Now Playing
Trillion-Dollar Selloff: Broadcom's Miss, SpaceX GPUs & HBM4E Race
Audio is available on Spreaker — see link below.
Broadcom's soft AI guidance triggered a $1.2 trillion semiconductor rout — and six stories underneath it reveal what's actually shifting in the AI hardware supply chain. SpaceX locks in 110,000 Nvidia GPUs, SK Hynix pulls HBM4E forward, and Nvidia taps debt markets despite $96B in free cash flow.
Audio is available on Spreaker — see link below.
Broadcom's AI revenue guidance came in soft, and the semiconductor sector lost one point two trillion dollars in a single session. That's not noise.
While Broadcom rattled the market, SpaceX moved from private speculation into public filing territory. The company confirmed a seventy-five billion dollar IPO at one hundred and thirty-five dollars per share, and simultaneously locked in a deal to rent one hundred and ten thousand Nvidia GPUs to Google for nine hundred and twenty million dollars per month through June twenty twenty-nine.
On the memory side, SK hynix pulled forward HBM4E sample shipments to June or July of this year, roughly six months ahead of schedule. Samsung is already shipping.
Nvidia raised at least twenty billion dollars in the corporate bond market this week, its first debt issuance since twenty twenty-one. The thirty-year paper priced at ninety basis points over treasuries, down from two hundred and twenty basis points four years ago.
Qualcomm is in talks to acquire Tenstorrent for eight to ten billion dollars. That price range is ambitious relative to Tenstorrent's current traction, but Qualcomm's handset business is under genuine pressure and this acquisition would represent a hard pivot into data center AI chip design at scale.
The broader capital picture is worth framing clearly. Four hyperscalers, Meta, Amazon, Google, and Microsoft, are expected to spend seven hundred and twenty billion dollars on chips and data centers in twenty twenty-six alone.
Chapter summary auto-generated from the verified script. Listen to the full episode for the complete content.