Nvidia CEO Jensen Huang told the Goldman Sachs Communacopia + Technology Conference on September 10 that the AI buildout remains in its early stages, even as chip prices, revenue and supply constraints all keep rising. He reaffirmed a $3 trillion to $4 trillion estimate for AI infrastructure spending by 2030 and said Grace, Blackwell and NVLink system shipments grew 27% month over month.
Huang told the conference that AI infrastructure spending could reach $3 trillion to $4 trillion by 2030, a figure he said follows from two shifts in computing: the move from retrieval-based systems to generative computing, and the end of Moore's Law. He argued both changes require more semiconductor capacity, not less.
Grace Blackwell shipments climb 27%
Nvidia's Grace, Blackwell and NVLink systems in 72-rack configurations grew 27% month over month, Huang said, citing internal reports. Cryptobriefing separately reported that Blackwell production rates have reportedly reached hundreds of units per week, with cumulative Blackwell GPU package shipments estimated in the low millions by late 2025.
Pricing has risen alongside volume. Huang said individual GPU systems now sell for as much as $8.5 million, up from a historical consumer GPU price of $399, with per-system pricing climbing from about $18,000 for Hopper to about $25,000 for Blackwell and about $40,000 for Vera Rubin. Nvidia's revenue reached $303 billion over the last twelve months, with gross margins near 75%.
Supply, not demand, is the constraint
Huang said Nvidia expects 70% year-over-year revenue growth, even though unconstrained demand is growing at more than 100%. He named packaging, DRAM, connectors, voltage regulators and wafer availability as upstream bottlenecks, while land, power and data center shell capacity remain the bigger downstream limit. Nvidia is tracking global land, power and shell availability through its network of neocloud partners, OEMs and cloud providers.
He pointed to Australia as an example, where Nvidia announced a 2-gigawatt data center commitment for 2027 representing $80 billion in infrastructure investment. Huang also said about $400 billion in venture capital has flowed into AI-native companies in the last six months.
Growth stock, value stock
Huang described Nvidia's compute systems as increasingly asset-backed, saying older architectures still generate rental revenue years after launch. According to Investing.com: "We are the world's first and only growth value stock." He pointed to future growth in cybersecurity, where Nvidia partners with CrowdStrike, Cisco and Palantir, and in physical AI, where he expects self-driving cars to show substantial progress within two to three years.
Sources: Investing.com, Crypto Briefing
Trading involves risk.
Risk Disclaimer
