Nvidia’s supply commitments rose to $390 billion in its latest quarter, with the company saying the increase was primarily for multi-year memory supply. That is a striking signal of how much AI hardware makers are committing to secure the components behind their next generation of computing.
NVIDIA Watch analysis
What happened
Nvidia’s commitments climbed from $279 billion three months earlier and $119 billion in April, according to an account by Daniel Sparks at The Motley Fool. Of the $390 billion, $203 billion falls in Nvidia’s 2028 financial year, which ends in January 2028. The company said the increase was primarily related to multi-year memory procurement.
Micron says it has 26 multi-year customer agreements, up from 16 a quarter earlier, with customers committing $32 billion, mainly as cash deposits. It is also working with Nvidia on a custom HBM4E chip. Nvidia has not disclosed how its supplier spending is divided, and Micron does not name its largest customers. Samsung and SK Hynix also sell high-bandwidth memory to Nvidia.
Why it matters
AI processors need memory as well as computing power. Nvidia’s rising commitments point to a supply chain being booked years ahead, while Micron’s customer agreements show suppliers are also asking buyers to commit to volumes and put money down. That matters to operators planning AI infrastructure, not just to anyone keeping a nervous eye on semiconductor shares.
The headline figure is not a Micron order book: Nvidia does not say how much goes to each supplier. Nor do commitments automatically mean every dollar becomes revenue on the same timetable. Still, the scale and the jump in spending make memory availability a central part of the AI build-out, rather than a footnote to the chips themselves.
Our read
The useful news here is the scale of the commitment, not the Motley Fool author’s stock pick. Nvidia is locking in memory supply well ahead, and Micron is describing deals designed to secure customer volumes. That looks like a serious attempt to manage supply, though it does not tell us who gets what or whether today’s demand will last. When the AI boom meets a component bottleneck, a purchase commitment is less glamorous than a new chip and rather more revealing.
What to watch
- Whether Nvidia discloses more about how its memory commitments are split between suppliers.
- Whether Micron’s customer agreements translate into delivered volumes and recognised sales.
- Whether Samsung and SK Hynix report similar demand or expand capacity to meet it.
Discussion spark: Should AI chipmakers lock in memory supply years ahead, even if that risks tying up capital and capacity before demand is certain?
Sources and evidence
- Source update (20 November 2026, 04:07 UTC)
Independent WittyWires tracker for public updates about NVIDIA. Not affiliated with or endorsed by NVIDIA; this is not an official account.