Nvidia (NASDAQ: NVDA) has made one of the largest supply commitments in corporate history, pledging $279 billion toward memory components to fuel its expanding AI data center empire.
The announcement came during second-quarter earnings, when CFO Colette Kress revealed the figure had more than doubled from $119 billion just three months earlier.
The sharp increase in supply commitments is primarily driven by demand for high bandwidth memory, which has emerged as a critical bottleneck across AI infrastructure build-outs globally.
Nvidia’s data center segment generated $89 billion in revenue during the second fiscal quarter of 2027, representing 117% growth compared to the same period a year earlier.
The company has guided for $108 billion in total revenue for the current quarter, with management projecting approximately 70% total revenue growth for fiscal 2028.
Of the $279 billion total commitment, $92 billion is scheduled for the remainder of fiscal 2027, followed by $87 billion in fiscal 2028 and $88 billion in fiscal 2029.
The scale of Nvidia’s buying power is effectively reserving the near-term memory market, signaling to manufacturers that demand is substantial enough to justify building new fabrication capacity.
Nvidia’s gross margin came in at 75% last quarter and is guided at 74% this quarter, with management acknowledging a further dip into the low 70% range is realistic by the end of fiscal 2027.
The bulk of Nvidia’s memory spending is expected to flow toward SK Hynix (NASDAQ: SKHY) and Micron Technology (NASDAQ: MU), the two companies at the center of high bandwidth memory qualification for Nvidia’s platforms.
Both Micron and SK Hynix currently trade at forward price-to-earnings multiples around 6, which analysts suggest may undervalue the multi-year revenue visibility that Nvidia’s massive purchase order provides.
Scaling the company’s Blackwell GPU systems and Vera CPUs creates compounding memory demand, as every additional rack and processor socket requires more HBM stacks and server DRAM layered on top.
Rather than absorbing supply risk through spot orders that can disappear within a quarter, Nvidia has chosen to pre-commit at scale, locking in components needed to keep Blackwell and Vera shipments on track.
For Micron and SK Hynix, this spending represents a multi-year offtake agreement that funds capacity additions at a level these companies would historically have only anticipated in the most optimistic scenarios.
Wall Street forecasts already anticipate healthy revenue growth for both memory specialists, but the duration and scale of Nvidia’s $279 billion order could drive results well above current consensus expectations.
