AI’s Next Act: From Subscriptions to Selling Outcomes
Frontier AI companies are shifting their business models away from subscriptions, APIs, and token sales toward directly selling research and commercial results. This evolution, discussed by Rick, suggests that the AI arms race will not halt even if short-term annual recurring revenue (ARR) growth slows. Instead, alpha is converging from a broad “all AI rises” narrative to specific sectors: storage, manufacturing, equipment, and biology. The market, however, is still pricing this new commodity using old software frameworks, creating mispricing and opportunity.
The Shift to Outcome-Based Pricing
Historically, AI companies monetized through subscriptions (e.g., ChatGPT Plus) and API calls priced per token. But as models become more capable, the value lies in delivering tangible outcomes—such as accelerated drug discovery or optimized supply chains. Rick argues that this transition will redefine revenue models and valuation metrics. Anthropic’s IPO valuation, for instance, reflects expectations of such outcome-based revenue, not just user growth.
Compute Arms Race Continues
Despite potential ARR slowdowns, investment in compute infrastructure remains robust. The race for AI supremacy is a multi-year cycle, and companies are building moats through proprietary data and specialized hardware. This capex-heavy phase benefits storage and equipment providers, as well as manufacturers of AI chips and servers.
Alpha Convergence
The “all AI” trade is over. Rick identifies four areas where alpha is concentrating:
- Storage: Data storage and management solutions for training and inference.
- Manufacturing: Companies that produce AI hardware components.
- Equipment: Semiconductor equipment makers and specialized machinery.
- Biology: AI-driven drug discovery and biotech platforms.
These sectors are poised to outperform as AI integrates deeper into the economy.
Memory Cycle and Crypto Risks
Rick also highlights the memory cycle—the cyclical demand for DRAM and NAND—which is influenced by AI workloads. Additionally, he warns of crypto risks, particularly how tokenized AI projects may be overvalued if they rely on outdated pricing models. The intersection of AI and crypto is still nascent, but projects that tokenize compute or data could benefit if they align with outcome-based pricing.
Forward-Looking Perspective
Investors should watch for AI companies that move beyond subscriptions to outcome-based contracts, as these will command premium valuations. The compute arms race will continue to fuel demand for hardware and storage, but alpha will be sector-specific. Crypto projects that tokenize AI infrastructure or data must adapt to new pricing paradigms to avoid being left behind. As the market matures, the old software framework will give way to a more nuanced valuation approach—one that recognizes AI as a general-purpose technology with diverse revenue streams.
