Anthropic's debut on the Nasdaq tomorrow represents a watershed moment for AI startups, but the company's staggering $1.4 trillion valuation and $88.5 billion IPO raise should prompt serious questions about whether the massive capital requirements of AI businesses can actually translate into sustainable profits. Yes, the numbers are jaw-dropping: $17 billion in Q3 revenue less than four years after Claude launched. Yes, Claude Code achieved the fastest enterprise adoption on record with 11,900 companies spending $100,000 or more in the trailing 12 months, up 12x year-over-year. But there's a number in Anthropic's S-1 that deserves far more scrutiny than the headline revenue figures—the nearly $900 billion in compute purchase and lease commitments.

That commitment dwarfs the IPO proceeds and raises an uncomfortable question: Can Anthropic actually monetize Claude and Claude Code fast enough to justify that level of infrastructure spending? The company's $17 billion revenue is impressive in isolation, but it's also the gross top line before the astronomical costs of running and scaling AI infrastructure. Anthropic must prove it can convert its enterprise traction into profit margins that justify betting nearly $900 billion on compute. For context, that's roughly 10 times the IPO's total capital raise.

The company's founders and investors—including Google, Amazon, Spark Capital, Menlo Ventures, Lightspeed Venture Partners, Coatue, and Sequoia Capital—are positioning this as a paradigm shift. And they may be right. Anthropic's focus on AI agents for knowledge work and scientific discovery by 2027, coupled with potential applications in biology, rare-disease research, and healthcare access for the 4.6 billion people the World Health Organization says lack basic health services, represents genuine transformative potential.

But Anthropic's own S-1 filing lays bare the risks that could derail that vision: cyber attacks, biological attacks, unpredictable model behavior, authoritarian surveillance, and economic disruption. The company's public-benefit structure—where the Long-Term Benefit Trust participates in governance without financial stake, weighing impacts on national security, economics, global health, and science—is intended to balance these tensions. That's commendable, but structural governance doesn't eliminate the underlying risks.

As Anthropic trades tomorrow, investors and observers should focus less on comparing its user adoption to OpenAI's ChatGPT milestone and more on whether the company can deliver returns that justify its compute commitments. The AI era may indeed generate massive new wealth, but Anthropic needs to prove its business model isn't just a spectacularly scaled bet on perpetually declining infrastructure costs and infinitely growing demand.