How did Anthropic go from a $61bn valuation to $170 billion in five months?

Welcome to AI Collision 💥,

In today’s collision between AI and our world:

• From $61bn to $170bn in a few months
• Follow the…?
• Jeans Picard

If that’s enough to get the capital pumping, read on…

AI Collision 💥

Follow the money. That’s been the golden rule of AI Collision from day one.

And right now, the money is thundering toward one of the most extraordinary (and still private) AI companies in the world, Anthropic.

According to multiple reports, Anthropic is close to finalising a $3 billion to $5 billion funding round, led by Iconiq Capital, that would place its valuation at a jaw-dropping $170 billion.

Let that number hit like it should…

A private company, with no publicly traded shares, no public revenue disclosures, and not exactly a consumer household brand name being valued at $170 billion. More than Rolls Royce. More than BP. More than the GDP of Morocco.

And if you’re thinking, “Wasn’t it just worth $60 billion a few months ago?” — you’d be absolutely right. In March this year they did a huge raise of around $3.5 billion at a $61.5 billion valuation.

So how on Earth did we get to $170 billion?

And is this the hottest AI company in the world now?

Anthropic was born in early 2021. Led by a group of OpenAI staffers, notably Dario and Daniela Amodei, who walked out of OpenAI over concerns about safety, transparency, and OpenAI’s growing closeness with Microsoft.

They didn’t just leave to do research papers and commit to academia, they formed Anthropic, with a single mission… build powerful AI systems that are aligned with human values.

Their flagship product is the Claude AI model, named after Claude Shannon, the godfather of information theory. Claude is the gentle, helpful, non‑hallucinating rival to OpenAI’s GPT.

And it’s designed from the ground up to be safe, thanks to what they call Constitutional AI, a training method that bakes in AI safety guardrails.

Anthropic has led serious work on mechanistic interpretability, digging into how language models form internal representations of meaning, even pioneering “monosemanticity” research that could one day allow us to debug a large language model, like we do with software.

But as good as Claude is, it wouldn’t matter if the capital markets didn’t care and believe in Anthropic.

And boy… the money believes.

Here’s a timeline on how Anthropic’s valuation has skyrocketed since launch,

• 2021: Series A – $124 million from Jaan Tallinn, Dustin Moskovitz, Eric Schmidt
• 2022: Series B – $580 million, including funding from Sam Bankman-Fried (FTX), Jaan Tallinn, and others
• 2023: Strategic investments from Amazon ($4 billion commitment), Google ($2 billion+), Salesforce, Zoom, and others
• March 2025: Series E – $3.5 billion raised at a $61.5 billion valuation, led by Lightspeed, with Bessemer, Cisco, Fidelity, General Catalyst
• August 2025 (pending): New $3–5 billion round led by Iconiq Capital, targeting a $170 billion valuation; participants reportedly include UAE’s MGX, Singapore’s GIC, and Amazon again

That’s nearly a 3x leap in valuation in under five months.

For context, OpenAI took over seven years to reach a comparable figure, and with Microsoft’s heavy push. Anthropic’s done it in four, while maintaining its good industry reputation.

Iconiq Capital, the wealth management firm to the likes of Mark Zuckerberg, Jack Dorsey, and Reid Hoffman gives Anthropic the unofficial stamp of Silicon Valley royalty.

But there’s an irony here.

Anthropic was formed to escape the commercial pressure cooker of OpenAI. But in chasing scale, compute, and the resources needed to train frontier models, it has become one of the fastest growing companies, and most valuable in a short space of time, in human history.

A public benefit corporation, with a long-term trust model to prevent outside takeovers… now accepting billions upon billions from some of the richest and most powerful entities on the planet.

That’s not a criticism. That’s the reality of building AI these days. If the billions (and sometimes trillions) aren’t behind you, then have fun being irrelevant.

Training Claude 3.5 reportedly took over 250,000 H100 GPUs. The next versions could require twice that, and then some. This is a race where only the well-capitalised survive.

As I mentioned at the start, this story isn’t just about Anthropic and it’s lofty valuation.

It’s about the broader pattern we’ve been tracking in AI Collision follow the money to see how this revolution rolls out.

Because the money tells you where the big wins will come from.

The trillion-dollar action is in things like,

• Foundation models that will power everything from search to surgery
• Sovereign wealth funds buying AI equity for national strategies
• Big Tech fighting for AI supremacy through strategic investment and big talent spending

Anthropic sits at the intersection of all four.

So, while this $170 billion number might seem wildly over the odds, it’s a flashing signpost that screams the capital flow isn’t slowing down it’s going exponential.

#AD

Rice Krispies, Pringles & Extraordinary Gains

James Altucher’s AI would have flagged a junk food stock last year, Kellanova.
Two weeks later, it spiked. Traders could’ve seen extraordinary gains in a short time.

That same Blue Spike signal just flashed again… but this time, the target is a small AI chipmaker.

James is revealing everything… how to get the name of the stock, the setup, and the timing… on August 7.

Capital at risk. Simulated past performance is not a reliable indicator of future results.

Boomers & Busters 💰

AI and AI-related stocks moving and shaking up the markets this week. (All performance data below over the rolling week).

Boom 📈

• Rambus (NASDAQ:RMBS) up 17%
• Western Digital (NASDAQ:WDC) up 12%
• Meta (NASDAQ:META) up 8%

Bust 📉

• Micron (NASDAQ:MU) down 3%
• Hewlett Packard Enterprise (NYSE:HPE) down 4%
• IBM (NYSE:IBM) down 4%

From the hive mind 🧠

• This would be pretty cool I think. Really hard to do, but incredible if possible. The other thing is if they’re thinking about doing this, then what do you think is next? I say, an AI datacentre, and then we send AI to far off planets with the moon as a launchpad.
• Is this Perplexity hiding its actions, or is this AI hiding its actions? I think understanding the difference is far more important than first consideration would suggest.
• Good chance that Elon Musk’s Grok ends up destroying the porn industry. And that might not be as big a win as some people might think.

Artificial Polltelligence 🗳️

Weirdest AI image of the day

ChatGPT’s random quote of the day

Thanks for reading, and don’t forget to leave comments and questions below,

Sam Volkering

Editor-in-Chief

AI Collision