November 10, 2026
The next choke point in the AI chip supply chain is shifting from memory chips to the specialized tools needed to produce them.
What happened
The emerging hardware bottleneck for artificial intelligence is moving beyond high-bandwidth memory chips to the manufacturing equipment that makes them, such as hybrid bonding and TSV tools. This signals that constraints on AI chip supply will persist even as capital spending accelerates. The shift moved roughly 500 billion dollars in market value.
Why it matters
A shortage of the machines that build HBM means that pouring money into new factories cannot immediately fix the chip deficit. It extends the timeline of the AI supply crunch and strengthens the pricing power of semiconductor equipment companies. The question is whether massive AI infrastructure investments can be delivered on schedule.
The case against
If demand for HBM proves more cyclical than expected, a rush to build equipment capacity today could lead to a painful glut later. Equipment makers may also face their own supply constraints or execution risks that limit how fast they can ship new tools.
Our read
Our view, grounded in evidence, is that through 2030 demand for frontier training and inference will exceed the supply of accelerators, memory, wafers, packaging, and interconnect. The equipment logjam is a direct consequence of that imbalance.
What settles it
Lead times and delivery schedules for TSV and hybrid bonding tools from major equipment manufacturers.