Nvidia says capacity can meet growth

Jensen Huang used the Computex stage in Taipei to deliver a narrow message: Nvidia says it can get enough chips and systems built to keep growing. The point mattered because the company sits at the center of the AI buildout, where demand for accelerators, servers and networking gear has repeatedly run ahead of supply. Huang said Nvidia has sufficient capacity for “robust” CPU and GPU growth despite continuing industry constraints. Reuters reported the remarks on June 2 and said analysts still see concentration risk in Taiwan even if the immediate bottleneck looks less severe. ### If Nvidia says capacity is there, what exactly is Huang trying to reassure customers about? Nvidia’s problem has not been demand. The question has been whether foundry, packaging and assembly partners can keep up as cloud groups and enterprises order more AI systems. Huang’s comments were aimed at that issue: not whether customers want more chips, but whether Nvidia can physically deliver enough of them. Computex gave Huang a useful venue for that reassurance because the event is held in Taiwan, where much of Nvidia’s manufacturing ecosystem is concentrated. Reuters said Huang framed Taiwan as central to Nvidia’s production footprint and to the broader AI supply chain. ### Why does Taiwan matter so much in this story? Taiwan matters because Nvidia designs chips but depends on partners there to manufacture and package them at scale. Taiwan Semiconductor Manufacturing Co. is the key node in that chain, and advanced packaging capacity has been one of the industry’s choke points during the AI boom. Huang has been explicit about that dependence. Reuters reported that he described Taiwan as the “epicentre” of the AI industry and linked Nvidia’s expansion plans to the island’s manufacturing base. That does two things at once: it reassures customers about supply access, and it underscores how much of the AI stack still runs through Taiwan. ### What was the investment message around Taiwan? Reuters said Huang referred to plans to invest about $150 billion a year in Taiwan. In the context of his Computex remarks, that figure was part of a broader argument that Nvidia and its partners are scaling the infrastructure needed for continued CPU and GPU growth. The investment language also fit a wider theme at this year’s Computex. Taiwanese officials and industry executives have been emphasizing the island’s role as a stable, efficient production base for global technology companies, while Nvidia has been presenting itself as a long-term builder inside that ecosystem. ### Does this mean the supply crunch is over? No one in the reporting said that. Reuters said analysts viewed Huang’s comments as evidence that near-term bottlenecks are easing, but they also said concentration risk remains because so much advanced chip production is tied to Taiwan and TSMC. A Futunn summary of Computex reached a similar conclusion, saying Nvidia had confirmed sufficient TSMC capacity while supply-demand conditions could remain tight through 2027. That does not contradict Huang’s message. It means Nvidia believes it has enough access to keep growing, not that the industry has suddenly become loose or diversified. ### Why are analysts still focused on concentration risk? TSMC is still the dominant producer of the most advanced chips, and Nvidia is one of the clearest examples of that dependence. Even if wafer supply and packaging improve, the structure of the supply chain has not changed much: a small number of companies in a small number of places still do the hardest manufacturing work. Reuters said analysts highlighted that risk after Huang’s speech. Their point was not that Nvidia faces an immediate shortfall, but that the AI supply chain remains geographically concentrated. For customers and investors, that keeps Taiwan’s capacity, resilience and political stability in focus. ### What should readers watch next? Taipei and Santa Clara are the two places to watch. In Taiwan, the next signal is whether TSMC and packaging partners continue expanding capacity fast enough to meet AI demand into 2027. At Nvidia, the next test is whether future product ramps and customer shipments match Huang’s June 2 assurance that capacity can support robust CPU and GPU growth.