DeepSeek Shows the Continued Commoditization of LLMs, as Predicted

In my previous extensive notes (Part 1 and Part 2) on Palantir ($PLTR), I articulated a core thesis: large language models (LLMs) are on an inexorable path toward commoditization.

The real value from AI does not lie within the models themselves but rather in the ability to process, deploy, and integrate them into enterprise operations. The emergence of DeepSeek has only solidified this position, providing yet another proof point for what Palantir’s leadership, and particularly Alex Karp, have been forecasting for quite some time.

The New AI Landscape: Commoditization and Palantir’s Role

AI models are increasingly becoming commodities, akin to raw materials like hydrocarbons. As Alex Karp noted in April 2023..

"People have it wrong. They think all the value is in the LLMs. AI models are like hydrocarbons in the ground. They need to be processed.”

The LLM alone does not create value; it must be refined, controlled, and aligned with the unique business logic of an enterprise.

This perspective has been underscored by DeepSeek’s rise. With a reportedly modest budget of less than $10 million, the Chinese research lab developed a GPT-4-class model, leveraging algorithmic efficiencies that challenge long-held assumptions about the high costs of AI development. Keep in mind though that skepticism remains regarding this reported figure, so take it with a grain of salt. Brad Gerstner of Altimeter Capital pointed out that “The reports they spent $6M vs. billions are fake news. The actual cost comparison is closer to $6M vs. $15M, and even this reflects falling compute costs over the past year.” Such context tempers the initial excitement over DeepSeek’s purported breakthroughs, but the point still stands nonetheless. That is, DeepSeek is a prime example of the commoditization of LLM's taking place right now.

Palantir: Beyond the Hype

Palantir has long positioned itself as a company that transcends the LLM hype cycle by focusing on real-world deployment. Unlike firms relying solely on proprietary models as a competitive moat, Palantir’s strength lies in its ability to process, secure, and integrate AI into enterprise workflows.

Palantir AIP: Any LLMs are simply inputs applied on the ontology (digital copy of the business) generating actionable, enterprise value.

Palantir’s Artificial Intelligence Platform (AIP) enables seamless orchestration of multiple models, such as ChatGPT, Claude, and LLaMA, tailored to specific use cases. This underscores Palantir’s emphasis on application rather than model creation. As highlighted in a recent tweet by Chad Wahlquist from Palantir, "LLMs are a commodity with Palantir AIP. Here's how easy it is to swap them out or use them together".

Why DeepSeek Matter as an Example of Continued LLM Commoditization

As DeepSeek demonstrates, the idea that proprietary models provide a sustainable competitive edge is increasingly outdated. Instead, the future belongs to companies that can operationalize AI effectively. DeepSeek’s emergence has shattered illusions of long-term differentiation among LLMs. Its implications are profound:

1. Drastic Cost Reduction: DeepSeek’s development of a near-frontier LLM with significantly lower capital suggests plummeting costs in AI creation.

   
   

2. Accelerated Catch-Up Cycle in the LLM Race: Other firms are already analyzing DeepSeek’s methods, shortening the time required to replicate and innovate on its breakthroughs.

   
   

3. Application & Workflow Layers Prevail: The battle is no longer only about the best model but about which company can integrate AI seamlessly into real-world operations.

   
   

4. Security & Ontology as Differentiators: Without structured frameworks like Palantir’s, commoditized models cannot deliver the governance and reliability enterprises demand.

   
   

The Market Implications: Winners and Losers

Financial markets are recalibrating to reflect the commoditization of LLMs. Nvidia, Microsoft, and AMD experienced immediate short-term declines as investors reconsider the economic dynamics of AI upon the arrival of DeepSeek.

• Winners: Companies emphasizing secure AI deployment and integration; such as Palantir, Oracle, and Google Cloud, are likely to thrive. These firms act as the connective tissue between AI models and enterprise workflows.

  
  

• Losers: Pure-play LLM providers banking on proprietary models as a durable moat face increasing pressure as pricing and differentiation erode.

DeepSeek’s rapid emergence serves as yet another confirmation that LLMs are on a trajectory toward complete commoditization. As model differentiation continues to erode, the defining factor in AI’s future will be how effectively these models are deployed and integrated into meaningful applications. The AI landscape is shifting away from exclusivity in model creation and toward innovation in usability, security, and governance. The companies that can efficiently harness AI to solve real-world challenges will define the next phase of technological transformation.

---

By J