Taiwan Semiconductor Manufacturing is raising its 2026 capital spending to $52–$56 billion（約9兆円） to support AI chip demand, and Citigroup projects AI data center spending will grow another 40% to 50% next year to around $1.5 trillion（約240兆円）. These moves signal confidence that AI infrastructure growth remains robust, easing fears of a slowdown that have recently pressured AI stocks.
Summaries like this, in your inbox every morning.
Sign up free →What happened
Taiwan Semiconductor Manufacturing raised its 2026 capital expenditures to between $52 billion（約8.3兆円） and $56 billion（約9兆円） to support AI chip growth, and Citigroup analysts expect AI data center spending to grow another 40% to 50% next year, bringing it to around $1.5 trillion（約240兆円）.
Why it matters
Concerns have mounted that AI infrastructure spending could slow, pressuring AI stocks. These signals from a major chip manufacturer and Wall Street analysts suggest the boom has real legs, because TSMC CEO C.C. Wei consulted extensively with customers and their clients before committing to the higher spending, and underutilized chip factories can severely damage a foundry's profits—meaning this is not a decision taken lightly.
What to watch
Citigroup analyst Vivek Arya attributed the projected growth to AI agent adoption, token growth, and current market supply constraints. The key question is whether actual spending follows these projections through 2026.
The recent pressure on AI stocks has stemmed largely from fears that hyperscalers—large cloud providers building out massive data centers—might reduce their infrastructure spending. Meta Platforms' announcement that it could offer cloud computing services with excess capacity fueled these concerns, raising questions about whether demand is truly as strong as believed. However, major players are now providing reassurance. TSMC's decision to commit $52–$56 billion（約9兆円） in 2026 capex is particularly significant because chip-making capacity is an irreversible investment: unused factories destroy profitability, so foundry leaders do not make such bets without confidence in customer demand over a multi-year horizon. Wei's approach of consulting not only his direct customers but also their downstream clients suggests he is building conviction across the entire AI supply chain.
Citigroup's projection of 40% to 50% growth in AI data center spending next year, reaching around $1.5 trillion（約240兆円）, extends this confidence narrative. The bank attributes this growth to three concrete factors: adoption of AI agents (autonomous AI systems that make decisions and take actions), token growth (the increasing amount of data AI models process), and supply constraints that keep demand high. Together, these signals from both equipment suppliers and Wall Street suggest the AI infrastructure boom is not a near-term bubble but rather a multi-year buildout, though investors are being advised to treat AI stocks as long-term holdings with solid fundamentals rather than short-term bets.
No comments yet. Be the first to share your thoughts!
Log in to join the discussion
Get curated AI news from 200+ sources delivered daily to your inbox. Free to use.
Get Started FreeFree · takes 30 seconds · unsubscribe anytime
1 minute a day. The AI essentials.
200+ sources · Email / LINE / Slack
