Home

Something quite strange happened last week. Anthropic’s Dario Amodei argued that AI companies need to “pace the frontier”, essentially slow down how quickly they make their most powerful models more capable. Sam Altman agreed with him.

Think about that for a second.

OpenAI and Anthropic are spending billions of dollars trying to beat each other. They are fighting for the same users, developers, talent and, ultimately, the chance to build AGI first.

And now they agree that perhaps everyone should slow down. This makes a lot more sense when you look at the AI race as a classic prisoner’s dilemma.

Assume OpenAI believes pushing AI too fast has real risks. Anthropic believes the same. Both would probably prefer a world where everyone spends more time testing models instead of shipping the next capability jump as fast as possible. This gives both enough time to prep up for enterprise adoption, which has been abysmally slow.

But neither can afford to slow down alone. If OpenAI slows down and Anthropic keeps going, Anthropic gets ahead. If Anthropic slows down and OpenAI keeps going, OpenAI gets ahead.

So both keep racing.

The result is quite sad. Both companies can end up doing something neither actually wants to do.

Which is why Dario’s proposal is interesting.

Except there is a third prisoner.

And this prisoner isn’t sitting at the table.

Over the last two years, Chinese labs have become a serious force in open AI. DeepSeek changed the discussion around the cost of building capable models. Alibaba’s Qwen family has become one of the most widely used open model families. Moonshot, Zhipu and others continue to push models that developers can download, change and run themselves.

I would not claim that these models have simply “beaten” OpenAI or Anthropic. That is not true across the board. But they have definitely changed the economics of the race.

OpenAI and Anthropic are no longer competing only with each other. They are competing with models that can get close enough on many tasks, cost far less to use, and in some cases can run without sending anything to OpenAI or Anthropic.

And that creates a much harder prisoner’s dilemma.

Sam and Dario can agree to slow down together.

They cannot make DeepSeek slow down.

They cannot make Qwen slow down.

This is where the AI safety argument starts getting very interesting.

Suppose the industry agrees that powerful AI models need outside tests before release. They need ongoing checks. Their makers need to know who can access them.

And if something goes badly wrong, the maker should have some way to restrict access.

An open-weight model cannot offer the same level of control after someone downloads it (who is going to take the blame in an org? the CTO? In that case, why would he/she even push open source models?)

Nobody has to say “ban open source.”

OpenAI and Anthropic can genuinely believe that advanced AI needs stronger safety rules. At the same time, those rules can strengthen the business model of OpenAI and Anthropic.

Both things can be true.

This does not mean Sam Altman and Dario Amodei are secretly trying to kill open source. There is no evidence for that. In fact, OpenAI itself has released open-weight models, and Dario has said he does not support banning them.

The more useful question is not what they intend. It is what happens if their view of responsible AI becomes the rule.

Because the definition of “safe AI” could slowly become a definition that favours models which remain under the control of their makers.

And then we have an even bigger problem.

Imagine OpenAI and Anthropic really do slow down. They spend more time testing frontier models. They delay releases when the risks look too high. They follow every rule they helped create.

Meanwhile, Chinese open models keep improving.

At some point, one of the American labs looks at the capability gap and asks the obvious question:

Why are we slowing down when they aren’t? And the prisoner’s dilemma starts all over again.

Except now it is not OpenAI versus Anthropic. It is closed frontier AI versus anyone who refuses to join the agreement. Which may be the real problem with trying to pace AI.

It works only if enough of the people who can push the frontier agree on what “pace” means. Sam and Dario can make peace with each other. They cannot make peace on behalf of everyone else.

And in a world where everybody wants to rule the world, that may be the part that matters most.

What’s your take?

Y Combinator | 21 min

YC Visiting Partner Vivian Shen sits down with Paul Graham at the original YC office in Mountain View to talk about startups, AI, ambition, and what makes great founders. This analysis reveals that while the scope and ambition of YC-backed startups have grown significantly, tackling complex problems from intercontinental logistics to cancer research, the fundamental drivers of success are unchanged. True ambition is an inherent quality in founders, often suppressed but not created. The startup journey is inherently difficult, making it unsuitable for those seeking easy credentials.

True Ambition is Inborn, Not Taught

Ambition is largely an inborn trait. While some individuals may appear to lack ambition, it's often a result of being conditioned to suppress their natural drive, particularly in environments that prioritize obedience over independent action.

This inherent quality is crucial for navigating the challenges of a startup.

Understanding that ambition is innate helps in identifying high-potential founders by looking beyond superficial presentations and recognizing the underlying drive. It also suggests that attempts to 'teach' ambition might be less effective than creating environments where it can flourish.

When evaluating founders, look for signs of inherent drive and a history of pursuing their own path, even if it was previously constrained. Mentors and investors should focus on unblocking suppressed ambition rather than trying to instill it from scratch.

Startups Offer No Shortcuts to Credentials or Coolness

The startup journey is characterized by brutal difficulty, demanding immense hard work, cleverness, and determination. It offers no shortcuts to credentials or 'coolness,' with any recognition coming only after years of arduous effort.

This contrasts sharply with the perception of startups as a trendy career choice.

This principle serves as a crucial filter for aspiring founders, deterring those with superficial motivations and highlighting the deep commitment required for success. It underscores that genuine motivation must stem from a desire to solve problems, not from external validation.

Founders should introspect deeply about their motivations. If the primary drivers are external validation or an easy path to success, they are likely to fail.

Investors should probe founder motivations to ensure they are rooted in genuine problem-solving and resilience.

If you want to seem cool, like starting a startup is just about the least efficient way to do it.

Formidable Founders Consistently Get What They Want

Formidability is not about charisma or specific skills, but about a proven track record of getting what one wants in any situation. This quality is highly attractive to investors because a formidable founder's success directly translates to the investor's success, creating a strong alignment of interests.

This provides a clear, actionable criterion for investors to identify high-potential founders, moving beyond subjective assessments to a results-oriented evaluation. For founders, it emphasizes the importance of demonstrating a consistent ability to execute and achieve goals.

Founders should focus on building a reputation for execution and goal attainment. Investors should prioritize founders who can demonstrate a history of achieving their objectives, as this is a strong indicator of future success.

I said, I think that it's someone who who gets what they want. the test, right?

AGI is a Spectrum, Not a Finish Line

The traditional view of AGI as a distinct point to be crossed is inaccurate. Instead, AGI manifests as a continuous, multi-dimensional progression where various aspects of AI achieve human-level or superior intelligence at different rates.

We are currently 'on the smear,' experiencing this uneven development.

This reframes the discussion around AI progress, moving away from binary 'achieved/not achieved' thinking to a more nuanced understanding of its gradual and multifaceted development. It helps manage expectations and guides research and investment towards specific capabilities rather than a monolithic goal.

Researchers and developers should focus on advancing specific AI capabilities rather than waiting for a single AGI breakthrough. Investors should evaluate AI companies based on their progress within this 'smear' rather than expecting an all-encompassing AGI solution.

line is actually this sort of smear. I think the best answer you can give is like we're on the smear.

Shipping Pace Remains the Ultimate Predictor of Startup Success

Despite the availability of powerful AI tools that can accelerate development, the fundamental differentiator for successful startups is their ability to rapidly iterate and release new offerings. This pace reflects not just production efficiency but also the capacity for generating and executing on new ideas.

This principle reinforces the enduring importance of execution and agility in the startup world. It clarifies that technology, while enabling, does not replace the core entrepreneurial drive to build and deliver quickly.

For founders, it provides a clear metric to focus on.

Founders must prioritize a culture of rapid shipping and continuous iteration. AI tools should be leveraged to enhance this speed, but the underlying commitment to quick delivery and idea generation is paramount.

Investors should evaluate a startup's shipping velocity as a key performance indicator.

Some might believe that with AI, the focus shifts from speed to pure innovation or quality, but this suggests speed remains king.

said is that the best predictor of success for a startup is the pace that they ship new stuff.

Watch the full conversation on YouTube