There is that old fable of searching for a lost coin under a lamppost: the coin is lost elsewhere, but the search is preferred where the light is.
Something similar is brewing in stock markets: Investors gawk at the supposed innovations in the semiconductor subsegments, where they find GenAI revolution expressions.
Let's use another tacky simile: in GenAI, the world has hit upon a cognitive elixir that keeps giving (https://bit.ly/3Iy5HAY). Unfortunately, this liquid is too inscrutable for listed markets, so the understandable plays are found in its picks-and-shovel bottlers and even bottle-cap manufacturers.
To a degree, the market zeitgeist is justified for the clarity brought to investment opportunities. However, it is reaching an extreme where many forget that the GenAI revolution makes many evolutionary developments of the semi-industry look fab. The cyclical industry has tremendous, new demand benefits, but it is risky to be over-swayed given the fast-developing, massive supply response.
The cutting edge in memories, semi-processing, fabricating, packaging, etc., has been improving steadily for decades. If terms like CoWoS, HBM, or photoresist chemicals are used to discuss a "GenAI innovation" play, it will pay to remember the above lamppost example.
For sure, the GenAI boom has made the whole semi-chain more critical than before. It is more important to follow the latest technological developments in GPUs, memories, or even some EDA tools and OSAT technologies now. But, the more upstream one moves in the semi-subsegments for supposed GenAI beneficiaries, the more commoditized the space becomes.
Even the best of the semiconductor industry, while riding the AI hype, risks becoming a victim of its own success, with unprecedented investments flooding in globally. The announcements, including vast sums, are from the existing majors and new players, funds, and nations. Semiconductor sovereignty is emerging as a critical topic for national security in a rising number of nations; in other words, one should expect the rise of subsidized and non-profit entities in all the most hyped segments in the periods ahead.
Those who expect the current technology leaders to retain supremacy and pricing power ignore the histories of at least a dozen seemingly unconquerable industry leaders from the past and overlook the capacity of giants entering the fray. None of the US tech giants are taking their GPU dependency lightly. Many are more than smarting from having to pay 75% profit margins as they see their size being overtaken by somebody supplying to them.
To conclude, when one watches a development like Devin (https://bit.ly/43h3kfq), it would pay to remember that knowing the details of changes in the GenAI field, however rapid, is far more important than following mostly usual progress within Semi's. And, if wafer bonders or specialty chemicals can be called GenAI plays, almost everything else too.
