The AI Boom Has a New Bottleneck. This Time It Is Inside the Chip
Weekly Stock Market Outlook
The first phase of the artificial-intelligence trade was about securing enough compute. The second was about building the networks, data centres and power systems required to use that compute at scale. The next constraint may be smaller, less visible and just as important.
As AI processors become more powerful, they also become harder to manufacture. More transistors must be patterned with fewer defects. Logic dies must be integrated with high-bandwidth memory. Multiple components must be assembled into one package without sacrificing yield, power efficiency or reliability. Once the system is built, enormous volumes of data still have to move through optical connections at increasingly high speeds.
That is the backdrop behind this week’s three companies: KLA Corporation (NASDAQ: KLAC), Amkor Technology (NASDAQ: AMKR) and Fabrinet (NYSE: FN).
KLA helps chipmakers find defects and control manufacturing processes. Amkor packages and tests increasingly complex semiconductors. Fabrinet provides the precision manufacturing and optical packaging behind high-speed communications equipment.
They sit at different points in the semiconductor chain, but the investment question is the same:
If each generation of AI hardware becomes harder to build, where does the value of manufacturing complexity accrue?
On December 12th, we will be bringing together investors and long-term thinkers at Fairmont Mt. Kenya Safari Club, Nanyuki for the Annual Henriot Capital Investor’s Summit. A full day of research, markets, meaningful conversations and shared experiences, built around one idea: long-term wealth conversations for future generations. Request an invitation below.
