Rafomac News, Tech & Trends That MatterArtificial intelligence investing is undergoing a critical shift. The hype around large language models and hardware is fading as commoditization sets in, pushing profit potential toward software companies with strong data moats.
Three Layers of AI Profitability — Which One Matters?
Artificial intelligence has often been viewed through the lens of three profit layers: the large language models (LLMs) that everyone talks about, the software applications built on those models, and the underlying compute infrastructure that powers it all. Until recently, enthusiasm and investment dollars flowed heavily into the LLM and compute hardware layers, with software somewhat sidelined.
But this dynamic is shifting. The LLM layer, once the star of the show, is rapidly becoming commoditized—essentially turning into a common tool rather than a unique asset. Hardware too, while having seen explosive growth since 2023, especially with Nvidia’s GPU dominance and the current memory cycle, is showing the same signs of becoming a commodity. This leaves the application software layer standing as the most promising space for long-term profit and pricing power.
Why Software with Data Moats Holds the Edge
The key to surviving the commoditization of LLMs and compute is owning a proprietary edge in data and software. Take companies like Palantir and Axon as examples. Palantir’s locked-in government and corporate contracts create an entrenched ecosystem hard to exit. Axon’s exclusive access to data from bodycams and emergency calls gives it a defensible position humans find tough to replicate or replace.
On the other hand, companies purely selling LLM subscriptions or basic AI software face an uphill battle. As LLMs become interchangeable—like choosing a cola brand—it’s the unique data and application layer that creates lasting value. This perspective aligns with recent discussions on the All-In Podcast, where the hosts agreed that LLMs will commoditize, but highlighted that OpenAI and Anthropic’s long-term play is embedding themselves deeply within software applications to combat that trend.
Real Estate AI and the Difficulty of Commoditization
Not all data is easily distilled. Real estate is a prime example where the AI tools haven’t yet found their full footing. Evaluating properties requires a blend of hard-to-collect information: renovation expenses, comparable sales, location nuances. Unlike generic AI that can churn out text or images, real estate AI faces a complex, fragmented data challenge that makes commoditization tougher and preserves the advantage for startups focused on proprietary models.
This intricacy in real estate AI data contrasts with more general AI aims and explains why AI-driven real estate valuation technology has not yet exploded despite broader AI enthusiasm. It’s a niche where data moats prove valuable and sustainable.
What Investors Should Watch in the AI Sector
For investors, the takeaway is clear. When evaluating AI companies—whether OpenAI, Anthropic, or others—we should focus less on their LLM subscriptions or the flashy new AI model launches and more on how strong their software applications and data moats are. Many software companies will integrate multiple LLMs, selecting whichever fits best at the time, much like a utility. The real question is which companies will protect and monetise unique data or lock clients in through proprietary software experiences.
The debate between AI optimists and cautious pros boils down to this: new investors might chase the next AI IPO or chipmaker, hoping to ride a wave to the moon. The seasoned view suggests that while hardware and LLM layers may see their margins squeezed, software companies with true data leverage will be the big winners over the next decade.
Market Risks and Upcoming Catalysts
Beyond AI, a few other factors could rock the markets in the near term. Tensions with Iran remain unresolved despite a pause in strikes, hinting at possible renewed conflict. The Federal Reserve’s upcoming meeting is another wild card, with a 34% chance of a rate hike that could spook investors. Plus, earnings reports from giants like Microsoft and Meta are on the horizon, which could further sway sentiment.
All told, it’s going to be a nerve-wracking week, requiring a keen eye on fundamentals. But the wisdom around AI is especially worth heeding—especially if you’re looking to invest smartly for the years ahead.
