“Seismic” is a grossly inadequate adjective for “change” when talking about the end of idea generation as a solely human activity. In the AI-era financial markets, this fundamental change has ushered in a tectonic shift that is upending the long-standing dynamics between hardware and software.
The Internet Age was defined by software supremacy and hardware commoditization. This has flipped in the GenAI era.
When based in Seoul and Taipei around the turn of the century, this analyst was befuddled by hardware commoditization, in particular. Despite the complexities of manufacturing and the limited number of key players, almost all hardware components outside the CPUs exhibited surprisingly low pricing power. The number of players in key components like memory and GPUs kept declining, but it never lifted its manufacturers’ profitability or market perceptions.
Meanwhile, successful software companies, fueled by network effects and platform advantages, enjoyed immense market valuations and seemingly immense pricing power in whatever revenue paths they chose.
Hardware is reclaiming its value, assuming the unprecedented margin power exhibited by a series of players is not a flash in the pan. While many software segments' underperformance is deemed temporary by their fans, few deny the impact of the boost in programming productivity and rapid cloning potential of any program, however complex.
Bits to atoms is likely a trend across segments when one broadens the perspective. The fundamental change is not machines simply being able to program. They participate in a rising number of cognitive domains and at a sharply rising level of contributions. The biotech and drug discovery are experiencing a similar flip. GenAI is democratizing ideation, which has been a bottleneck since time immemorial, with the limited number of humans capable of generating ideas for drug targets and molecules.
GenAI’s ability to generate a multitude of potential drug candidates is shifting the focus downstream. The key now is not just the generation of ideas but the efficient and cost-effective testing and execution of these AI-generated ideas. Companies with robust research and development infrastructure in low-cost jurisdictions will be best positioned to capitalize on this abundance of ideas.
Only time will tell whether machine ideation causes a long-term flip in the value balance of sectors like materials versus financials or production versus idea-driven services. One gets a feeling that GPU and memory stocks are not the last stocks to move to a new valuation range.
