Large models are becoming "free infrastructure," with real Alpha opportunities at the application layer?
Senior investor Sparks believes that large language models (LLMs) are rapidly being commoditized, much like broadband or tap water, and are being "given away for free," with the models themselves not being the ultimate value creators. The real investment opportunities have shifted from the infrastructure layer to the application layer. The biggest opportunities in the next 2-3 years lie in innovative applications that use AI technology to disrupt traditional industries
A senior investor, speaking through a hedge fund CIO, suggested that the investment focus in the AI field should shift from infrastructure to the application layer.
According to an article published by Eric Peters, Chief Investment Officer of One River Asset Management, on November 24, 2025, entrepreneur and seasoned investor Sparks pointed out that there is a misconception in the current market regarding large language models (LLMs); the true long-term investment value lies not in building these models themselves, but in the application ecosystem built on top of them.
Sparks compared the current state of LLMs to the early stages of broadband network construction. He believes that LLMs are rapidly commoditizing, and these models are "essentially being given away for free," much like the beloved tap water.
"When broadband was built, the real money was not made by the operators, but by the companies that built businesses on top of broadband."
Sparks believes that the current market landscape is characterized by several large language model developers (such as OpenAI) fiercely competing to outdo each other on nearly identical functionalities, which is akin to "simultaneously creating 10 Googles."
In his view, LLM developers are playing the role of operators from back in the day; they provide the basic tools, but the most substantial profits in the value chain will flow to those application developers who best understand how to use these tools to create real commercial value.
This judgment directly influences his investment strategy. Sparks stated that he would rather "be the one making real money using search engines" than be a developer of search engines.
He is focusing all his "forward-looking energy" on finding opportunities for the next 2-3 years (which he refers to as "T+3"). He believes that the biggest opportunities lie with startups that can deeply integrate AI capabilities with specific industries, thereby creating significant efficiency improvements and transforming business models.
Regarding the current market's enthusiasm for AI infrastructure, Sparks takes a cautious stance. He compares NVIDIA's valuation of up to $5 trillion to Cisco's during 1999, suggesting that this is more of a "backward-looking" valuation, reflecting achievements that have already occurred rather than future potential.
He predicts that although the U.S. will invest $500 billion in building data centers over the next few years to meet "crazy" demand, this round of construction resembles a "small boomlet," with the capital and attention invested having already gone "offsides."
