The truth value of any answer is whether the speaker could have said the reverse. When one reads about Eric Schmidt’s just published talk, where he reinforces the value of computer science knowledge, one wonders if the legend could have said more.
The glaring omission in many Valley voices is the critical role of hardware manufacturing, or the layers below chip design companies where the current monopolistic powers reside and are likely to be the birthplace of most of the next five trillion Dollar companies.
A prevailing sentiment is that semi-manufacturing and associated technologies are under total US ownership, a narrative at odds with reality despite the US ability to exert political control. Almost all the cutting-edge manufacturing is outside the US, and getting them onshore is proving challenging. The reasons are in the work practices needed in this precision field where the US is falling farther behind, the way Mr Schmidt feels Google is falling behind due to WFH. As last week’s NYTimes article shows (“What Works in Taiwan Doesn’t Always in Arizona”), making the highly paid Engineers work in teams and with assembly line manufacturing discipline requires more than having them leave home.
In summary, despite TSMC’s $65bn investment and a $6bn grant, the giant is struggling to establish what it has been able to easily elsewhere. What is needed is a unique blend of white-collar expertise and blue-collar discipline, a combination not easy to execute in the US.
Semi-manufacturing innovations are cumulative. Individually, thousands of tiny process advancements occur on the shop floors of such factories. It’s not glamorous research, but their collective power makes it insurmountable for anyone to join midway without walking similar paths through previous-generation processes.
To prevent the Valley from facing a long secular decline, its business leaders must be mindful of the commoditization risks in software because of GenAI and the rising pricing powers deep upstream. Yes, this could again prove cyclical, unlike what this writer is saying, but it may pay to discuss the risks to ensure protection.
As discussed yesterday, the US businesses, and indeed those elsewhere, must reassess strategic partnerships with the upstream, transforming them from vendor relationships to something deeper. Merely relying on the leadership powers of EDA toolmakers or fabless designers is not enough, as these layers do not hold the same monopolistic advantages in most cases, as evidenced by the rising prowess of players in China or those like Mediatek or Tokyo Electron.
The future of tech isn’t about code; it’s about chips. As influencers whose interviews are heard and posts read recognize this, their activities will give rise to realities we cannot foresee. As things stand today, many changes are needed in topics discussed, even when surrounded by the necessary platitudes.
