Nvidia, Micron post blowout AI earnings amid Treasury yield spike

Nvidia and Micron occupy complementary positions in the AI infrastructure stack, each defending against the market turbulence triggered by the 10-year Treasury yield climbing to 4.65%. Nvidia's earnings underline its dominance in the accelerator layer: the Blackwell platform is ramping at what CEO Jensen Huang called the fastest pace in company history, with Microsoft's Fairwater site alone running hundreds of thousands of Blackwell GPUs. That velocity is reflected in guidance that projects $91 billion in Q2 revenue and a $1 trillion cumulative forecast through 2027 for Blackwell and Rubin.

Micron's story is structurally different but equally material. The company supplies the HBM (high-bandwidth memory) and DRAM that Nvidia's accelerators cannot function without, and it has systematized what has historically been a brutally cyclical business. The 16 Strategic Customer Agreements represent a deliberate shift: rather than riding memory price swings, Micron has locked in roughly $100 billion in minimum revenue with price floors anchored well above past peak margins. HBM4 revenue has already surpassed $1 billion and is doubling the ramp rate of HBM3E, signaling that demand for the highest-spec memory is accelerating faster than legacy products.

The yield environment poses asymmetric risk. Nvidia, trading at 25x forward earnings, has shed 3.75% in the past week; Micron, at just 6x forward earnings and up 241.59% year-to-date, actually gained 2.58% last week. For investors, the divergence reflects Micron's valuation cushion and the visibility its contracted revenue provides, while Nvidia's premium multiple makes it more vulnerable when rates rise and growth multiples compress. Both names hinge on sustained hyperscaler capital expenditure; any sign of HBM pricing pressure or capex pullback would be a critical inflection point.