Business 5 min read By Callum Montgomery
Nvidia Guides for 70% Revenue Growth in Fiscal 2028 as Amazon Agrees to Buy 2 Million GPUs
Nvidia has guided for 70% revenue growth in fiscal 2028, far exceeding Wall Street's 44% forecast, following Amazon's agreement to purchase 2 million GPUs. The guidance raises questions about sustainability of demand for AI chips.
Nvidia has guided investors toward 70% revenue growth for fiscal 2028, a figure that sharply exceeds the 44% that Wall Street analysts had modelled, as the company confirmed that Amazon has agreed to purchase 2 million of its GPUs. The disclosure, made by chief executive Jensen Huang, signals that the artificial intelligence infrastructure buildout remains on an aggressive trajectory despite concerns about a potential slowdown in data centre spending.
The guidance implies that Nvidia expects revenue to reach roughly $380 billion in the fiscal year ending January 2028, up from an estimated $224 billion in the current period. Amazon's commitment to acquire 2 million accelerators, believed to be the next-generation Rubin architecture chips, represents one of the largest single procurement deals in the history of the semiconductor industry and provides a tangible anchor for the company's growth projections.
Wall Street had been cautious heading into the guidance update, with many analysts questioning whether the explosive demand for AI compute seen since 2023 could be sustained as hyperscale cloud providers begin to digest their enormous capital expenditure programmes. The 44% consensus forecast reflected a view that growth would inevitably decelerate as the installed base of AI infrastructure matures and enterprises become more selective about deployment.
Huang's 70% target suggests that the company sees no such moderation on the horizon. Speaking to investors, he framed the outlook around the transition from training large language models to inference workloads, which require substantially more compute as AI systems are deployed at scale in production environments. He also pointed to sovereign AI initiatives, where national governments are building their own computing capacity, as a new demand pool that did not exist in previous cycles.
The Amazon agreement is central to the credibility of the forecast. A purchase of this magnitude locks in a significant portion of the projected revenue and demonstrates that the largest cloud providers remain willing to commit capital to Nvidia's roadmap years in advance. Amazon Web Services has been expanding its own custom silicon efforts, but the decision to procure Nvidia GPUs at this scale suggests that the company still views Nvidia's ecosystem and software stack as essential for serving its enterprise customers.
Questions remain about whether the guidance is achievable. Supply chain constraints have historically limited Nvidia's ability to ship as many units as demand would justify, and the company has faced scrutiny over allocation practices. Manufacturing capacity for advanced packaging and high-bandwidth memory remains a bottleneck across the industry, though Nvidia has worked to diversify its supply base.
There is also the question of customer concentration. A small number of hyperscale operators account for a disproportionate share of Nvidia's revenue, and any pullback in their capital expenditure plans would have an outsized effect on the company's results. Amazon's commitment mitigates this risk in the near term, but it does not eliminate the possibility that other major customers could delay or reduce their orders if the economics of AI deployment fail to materialise as expected.
The guidance also has implications for the broader market. Nvidia's performance has become a bellwether for the AI trade, and its ability to deliver on such an ambitious target will influence sentiment across the technology sector. A miss would likely trigger a repricing not only of Nvidia's shares but also of the suppliers and cloud providers that have hitched their strategies to the AI boom.
For now, Huang's confidence appears underpinned by order books rather than hope. The Amazon deal, combined with similar commitments from other hyperscalers, provides a degree of visibility that is rare in the semiconductor industry. Whether the company can execute on the operational side remains the key variable, but the demand signal is unambiguous.
