much of that revenue survives the cost of the computers it runs on.
That is the second breakdown, not a paragraph at the end of this one. It needs its own sources, and the numbers move fast enough to need their own dates.
The long version
The same six questions, at length: the full funding history with every valuation and the run-rate behind it, how Constitutional AI and the Responsible Scaling Policy actually work, who legally controls the company, and the money story the reel had no room for. Twelve pages, seventeen sources, every claim numbered, plus a page on what could not be confirmed — including a widely repeated revenue figure that contradicts Anthropic’s own filing.
If you run a business
I run Leadoo, which helps Malaysian service businesses put this kind of thing to work — answering enquiries, following up, being findable. If that is relevant to you, message me. If you are just here for the breakdown, that is genuinely the point of the page.