An Unexpectedly Profitable Way to Save the World

There is something inherently awkward about setting out to save the world and accidentally becoming a decacorn in the process. It is rather like trying to write a stern letter to the local council about a loose paving stone and having them respond by naming a small moon after you. Anthropic, a company that began its life with the quiet, studious intensity of a group of people who really, really didn't want the robots to eat us, has found itself in just such a predicament. Recent reports suggest the firm is on track to generate some $10.9 billion in revenue this quarter. For a company that prides itself on being a 'Public Benefit Corporation,' this is an awful lot of benefit to have to manage without accidentally buying a private island.

One must imagine the internal meetings at Anthropic are currently a delicate dance of existential dread and fiscal exuberance. It is difficult to maintain a somber tone about the impending silicon-based apocalypse when the accounts department is frantically trying to find enough digital filing cabinets to store the incoming cash. There is a certain irony in the fact that the more one talks about the dangers of artificial intelligence, the more the market seems determined to throw money at it. It is as if the world has decided that if we are going to be replaced by algorithms, we might as well ensure the algorithms have an excellent credit rating.

(I once spent an afternoon trying to convince a self-checkout machine that I had, in fact, paid for my leeks. It remained unconvinced, despite the clear evidence of the receipt. I suspect that if I had offered it ten billion dollars, it might have been more amenable to my perspective, or at least allowed me to leave the store without the alarm sounding.)

The velocity of this revenue growth is, by any reasonable standard, slightly ridiculous. Traditional software companies usually spend decades slowly climbing the mountain of profitability, pausing occasionally to have a sensible lunch and perhaps a round of golf. Anthropic, by contrast, appears to have strapped a rocket to the mountain and decided to simply fly over it. This creates a unique set of bureaucratic challenges. How, for instance, does one ensure that a ten-billion-dollar revenue stream remains 'safe'? Is there a specific type of guardrail for a balance sheet that is growing faster than the speed of light? One suspects the safety team is currently looking for a way to 'align' the profit margins with human values, which is a noble goal, though perhaps difficult to explain to the shareholders.

There is also the matter of the 'Public Benefit' status. In the grand tradition of institutional absurdity, we are now faced with the prospect of a company that is legally obligated to do good, while simultaneously being financially obligated to dominate the known universe. It is a bit like a charity shop that accidentally discovers cold fusion in the back room and has to decide whether to keep selling second-hand cardigans or simply solve the global energy crisis. Anthropic has chosen a middle path: they will continue to build very smart models that tell us how to be better humans, and they will charge us a very large amount of money for the privilege.

(Reflecting on this, I am reminded of a time I tried to organize a neighborhood watch program. We spent three weeks discussing the proper color for the high-visibility vests and completely forgot to actually watch the neighborhood. We were, however, very safe in our deliberations.)

Human behavior in the face of such technological shifts is always a source of quiet amusement. We are currently witnessing a global scramble to secure a piece of the AI pie, even as we collectively worry that the pie might be poisoned. It is a classic case of wanting to have our cake and eat it, while also ensuring the cake doesn't develop a consciousness and decide it would rather eat us. The financial markets, ever the pragmatists, have decided that the risk of a sentient cake is a small price to pay for a 15% jump in share price. It is a comforting thought, in a way; even at the end of the world, someone will be there to check the quarterly earnings.

Ultimately, the rise of Anthropic as a financial powerhouse tells us more about ourselves than it does about the technology. We are a species that values safety, but we value a good return on investment even more. We want the future to be secure, but we also want it to be delivered in high-definition with a subscription model. As Anthropic navigates its new life as a ten-billion-dollar safety manual, we can only hope they remember their original mission. After all, it would be a shame to save the world and then realize we can't afford the licensing fees to live in it.