Nvidia projects $1T AI capex 2027

Nvidia’s latest earnings call did two things at once: it confirmed that demand for AI infrastructure remains unusually strong, and it pushed the spending debate further out into 2027 and beyond. On May 20, the company reported first-quarter fiscal 2027 revenue of $81.6 billion, up 85% from a year earlier, with data center revenue of $75.2 billion. (investor.nvidia.com) Jensen Huang, Nvidia’s chief executive, said the “buildout of AI factories” was accelerating, while finance chief Colette Kress told investors AI had moved from optional experimentation to required infrastructure. (cnbc.com) The $1 trillion figure attached to 2027 is the part of the call that has resonated most on Wall Street. Kress said analysts were now forecasting hyperscale capital expenditures to exceed $1 trillion in 2027. (investor.nvidia.com) CNBC, citing the earnings call, reported that Huang went further, saying hyperscaler capex was already at about $1 trillion and was growing toward $3 trillion to $4 trillion, while Kress said AI infrastructure spending was on track to reach that annual range by the end of the decade. ### Where did the $1 trillion number come from? Colette Kress framed the figure as an external forecast rather than a formal Nvidia company target. On the May 20 call, she said “analysts now forecast” hyperscale capex will exceed $1 trillion in 2027. That matters because the number refers specifically to the largest cloud and internet companies’ capital spending, not total global technology investment. (investor.nvidia.com) CNBC reported that Huang separated hyperscaler capex from other parts of the market, including neoclouds and other supercomputing segments. Read narrowly, the claim is that the biggest cloud operators alone could cross the trillion-dollar mark in annual capital expenditure before the rest of the AI buildout is counted. (cnbc.com) ### What did Nvidia say about AI demand itself? Kress told investors that “AI is no longer a nice-to-have” and called it “a necessity for enhancing productivity across all industries and roles,” according to reports citing the call. That language is important because Nvidia is no longer describing demand as limited to model training by a handful of labs; it is describing AI as a broader corporate infrastructure purchase. (cnbc.com) Nvidia’s own numbers backed that argument. The company said first-quarter data center revenue rose 92% from a year earlier to $75.2 billion, while networking revenue reached $14.8 billion and compute revenue hit $60.4 billion under its previous reporting structure. Nvidia also said it now has more than 80 partner data centers above 10 megawatts, nearly double a year earlier. (cnbc.com) ### Why are investors extending the story to TSMC, Micron and AMD? The spending thesis does not stop at Nvidia’s own chips. If hyperscalers and enterprises keep expanding AI infrastructure, the beneficiaries would also include foundries, memory suppliers, networking vendors and server makers that sit upstream or alongside Nvidia in the supply chain. That is why the $1 trillion capex discussion quickly pulled in companies such as Taiwan Semiconductor Manufacturing Co., Micron and AMD in market commentary after the call. (finance.yahoo.com) Nvidia’s disclosures point in the same direction. The company said total supply commitments, including inventory purchase commitments and prepaids, rose to $145 billion, a sign of how much hardware, packaging, memory and networking capacity must be lined up to support continued growth. (investor.nvidia.com) ### Why did the stock fall even after strong results? Nvidia shares closed at $211.14 on May 29, down 1.45% for that session, according to market data. The post-earnings debate has centered less on whether AI demand exists and more on how much future growth was already reflected in the stock after a multiyear run-up. The company’s market value, recent revenue growth and investor expectations help explain that tension. Nvidia reported quarterly revenue of $81.6 billion and announced an additional $80 billion share repurchase authorization, but a company can beat estimates and still face a higher bar if investors were positioned for even stronger guidance or faster upside beyond what management delivered. (cnbc.com) That is an inference from the market reaction rather than a statement Nvidia made. (fool.com) ### What comes next that investors can actually track? June 4 is Nvidia’s next dated shareholder marker because that is the record date for its increased quarterly cash dividend. The company said the dividend will be paid on June 26, 2026. (finance.yahoo.com) The next operating checkpoint will be Nvidia’s second-quarter fiscal 2027 results, which will test whether the company can keep converting broad AI spending forecasts into reported revenue, data center growth and supply expansion. Nvidia’s first-quarter call also said production shipments of Vera Rubin were set to begin in the third quarter. (fool.com) (investor.nvidia.com)