A GPU can be in stock and still be out of reach if a shipment needs export approval.
Nvidia announced on Nov. 17 that its board formed a standing committee to oversee national security policy, export compliance and global trade risks.
The announcement changes who oversees those risks inside Nvidia. For IT teams planning an AI rollout, the question is whether their suppliers can deliver the hardware or cloud capacity they’ve promised, and what happens if that changes.
Who oversees Nvidia’s export compliance
Nvidia’s National Security and Export Control Oversight Committee will be chaired by Suzanne Nora Johnson. Fellow directors Stephen C. Neal and John O. Dabiri will also serve on the committee.
The committee will meet at least quarterly, including with management and in executive session. It will also consult outside experts in law, national security and trade compliance.
Nvidia’s National Security Team and Global Trade Compliance Team will report to the committee for board oversight and to interim CEO Colette Kress on day-to-day operations.
The committee will scrutinize the company’s compliance programs, oversee efforts to strengthen the two teams, and keep national security risks firmly on the board’s agenda. Board chairman Harvey Jones said Nvidia wants to use “every tool at our disposal” to safeguard national security while supporting technological innovation.
Nvidia faces scrutiny over chip exports to China
Nvidia announced the committee six days after Jensen Huang stepped down as CEO, president and board member, effective Nov. 11. Nvidia’s announcement gave no indication that his departure prompted the decision.
The board’s move comes amid US–China tensions over AI chip exports. Reuters reported that Singapore-based cloud company Megaspeed is under investigation over the possible diversion of restricted Nvidia chips to China. Megaspeed denies wrongdoing.
What IT teams should check before committing to GPU capacity
If an AI rollout depends on a specific GPU or cloud region, ask procurement and trade compliance to review the terms before you sign. Start with the commitment your project depends on most.
- Buying GPUs? Confirm what will ship, where and when. Ask the vendor to confirm the exact model, delivery destination and lead time in writing. Check whether delivery depends on a pending approval and whether you can cancel or accept a substitute if the date slips. If you’ve designed around Nvidia’s Vera Rubin platform, have engineering assess a substitute before procurement signs off.
- Reserving cloud GPUs? Confirm where workloads will run and who can use them. Ask the provider which region will host the capacity and whether your teams can access it from the locations they work in. If the provider offers capacity in another region, check data residency, latency and migration costs before relying on it.
- Decide who can approve a fallback. Have procurement track supplier notices, engineering validate alternatives and trade compliance review changes in destination or access. Agree on the date that triggers a switch so the team can act before a delay threatens the rollout.
If your launch depends on a single supplier or cloud region, get the fallback in writing before committing the budget.