Gold prices hit record high above $3,650/oz on rate cut bets, haven appeal
Investing.com -- Nvidia’s supply chain remains under pressure as demand for its AI chips continues to surpass availability, JPMorgan said following an investor meeting with Toshiya Hari, the company’s vice president of investor relations and strategic finance.
“Demand continues to outweigh supply, keeping lead times stretched but stable,” JPMorgan noted, adding that these remain measured “in quarters, not months,” underscoring the strength of AI spending more than two years into the current cycle.
Nvidia reiterated that its upcoming Vera Rubin platform is on track for a second-half 2026 launch despite industry speculation of delays.
The firm highlighted that Nvidia’s FQ2 inventory increase, up 33% quarter-on-quarter to $8.7 billion, was in support of the Blackwell Ultra (BWU) ramp rather than a build-up of older GB200 chips.
“The vast majority of the finished goods… has already shipped this quarter,” analysts wrote.
Networking is said to remain another growth driver, with attach rates in the “high-70s%/80% range.”
JPMorgan said management emphasised that Nvidia does not bundle networking with compute but provides customers with system design flexibility.
On China, Nvidia has received U.S. government approval to export its H20 GPUs, but no purchase orders have been booked.
According to JPMorgan, Nvidia has enough inventory to support $2 billion to $5 billion of H20 revenue in FQ3 if orders come through, which could imply guidance of $56 billion to $59 billion.
Looking ahead, JPMorgan believes rack production will inflect higher in the second half of fiscal 2026, with supply chain partners boosting capacity and an “ASP uplift from a mix shift towards BWU” adding further support.