TSMC's CoWoS advanced packaging capacity—not GPU supply—is now the binding constraint on AI infrastructure timelines, with data centre delivery delays potentially stretching into 2027. Today's briefing also covers Intel's 7% stock climb on 18-angstrom foundry momentum, Samsung and SK Hynix racing on LPDDR6 for on-device AI, and a phonon laser breakthrough from the University of Colorado Boulder.
Audio is available on Spreaker — see link below.
The AI infrastructure story has shifted. The constraint isn't GPU supply anymore.
Intel had a strong week on a different front. The stock climbed roughly seven percent to around one hundred twenty-four dollars fifty, driven by analyst upgrades and visible progress on its eighteen-angstrom process node.
While the data centre is constrained by packaging, the mobile side of AI hardware is moving fast on memory. Samsung completed validation of its sixteen-gigabyte LPDDR6 chip running at ten-point-seven gigabits per second on Snapdragon eight Elite silicon.
One development worth flagging from the research side. A team at the University of Colorado Boulder created a surface acoustic wave phonon laser operating at around one gigahertz.
The packaging constraint at TSMC is the thread that runs through most of what matters right now. Until that capacity expands, the AI infrastructure buildout is moving slower than GPU production alone would suggest.
Chapter summary auto-generated from the verified script. Listen to the full episode for the complete content.