Nvidia nears $5t valuation after AI chip surge

Nvidia neared a US$5 trillion valuation on October 28 after announcing US$500 billion in bookings for its AI processors and a plan to build seven supercomputers for the US Department of Energy.

The Santa Clara-based chipmaker’s shares closed up nearly 5%, raising its market value to US$4.9 trillion and briefly touching US$4.9 trillion.

Nvidia is a key supplier in the global AI industry and continues to make deals amid ongoing US-China tech trade tensions.

The largest of the new supercomputers will be built with Oracle and will feature 100,000 Nvidia Blackwell AI chips, while the group of supercomputers will support some US nuclear weapons programs.

Nvidia’s stock has surged 50% in 2025, surpassing US$4 trillion in market capitalization for the first time in July.

🔗 Source: Reuters

🧠 Food for thought

Implications, context, and why it matters.

Nvidia’s $500B in ‘bookings’ lacks transparency on terms and timing

Nvidia says it has US$500B in ‘bookings’. It gives no definition, term lengths, cancellation rights, or customer mix which makes it hard to judge firm revenue versus a pipeline.
Oracle’s Remaining Performance Obligations (RPO), a standard metric for contracted but not yet recognized revenue, rose from US$138B to US$455B in one quarter 1. That equals US$317B in new contracted AI infrastructure revenue in three months. Oracle gives RPO detail while Nvidia skips how bookings map to backlog, revenue timing, or prepayments.

US Department of Energy (DOE) supercomputer projects signal procurement surge for infrastructure vendors

US Department of Energy (DOE) will add AI supercomputers at Oak Ridge National Laboratory (a federal research lab), with Lux targeting early 2026 and Discovery later in the decade 2. These builds need power and liquid cooling.
The agency opened the Oak Ridge Reservation (a DOE-managed federal site surrounding the lab) for private AI data center proposals, with submissions due December 1, 2025 3 4. A government-wide Request for Information named 16 DOE sites for AI infrastructure and on-site power, including advanced nuclear and enhanced geothermal. Vendors can pursue cooling and power work. They can also seek facility or grid contracts.
The Lux supercomputer uses a roughly six month delivery cycle with AMD, Hewlett Packard Enterprise (HPE), and Oracle 5. This public private model can speed deployment. Investors and operators should watch similar deals as the DOE scales this template.