Nvidia executives this week projected that global data-center capital expenditures will swell to between $3 trillion and $4 trillion by 2030, underscoring the company’s outlook for explosive AI infrastructure build-out. The firm has already sold out its current GPU production capacity and plans to ramp wafer starts at its key foundry partners by more than 50 percent over the next 12 months. This tight supply/demand dynamic is expected to drive Nvidia’s data-center segment revenue growth well north of 60 percent year-over-year through fiscal 2027, based on management’s internal forecasts, and supports continued margin expansion as fixed costs are absorbed by higher volumes.

In remarks delivered on Bloomberg Television, CEO Jensen Huang sought to rebut a U.S. congressman’s assertion that Nvidia offered technical support exclusively to China’s DeepSeek venture. Huang emphasized that Nvidia "supports every developer" leveraging its CUDA ecosystem, noting that more than 2 million engineers globally use its software toolkits for AI model training and inference. He reiterated that Nvidia’s policy is to provide identical driver updates, performance patches and developer guidance regardless of geography, a stance he said preserves neutrality and mitigates the risk of antitrust scrutiny.

Eight industry sources revealed that OpenAI, dissatisfied with certain performance characteristics of Nvidia’s latest Hopper-architecture accelerators, has been exploring alternative GPU platforms since late last year. While OpenAI continues to purchase hundreds of petaflops of Nvidia hardware monthly, these discussions reflect growing caution within its engineering ranks. Concurrently, internal pushback at Nvidia over a proposed investment plan ‘‘up to’’ $100 billion in OpenAI has emerged, with some board members questioning the strategic rationale and potential strain on Nvidia’s balance sheet should the funding exceed $50 billion over the next five years.

Analysts downgraded Nvidia ahead of its fourth-quarter earnings report, flagging two key risk factors: potential gross-margin compression due to higher mix of lower-priced China-sold GPUs, and conservative first-quarter fiscal 2027 guidance that may exclude revenue from Chinese cloud customers. Street models currently assume a 74.5 percent non-GAAP gross margin for the quarter, but management has cautioned that tariff impacts and increased memory costs could shave 200 to 300 basis points. Any guidance shortfall, especially if China revenue is disallowed, could trigger a post-earnings pullback despite the company’s strong long-term growth trajectory.