Anthropic’s coming IPO is a risky bet. Costs are rising and cheaper, capable competition is increasingly available. I wouldn’t invest any money I couldn’t afford to lose. But there could be tremendous, once-in-a-generation upside. I believe Anthropic will lead a business transformation more profound than the Industrial Revolution.
I’ll share my coffee-and-naive-optimism-driven thinking on these points: Vision, Revenue, and Risks.
Vision
I think “smart” people analyzing Anthropic through the lens of token maxing are missing a critical point.
The vision Anthropic expresses in its S-1 is directionally right: AI is not an application or an employee. It will augment and replace both as smart leaders weave AI tools into management, execution, and service functions. Cheaper, open-weight models and tools may drive operational, logistical functions, but all data will flow into a trusted AI-powered brain that continuously provides insight into execution while identifying risks and opportunities. This brain will be the basis for all strategic decisions.
I doubt a major company will trust this to a free, open-weight system even if delivered by a trusted partner with a warranty.
Early on, Anthropic focused on the enterprise and has been continually rewarded with tremendous growth within these customers. It’s also developed, I believe, a rapidly growing skill set for helping customers with this transformative tech. I assume their two-year-old Customer Success function is turning this experience into mature models and playbooks.
In the very near future, like next week or next year, AI-enabled employees will become the Tony or Antonia Stark of their functions. I’m betting Anthropic will become a factory for delivering these highly valued team members.
It may take business leaders months, or a year, to learn how to wield this transformational technology, but once they begin, it will happen at AI speed.
Revenue
Most analysis has focused on this critical component. The question is simple: can Anthropic grow revenues fast enough while avoiding self-owns (attractive spys, etc)?
I’m supremely unqualified to answer that.
According to its S-1: By September 2026, its customers spending at least $100,000 annually included every Fortune 10 company, more than half the Fortune 500 and over 30% of the Global 2000. It believes that many of these companies have barely begun deploying AI throughout their organizations. In other words, Anthropic believes it can grow substantially even without acquiring new customers, simply by expanding its presence inside existing ones.
It seems to be working. The company reports 900% net retention–that means as customers use Anthropic’s AI products, they want more and pay more. This is the goal of any business.
I spent my career in startups and smaller companies that sold to enterprises. One thing is very true: big companies hate change and risk. If Anthropic becomes integrated into an organization’s operations and delivers sufficient value reliably, displacing it will be very difficult.
Risks
There are many, but here are a few I’ve thought about.
Competition
Free and cheaper competitors are readily available and multiplying. Open-weight models are increasingly powerful. Hugging Face, the leading AI community for open-source solutions, makes it easy to find and deploy capable solutions.
However, as someone whose hobby is AI, I experience daily just how much time and energy (brain on fire) it takes to work with a SOTA coding agent like Claude to make a functional application. Sure, many AI coding agents can build an app but when you try to use it, you understand how many decisions you have to make to turn that app into something useful.
No Moat
If you think of Anthropic as a token vendor, you are right. If you see it as an emerging solution for enterprise management, then I completely disagree. Its early focus on the enterprise has given it a lead that will be hard for others to overcome before it becomes the standard for managing innovative businesses that operate faster and more effectively.
Costs and Citizen Sentiment
No denying costs are a huge risk. Equally true is that with continued investment, costs decrease. Anthropic and others are finding ways to decrease training costs, for example. I liken AI to solar energy, which has become one of the cheapest forms of power while improving its efficiency.
Anthropic’s hybrid approach to data centers is smart: it uses hyperscalers and is building its own. It will learn to navigate this as AI becomes more and more embedded in society. I don’t know the answers, but I’m guessing we can have clean, quiet cities and sufficient compute.
Skynet Will Kill Us
Not really an Anthropic-specific risk, but I hear it a lot when folks talk about its IPO. It’s more likely Anthropic will emerge as a hero as companies rely on more sophisticated tools and experts to combat AI-powered threats. One thing is certain: more than rogue AI, I worry about BAD HUMANS. They will do what they always do: use a new technology to do serious damage for personal gain.
Bottom Line
AI will forever transform our world, and there is no going back. That doesn’t mean Anthropic will succeed. We could find out Dario and his team are secret furries hellbent on destroying the world. I doubt it. I’m optimistic that AI, in the right hands, can bring tremendous benefits to people, companies, and societies that embrace it and make laws for its use.
I think a small bet on Anthropic could be a good long-term investment.




