
Anthropic just went public in a blockbuster trillion-plus-dollar IPO. But is the $ANT stock a buy? In evaluating Anthropic as a potential investment, the big question is why they chose to go public now.
Anthropic's founder/CEO Dario Amodei is super scaling-pilled. He believes that as long as you keep scaling more compute, you'll get more intelligence. So far, that has proven to be correct. AI models keep getting better and better as the dataset and compute base get larger and larger. To win the race to AI superintelligence, he believes Anthropic needs the largest model with the most computation.
Dario is also super AI safety-pilled. That is why he left OpenAI, believing they weren't focused enough on safety and alignment. If Dario is so concerned about AI safety, why would he take his company public, where they will face more pressure from investors to forego safety for profits? And why IPO now, when AI stocks have crashed over the past two months? Originally a potential $2 trillion IPO, ANT is now closer to $1 trillion.
I think there are two possibilities for Anthropic to go public now.
First, Dario has decided that to win the AI race (against OpenAI, Google, China, and other actors he deems dangerous), Anthropic needs the extra funds that can come from public markets to accelerate AI progress and get the money to build a large enough cluster of compute to create superintelligence. It is winner-take-all. The first to superintelligence is the last. Many AI experts, particularly in the rationalist sphere the Anthropic folks come from, pinpointed 2027 as being the key year AGI takes off. That's why Anthropic is going public now—to try to win the AI race by scaling to superintelligence ASAP. While rushing to AGI isn’t the most safe route, Anthropic views a competitor getting there first even more dangerous, so it is worth the utilitarian risk. And Anthropic’s founders are notoriously part of the Effective Altruist movement (EA), rooted in utilitarianism. In Dario’s mind, “I alone can fix AI.”
But the second possibility is that Dario and Anthropic have finally started noticing diminishing returns with scaling. Perhaps despite having more compute, the most recent (currently private, internal to Anthropic) AI models are not dramatically better. Anthropic realized the current frontier is more or less as advanced as AI models will be. There may be modest improvements around the edges going forward, but this form of AI (Large Language Models) is not the ultimate route to superintelligence.
If that is the case, Anthropic is going public now to try to cash out on their sky-high valuation—even though it's a trillion less than it was two months ago—before it collapses even further when everyone else realizes the scaling curve has tailed off. It won't be a complete collapse because Anthropic still has great AI models that are highly capable and will be useful, but Claude won't get dramatically better from here. Scaling LLMs won't yield superintelligence.
If that is the case, only those at the frontier (OpenAI and Anthropic) would know it. Maybe that’s why OpenAI delayed their IPO. Or maybe only Anthropic knows it, and OpenAI hasn’t discovered the end of scaling yet. Either way, this is the Anthropic founders’ best chance to cash out and make the most money possible to devote to their Effective Altruism causes before everyone else realizes that the scaling law hypothesis has reached its endpoint. SBF already proved EAs are willing to scam public investors out of money if they think they can spend that money on greater good.
If scaling has reached its limit, that would mean that other AI companies can catch up to where Anthropic and OpenAI are now. The current frontier is basically the apex, meaning Anthropic will soon lose their lead. Rather than a winner-take-all scenario, there will be many different successful AI companies, which would lower Anthropic's market share and valuation. There won't be the doomer scenario where super AI kills us all, but neither will there be the utopian scenario where AI makes us immortal.
At least those are two possible futures. The question of whether to invest in Anthropic is which of those scenarios is true. Will scaling bring superintelligence, or has scaling reached its end? Though even in the first case, that still doesn't mean Anthropic will win the race against OpenAI (or China or Google or Meta or Grok). But if they do, $ANT may be the best investment ever. If not, Anthropic is just another AI company. Only time will tell.