Nvidia executives told investors on the company’s fiscal second-quarter earnings call that global spending on artificial-intelligence infrastructure could swell to between $3 trillion and $4 trillion by the end of the decade, underscoring what CEO Jensen Huang called “the beginning of an industrial revolution.” CFO Colette Kress added that cloud providers and enterprises are likely to allocate about $600 billion to data-center infrastructure and compute equipment in 2025, almost double the outlays seen two years earlier. The company expects that surge in capital expenditure to sustain demand for its graphics-processing units and networking gear. Revenue from Nvidia’s new Blackwell architecture rose 17% from the prior quarter, with roughly half of the sales going to major cloud service providers. Nvidia also released a $180 million reserve after selling about $650 million of H20 chips to a customer outside China, saying no H20 sales were booked to China-based clients during the quarter. For the three months ended in July, Nvidia posted a GAAP gross margin of 72.4% and a non-GAAP margin of 72.7%. It forecast third-quarter margins of about 73.3% on a GAAP basis and 73.5% on a non-GAAP basis, each plus or minus 50 basis points, and said it still expects to finish the fiscal year with non-GAAP margins in the mid-70% range.