Nvidia is on the cusp of a new era in AI hardware. The company’s GB300 servers, built on the Blackwell architecture, are set to ship in the second half of 2025. Supply chain sources say the high demand for these servers is changing priorities in the technology sector, even outpacing traditionally top-tier products like Apple’s new iPhone model.
Nvidia to focus on AI servers
Taiwan-based Foxconn took the largest share of the production of GB300 systems. The company has taken on a significant portion of Nvidia’s orders. These servers feature 72 Blackwell AI GPUs in their top-tier configuration and are currently being developed with the claim of being one of the most powerful AI platforms on the market.
The GB300, which shares similarities with Nvidia’s previous-generation GB200 models, does not require a significant transformation for Foxconn in terms of manufacturing infrastructure. The company’s revenue from AI systems is expected to account for more than half of its total server revenue.
Foxconn is not the only company producing for Nvidia. Other major contractors, such as Quanta, Wiwynn, and Inventec, are also involved in the production and shipping of GB300 systems. Quanta began testing GB300 systems with the same customers after shipping GB200 models to customers in the second quarter of the year. The company plans to ship GB300s by September. A similar schedule is expected for Wiwynn and Inventec.
Information from the supply chain suggests that GB300 systems have surpassed even iPhones in shipping priority. This shift clearly demonstrates the scale of demand for AI hardware and Nvidia’s weight in the industry. The interest in high-performance GPUs has shifted the focus not only of the tech giants but also of the entire manufacturing chain.
This surge in demand for AI systems has also prompted Nvidia’s competitors to take action in the market. While Amazon and Alphabet are developing their own custom chips, third-party designers such as Broadcom and Marvell have also started taking new orders.
{{user}} {{datetime}}
{{text}}