Nvidia Posts Another Blockbuster Quarter, But Debt is Rising in AI Frenzy

Nvidia’s second fiscal-quarter numbers were the kind that make a stock jump before anyone has finished reading the release. Revenue hit $96.2 billion, more than double the year-earlier level, and the company said the next quarter should reach about $108 billion, plus or minus 2 percent. That forecast was enough to push the shares higher after hours, then again in premarket trading, because the market had been looking for something closer to $104.2 billion.

The core business is still the data center, and it is still the machine doing most of the work. That segment brought in $89.0 billion, operating profit reached $63.7 billion, and gross margin held at 75.0 percent. Nvidia also said it returned about $26.0 billion to shareholders through buybacks and dividends, while roughly $99.0 billion remains under the current repurchase authorization. The next dividend, $0.25 a share, is due in early October.

What matters beyond the headline beat is how broad the company’s AI push has become. Vera Rubin is already in full production and running inside racks at CoreWeave, Google Cloud, Microsoft Azure, Oracle Cloud Infrastructure and Nebius. Nvidia also says the Groq 3 LPX accelerator is in full production after its $20 billion Groq acquisition, and that Vera, its first in-house CPU for AI agents, has early adopters including SpaceXAI. Then there is Amazon: AWS and Nvidia said they are expanding their partnership by two million GPUs across Blackwell Ultra, Rubin and Rubin Ultra over the next two years.

The numbers underneath that growth are less tidy. Nvidia’s own forecast contains no China data center revenue at all. Gross margin is guided down to 74.0 percent, operating expenses are rising 55 percent year over year, and adjusted profit trails GAAP profit this time because of items Nvidia excludes, not the usual way around. Debt has climbed from $8.5 billion to $33.4 billion, on top of multiyear infrastructure commitments worth about $366 billion. This is what an AI boom looks like once the bill comes due.

The company is now less a chip seller than a financing and infrastructure hub with silicon attached. That can work, right up until the power, land and customer concentration start looking like the real bottlenecks. Europe should pay attention: when the biggest AI buyer in the room is also helping design the rack, the financing, and the supply chain, open competition gets a lot smaller.