Public transcript

Anthropic CEO Dario Amodei on AI race: Focused on making our models as smart and capable as possible

By: CNBC Television
Length: 7 min
Language: English
Text from: the video's captions
Transcribed: 12 October 2026

Read the full transcript Open on YouTube

Notes in English.

Summary

Anthropic CEO Dario Amodei explains how Anthropic's enterprise-first strategy differs from consumer-focused rivals like Google and OpenAI. He discusses the company's rapid revenue growth, the need for AI benefits to diffuse broadly, and the capital-raising challenges created by uncertain future revenue. He also speculates that the AI industry could eventually capture trillions of dollars in annual revenue.

Key points

Amodei says Anthropic focuses on safety and reliability, which is synergistic with enterprise customers, and the split is roughly 80/20 toward enterprise over consumer.
Anthropic avoids ads, short-form video, and engagement maximization, instead focusing on making models as smart and capable as possible for business use.
Amodei argues consumer-optimized competitors lack the same incentive to make models broadly smarter and more capable.
Anthropic's revenue grew from 0 in 2023 to 100 million in 2024, to 1 billion in 2025, and now to almost 10 billion.
Amodei says AI must diffuse beyond Silicon Valley startups to large enterprises and the developing world, including work with the Gates Foundation.
He estimates total labor in the economy is about $50 trillion a year, and AI industry revenue could plausibly reach $5 trillion a year.
Higher margins and more predictable enterprise spending, plus periodic fundraising, buffer Anthropic against uncertainty about how much compute to buy.
Amodei acknowledges he never expected to spend so much time raising capital.

Questions it answers

00:00Anthropic's different path

How does Anthropic's approach differ from other AI companies?

Amodei says Anthropic's focus on safety and reliability is synergistic with enterprises, so it pursues an 80/20 enterprise-over-consumer strategy and focuses on making models as smart and capable as possible rather than maximizing engagement.

Anthropic's split is roughly 80/20 enterprise over consumer.
It avoids ads, short-form video, and engagement maximization.
Its work targets finance, coding, pharma, manufacturing, and energy.

02:02Divergence from consumer AI

Will consumer-focused models catch up to Anthropic's models?

Amodei argues there is a divergence: consumer-optimized models prioritize being engaging and entertaining, and he believes those players lack enough incentive to get smarter and more capable across the board.

Consumer models optimize for engagement and entertainment.
Amodei thinks rivals lack incentive to become broadly more capable.

02:42Revenue growth and the bubble question

Is AI investment a bubble if adoption doesn't broaden?

Amodei cites Anthropic's revenue growth from 0 in 2023 to almost 10 billion now, saying if that curve continues the investments will be justified, while stressing the need for diffusion to large enterprises and the developing world.

Revenue went 0 to 100 million in 2024, to 1 billion in 2025, to almost 10 billion.
Enterprise adoption is harder than startup adoption.
Anthropic has started work with the Gates Foundation.

03:59Ten-year revenue estimate

How big could AI revenue get over ten years?

Amodei says all labor in the economy is about $50 trillion a year, and if AI captures even 10% of that, industry or single-company revenue could reach $5 trillion a year, something unseen in history.

Total labor is estimated at $50 trillion a year.
10% capture could mean $5 trillion a year in revenue.

04:59Capital raising and uncertainty

How does Anthropic manage raising capital amid uncertain revenue?

Amodei describes a cone of uncertainty where next year's revenue could be 10, 20, or 100 billion while data centers must be bought ahead of time; buffers include higher margins, predictable enterprise spending, and periodic fundraising.

Enterprise spending is more predictable than fickle consumer spending.
Anthropic raises money periodically while growing.
Amodei didn't expect to spend so much time raising capital.

Notes

Anthropic's Enterprise-First Strategy

Amodei says Anthropic has thought in terms of safety and reliability since the beginning, which is synergistic with enterprises that value reliability, trust, and long-term relationships.
The company's focus is roughly 80/20 enterprise over consumer.
This lets Anthropic avoid ads, short-form video, and engagement maximization, focusing instead on making models as smart and capable as possible.
Target use cases include finance, coding, pharmaceutical companies, manufacturing, and energy — work Amodei describes as directly driving GDP.

Divergence from Consumer-Focused Competitors

Amodei sees a divergence: consumer-optimized models aim to be engaging and entertaining, and he argues those players lack strong incentive to get smarter across the board.
He believes Anthropic has been able to focus more on general capability.

Revenue Growth and the Bubble Question

Addressing Satya Nadella's point that AI could be a bubble if adoption doesn't broaden beyond big tech and wealthy nations, Amodei cites revenue growth: 0 in 2023, 100 million in 2024, 1 billion in 2025, and almost 10 billion now.
He says if that curve continues, the investments will be justified.
Diffusion matters: Silicon Valley startups adopt AI immediately, large enterprises are harder, and developing countries need access too — Anthropic has begun work with the Gates Foundation.

Long-Term Revenue Potential

Amodei estimates all labor in the economy is roughly $50 trillion a year.
If AI captures even 10% of that, industry or even single-company revenue could be $5 trillion a year — unprecedented in world history, creating growth as well as problems.

Capital Raising and the Cone of Uncertainty

Amodei describes a "cone of uncertainty": revenue could be 10 billion, 20 billion, or 100 billion next year, yet data centers must be bought in advance.
Buffers include higher margins and more predictable enterprise spending compared to fickle consumer spending, plus periodic fundraising.
He admits he never expected to be a capital raiser.

Text from the video's own captions. Summary and notes written by AI from the transcript, so check anything important against the recording.