Intel's foundry ambitions get their biggest validation yet as Google and Nvidia run real design trials on 18A — but yield risk remains the unresolved test. Plus: TSMC crosses $1.4T market cap, Samsung discloses HBM4E specs, and Jensen Huang declines a Senate grilling on China export controls.
Audio is available on Spreaker — see link below.
Intel's stock jumped twelve percent this week after reports emerged that Google has ordered more than three million TPU chips by twenty twenty-eight using Intel packaging, and that Nvidia is running early trials of Intel's eighteen A process for multi-chip designs. That's not a rumor.
The yield question is unresolved, and that's the real test. Google TPU orders and Nvidia trials confirm interest.
While Intel works to close that gap, TSMC just posted a market cap of one point four two seven trillion dollars, up over one hundred percent year on year, ranking it ninth among the world's most valuable companies. That growth reflects the AI chip demand surge directly.
The memory layer is moving just as fast. Samsung has disclosed that its HBM four E architecture hits sixteen gigabits per second per-pin speed and four terabytes per second of bandwidth.
Away from the foundry competition, AMD committed two billion pounds to UK AI infrastructure over five years, backing Cambridge's Zenith and Sunrise AI systems and Oriole's photonic networking lab. The signal here fits a broader pattern.
The near-term watchpoints are specific. Watch Intel's eighteen A yield data as Google and Nvidia trials mature.
Chapter summary auto-generated from the verified script. Listen to the full episode for the complete content.