When this deck was presented, Anthropic had just over 100 employees, a waitlist of 2,500 and no public product.

In May 2026, the same company closed a $65B Series H at a $965B post-money valuation and reported a $47B revenue run rate.

This is the deck Anthropic used to sell itself in early 2023. Not to investors, but to the people it wanted to hire.

It matters because it captures a frontier AI lab in the exact moment before its breakout, with almost nothing to show except a team, a mission and a few strong signals of demand.

You'll learn how Anthropic built a credible story with no revenue, what it deliberately left out, and which moves you can reuse in your own deck.

Below is the full breakdown, slide by slide. The complete deck is available to download at the end.

Inside you'll find:

Most people will open this file expecting a fundraising deck. It isn't one.

There is no market size slide, no business model, no pricing, no competition map, no financials and no ask. Instead, the story ends with an office full of plants and a list of open roles across engineering, security, business and product.

The clues also date it. Claude appears in private beta, and the Series C is described as "in talks" for 2023. That round was announced on May 23, 2023: $450M led by Spark Capital. The public Claude chatbot would only arrive in July 2023.

→ A deck for talent is still a pitch. The audience changes the ask, not the job: convince smart people with options that this is where they should bet their time.

That makes this deck useful for any founder. Hiring your first 20 people is a sales process, and most startups treat it as an afterthought.

The introduction is almost aggressively short: an AI safety and research company, a Public Benefit Corporation, based in San Francisco. Three lines.

Then comes the mission slide, which does something most founders struggle with. It compresses the entire company into three adjectives.

Reliable, steerable, interpretable. Each word is a promise, and each one implicitly contrasts with what the market feared about AI at the time: systems that are unpredictable, hard to control and impossible to understand.

→ The most telling detail is what's missing: the word "powerful". In a race defined by capability, Anthropic chose to position itself on trust.

Notice also the PBC mention on the second slide. For a certain type of candidate, the legal structure itself was a signal that the mission wasn't just marketing.

With no public product, Anthropic needed a reason for people to believe. It found two.

The first was pedigree. The "cutting edge" slide opens by reminding the audience that its founders led the development of GPT-3, then points to its Helpful, Honest and Harmless framework and more than 15 safety papers. On the right, it lists four of them, including the paper introducing Constitutional AI.

The second was third-party validation. The next slide doesn't feature a single claim written by Anthropic. It shows a Marginal Revolution post about Claude passing a law and economics exam, a TechCrunch headline and two posts from Riley Goodside, then a staff prompt engineer at Scale AI.

→ The boldest choice on this slide is the TechCrunch headline. It says Claude improves on ChatGPT but still suffers from limitations. Anthropic chose to show it anyway.

For a company positioning itself on honesty, putting a mixed review on its own slide is a coherent move. It also makes the positive quotes more believable. Goodside's line that Claude feels "simultaneously safer and more fun" lands harder next to an honest critique.

Early-stage founders often try to fill the traction gap with market-size slides. Anthropic went the other way: it showed evidence that people were already coming to it.

None of these is revenue. All of them are pull. Inbound demand and a waitlist tell the audience that the market is asking for the product before the company pushes it.

The fundraising slide adds a second kind of momentum. It shows a Series A of $124M in 2021, a Series B in 2022 and a Series C on the way, with a headline claiming roughly $1B raised across the first two rounds.

One nuance worth knowing: public records put the Series B at $580M, which puts the two rounds at around $700M combined. The slide's rounded figure likely folds in other capital raised around that period. If you use this deck as a reference, don't copy the rounding; investors will check.

Then the story turns into a funnel. People want Claude, demand exceeds capacity, and therefore there are many open roles.

→ This is the cleverest slide in the deck. It converts a constraint (not enough capacity) into an invitation. The problem becomes the reason to join.

The bet paid off quickly. According to reporting by Alex Kantrowitz, the company had tried to stay under 150 employees, and after the Claude chatbot launched, it was soon hiring more people in a day than it had employed in its whole first year.

If you showed this deck to an investor today, the list of missing pieces would be long:

Most of this is fine for a talent audience. Candidates care more about mission, team and momentum than about unit economics.

→ But don't confuse the exception with the rule. Anthropic could skip the business slides because its founders had built GPT-3. Most founders can't borrow that level of credibility, so they have to earn it with numbers.

The useful exercise is to read this deck twice: once as a model for how to tell a story, and once as a checklist of what you'd need to add before showing it to a VC.

You don't need a frontier lab to use what this deck does well. Here are five moves any founder can apply this week:

And one move to avoid: don't round your numbers up. The rest of the deck earns trust, and a single inflated figure can spend it.

Ten slides, zero revenue and a clear story were enough to help a 100-person company attract the talent that would take it to a $965B valuation.