The latest outlook from both companies suggests that hyperscalers such as Microsoft, Amazon, and Meta Platforms are still heavily investing in data centre expansion, with spending projected to exceed $600 billion this year as competition to dominate AI capabilities intensifies. This sustained investment is driving strong demand for advanced chips used in training and running large language models.
At the centre of this momentum are leading chip designers including Nvidia, Advanced Micro Devices, and Broadcom, all of which depend heavily on TSMC for the production of cutting-edge processors. The Taiwanese foundry’s dominant position in advanced chip manufacturing continues to make it a critical bottleneck—and a key beneficiary—of the global AI build-out.
TSMC chief executive C.C. Wei reinforced the strength of underlying demand, noting that cloud service providers remain firmly committed to expanding their AI infrastructure footprint. Speaking on an analyst call, he emphasised that customer signals remain robust, prompting the company to raise its annual revenue forecast and increase capital expenditure to expand production capacity.
ASML, which supplies the highly specialised lithography equipment required to manufacture advanced semiconductors, also upgraded its revenue outlook. The company’s leadership described demand conditions as persistently strong, highlighting that order flows remain elevated across AI-related applications.
Industry observers argue that these upgrades reflect a broader structural shift rather than a short-term cycle. According to investment analysts, ASML’s performance in particular signals continued health in the semiconductor ecosystem, even amid growing speculation about a potential “AI bubble”.
However, the nature of demand is evolving. While initial AI enthusiasm was driven largely by training large models, spending is increasingly shifting toward inference workloads—where models are deployed to generate real-time responses. This transition is further intensifying the need for high-performance, energy-efficient chips.
Despite the strong outlook, capacity constraints remain a significant challenge. Both ASML and TSMC acknowledged that supply limitations across the semiconductor value chain are preventing faster scaling. The industry’s reliance on a small group of advanced manufacturers means chip availability is tightly linked to long-term production planning and capital investment cycles.
ASML’s chief executive Christophe Fouquet warned that supply is likely to remain under pressure for the foreseeable future, with constraints affecting not only artificial intelligence hardware but also broader electronics segments such as smartphones and personal computers.
TSMC echoed similar concerns, noting that production capacity remains tight even as it accelerates expansion efforts. The company is increasing capital expenditure in a bid to meet surging customer demand and reduce bottlenecks in advanced chip output.
Overall, the latest updates from both firms underscore a semiconductor industry operating at near-maximum capacity, driven by an unprecedented wave of AI-related investment. While demand remains strong, the pace of future growth may ultimately depend on how quickly global supply chains can expand to keep up.