AI Backlogs Top $2 Trillion

NVIDIA is on fire, raking in over $96 billion last quarter with its data-center chips, but an industry-wide chip shortage is fueling both record sales and sky-high prices — and that’s enraging gamers. With $5,000 desktop GPUs now the norm and data-center buyers absorbing 70% of global memory, the stakes are clear: NVIDIA is betting everything on its “full-stack AI factory platform,” but can they actually ship enough product to meet demand? Supply constraints, $279 billion in obligations, and hyperscaler reliance have NVIDIA racing to secure capacity, rallying heavyweight partners like Apollo and SoftBank for billion-dollar infrastructure bets.

But here’s the catch: Broadcom is coming for NVIDIA’s crown with custom AI chips that claim to deliver cutting-edge model performance at half the price of a GPU. If Broadcom’s XPUs gain traction, NVIDIA risks losing future wallet share just as pricey memory and component shortages hit their own ecosystem hard. Meanwhile, NVIDIA’s $3.5 billion MediaTek investment aims to keep custom silicon “inside CUDA,” launching new Arm-based PCs this fall — but Windows on Arm still means tricky software compatibility and big price tags.

Featuring insights from Bloomberg, IDC, and The New York Times, this episode unpacks how NVIDIA’s strategy hinges on landing new PC wins before Broadcom’s challengers scale up — all while navigating supply, pricing, and geopolitics.

Powered by Apisod.com