For a revolution to fulfill its potential, it must become less about its upstream creators and more about the downstream applications, a point lost in today's financial markets.
Drawing parallels from the early days of the internet to the current landscape of AI investments, particularly in the semiconductor industry, presents a compelling narrative. The initial excitement around the internet saw investments heavily focused on infrastructure and key players like Cisco, similar to today's emphasis on GPU manufacturers and semiconductor companies in AI. Like during the bubble stages of the '90s, cyclical sector analysts have again reduced a new technology to investment in commoditized upstream picks and shovel plays like wafer bonders, packagers, and IC design-makers.
Those who invest a tad more downstream in LLM makers are also at risk because of the extreme commoditization of the space. There are hundreds of LLMs being constructed worldwide, unlike what happened in searches, iPhones, or social media providers. Most are being constructed in a short period with small teams at tiny budgets and still are able to beat the best LLMs a few months earlier. Anything that can be made with such little effort could be copied and trumped with little effort, regardless of valuations ascribed to them by enthusiasts and the money raised.
Internet-era lessons are more relevant if we move further downstream to analyze the potential of chatboxes and copilots. Early internet applications, like Hotmail, proved extremely low-end when Amazon, Google, and Facebook appeared years later. Like chatboxes today, Netscape offered a new form factor for consuming a new technology to awe the world. ChatGPT does not have to meet the same fate, but its form factor alone will not provide any long-term business protection.
This writer firmly believes in GenAI's ability to change the world across sectors and themes by spurring innovations. Investors that share GenAI optimism should be careful in implementing through picks and shovels aka its cyclical upstreamers who are market darlings, and most of them with no genuine new innovations.
