Anthropic: how to build a mission that lasts

We write these Culture Field Notes to surface real-time examples of culture as performance drivers. Our aim: to learn out loud from organizations we've studied closely, and share what might help others building enduring, differentiated companies.

As investors, we've learned to watch for a particular type of founder: those who've lived through cultural breakdowns firsthand. They engineer their next companies to deliberately avoid the cultural pitfalls they witnessed up close.

"We've seen the story so many times," reflects Dario Amodei, co-founder and CEO of Anthropic. "There's a tech company, it made all these idealistic promises, and then it hit the realities of the market. That's the story of social media and that's basically the story of OpenAI. We are absolutely determined to have it not happen to us too."

Amodei is exactly this type of founder. At OpenAI, he helped build one of the world's leading AI labs, then watched its ideals strain under competing visions and commercial pressures. When he and key members of the GPT-3 team left to start Anthropic, it wasn't just talent leaving – it was a cultural split over something urgent. Many experts believe extremely powerful AI systems may arrive within one to three years, while the tools to reliably interpret and steer them could take five to ten. Without those tools, even well-intentioned AI systems could pursue goals in ways their creators never intended, and developers would lack defenses against bad actors who might exploit powerful models for harm. This timing gap made Anthropic's mission – to build frontier AI systems that are more understandable and steerable – feel less like corporate positioning and more like necessity.

What makes Anthropic fascinating from our perspective isn't just their technical work. It's how they've turned mission integrity into competitive advantage at a time when AI model development depends on retaining specialized researchers who are exceptionally scarce. Many first saw Anthropic’s focus on safety as a handicap that would slow them in a race defined by speed. Instead, it became the reason researchers choose them over bigger paychecks.

The data tells the story: Engineers are roughly eight times more likely to move from OpenAI to Anthropic than the reverse, and nearly eleven times more likely to leave DeepMind for Anthropic.² While Meta's massive compensation packages have proven compelling to some in the industry, Anthropic is remarkably competitive in the broader talent competition despite being vastly outspent.

This talent advantage reveals something important about sustainable cultural advantage. Plenty of tech companies have compelling missions that attract great people. What makes Anthropic different – and what gives us confidence as investors – is how they’ve systematically engineered their mission to last.

The hard part isn't writing a mission that attracts talent. It's keeping it from drifting when pressure mounts.

When missions bend

The pattern Amodei described plays out with predictable mechanics: tech companies launch with bold promises, only to see missions give way under pressure – impatient investors, new revenue opportunities, partnerships that pull incentives off course. Drift doesn't hit all at once; it creeps. A compromise here, a pivot there, each decision seemingly rational in isolation but collectively eroding the original vision.

Companies can still succeed while their missions drift – Big Tech has proven that – but they pay a price in coherence, trust, and alignment. Those are exactly the weapons younger challengers use to take them on. Mission-driven talent becomes skeptical, customers lose faith, and organizations fragment around competing priorities.

Smart challengers succeed by deliberately avoiding the specific cultural weaknesses that tax incumbents. Adyen’s founders exemplify this approach. Having witnessed bureaucratic dysfunction at legacy payments firms firsthand, they didn’t just launch a new company, they built their next company to avoid those same pitfalls – creating a flat, execution-first culture that stripped away friction. Even the name means "start again" in Surinamese.

Anthropic shows the same instinct, pointed at a different risk. The company represents a preemptive defense against mission drift – anticipating pressures that cause it and building safeguards from day one.

Three defenses against drift

Clarity over abstraction

From the start, Anthropic's founders rejected lofty slogans for something testable: "We aim to build frontier AI systems that are reliable, interpretable, and steerable."

"We didn't take the attitude of, 'What's the most crazy, idealistic thing we can say to recruit as many people as we want?'" Amodei explains. "That we very deliberately did not want to do."¹

The language isn't built for press releases – that's the point. Missions that overpromise invite cynicism and broken promises. Anthropic chose a tighter frame so progress can be tracked, leaders can be held accountable, and everyone knows what "good" looks like.

Mission and business as allies

Rather than treating safety and commercial success as uncomfortable tradeoffs, Anthropic's founding documents put them on equal footing as mutually reinforcing aims. Revenue funds deeper safety research. Market relevance expands influence over industry standards. Greater market share means more safety-aligned systems in use.

The alignment isn't always neat. When safety and business pull in different directions, Anthropic chooses carefully when to be idealistic and when to be pragmatic, earning internal trust through transparency and honest adaptation rather than either compromising principles quietly or letting idealism paralyze progress. As one former employee explained: "These are the kind of people that just want to be treated as intelligent, rational, high-integrity people. When decisions need to be made, any smart person would have the resources to go all the way to first principles so they could believe them for their own reasons."³

By refusing to let their pro-social mission stand apart from business reality – and adapting honestly when conflicts arise – Anthropic made their mission easier to live and less likely to sow division.

Structure over intent

Culture alone won't stop drift. Anthropic backed intent with governance, incorporating as a Public Benefit Corporation that gives the board authority to put safety ahead of returns when needed. Within four years, an independent Long-Term Benefit Trust will control a majority of board seats through special "Class T" shares – a phased safeguard against future misalignment.

"The way we've structured Anthropic, with these deliberate checks and balances, makes it much harder for something like [OpenAI's turmoil] to occur,"¹ Amodei notes.

This may look extreme, and not every company needs it. But when intent falters, governance provides a backstop.

How defense became offence

These safeguards faced their first major test in 2022, when Anthropic had to choose between speed and safety – and turned what looked like a costly delay into their defining advantage.

Anthropic was positioned to beat OpenAI to market with the first public large language model. Instead, OpenAI launched ChatGPT and seized the spotlight while Anthropic held back to complete safety checks. Many concluded Anthropic had been too cautious – proof that safety would slow them down in the AI race.

Inside the company, the decision was never seen as a mistake. It was proof of their commitment: moving fast matters, but only if you can do so responsibly. Rather than concede defeat, teams were determined to prove safety and speed could coexist. They adapted, running safety reviews and product development in parallel – an approach that let them maintain speed without compromising safety principles, a demanding rhythm only sustainable with deep alignment.

One employee described the intensity: "We would parallelize and do it as fast as we possibly can, and give people a few days off later because the two weeks before damn near killed everyone, but the safety tests need to happen."³

That same alignment showed up beyond safety work, driving discretionary effort on the business side too. "Someone would turn out some really cool thing on the weekend and dump it in Slack…by Monday, people were already thinking about productization."³ This kind of extra effort demonstrated how mission alignment also accelerated commercial progress – turning what many saw as a constraint into a meaningful advantage.

What began as operational drag became competitive drive.

The talent advantage

The mission that initially slowed Anthropic has since become their edge in competition for AI researchers, allowing them to compete in a way that doesn’t rely on outspending rivals.

The company hasn't been immune to talent losses – Meta's extraordinary offers have proven compelling to some. But Anthropic's people are motivated by more than compensation. "These were people who were not just trying to pump their equity prospects... They were jazzed about AI in a technical, philosophical way that would be hard for other people to relate to,"³ explained one former employee. As Amodei put it, Meta is trying to buy alignment that can’t be purchased⁴ – the kind of philosophical commitment that creates stickiness compensation packages alone struggle to replicate.

For us as investors, this is the heart of the asymmetric story. Anthropic cannot outspend Meta, but its mission gives it a credible way to keep the scarce researchers it needs to succeed. What began as a perceived handicap became a critical edge that extends beyond today’s market conditions.

Why mission durability matters

This talent advantage illustrates a broader principle about competitive durability in volatile markets. We've lived the narrative whiplash of AI markets – the "Deepseek moment" convincing everyone efficiency would define the field, followed only months later by markets cheering Broadcom, Oracle, and Nebius as beneficiaries of unrelenting compute demand. Such reversals are the reality of investing at the frontier of technological disruption. Accepted truths expire quickly, business models are barely formed, and conventional wisdom adjusts to justify whatever outcomes just played out. This is why we've learned to focus on sources of compounding rather than expiring knowledge. While market predictions quickly become obsolete, the understanding of cultural dynamics gets stronger with each application, providing a stable lens for evaluating companies amid the noise. In environments rich with opportunity but clouded with uncertainty, this means looking for enduring cultural advantages, not just today's market predictions.

Many companies have compelling missions. Fewer have built those missions to last. Anthropic's structured approach to mission durability represents one powerful form of cultural advantage – giving us conviction that their talent advantage won't erode during an era of model R&D when elite researchers are a critical competitive factor.

What we've observed goes beyond mission design – it's about sustaining competitive advantage through changing conditions. While narratives shift and technical approaches evolve, the underlying need for aligned, mission-driven talent remains constant. Anthropic's cultural coherence represents something rare – an organization that gets stronger under pressure rather than fragmenting.

This insight extends beyond investment analysis. For founders, it’s the difference between building a mission that attracts great people today and building one that keeps them tomorrow.