TSMC partners with Nvidia on AI-driven chipmaking as 2nm production ramps up

Taiwan Semiconductor Manufacturing Company has entered into a partnership with Nvidia aimed at applying AI and accelerated-computing techniques directly to its chip-manufacturing processes, a move both companies say is intended to improve production efficiency as demand for advanced AI chips keeps climbing. The announcement comes as TSMC's most advanced 2-nanometer manufacturing node moves into volume production, with the company targeting roughly 100,000 wafers per month of 2nm capacity by the end of 2026 — putting it an estimated 12 to 18 months ahead of its closest rivals, Samsung and Intel. Major chip designers including Apple, Nvidia, AMD and Google have all already secured allocations of TSMC's 2nm capacity, and the company expects revenue from that node alone to surpass its combined 3-nanometer and 5-nanometer business by the third quarter of 2026. Samsung, by comparison, is reported to be investing roughly $40 billion in its 2026 capital expenditure and aiming for around 21,000 wafers per month of 2nm output by year's end — a fraction of TSMC's planned scale — while Intel's foundry business continues navigating a broader restructuring and has yet to land major external customers at the leading edge. The widening gap has fueled reports that even Apple, historically one of TSMC's most loyal customers, has been quietly exploring whether Intel or Samsung could take on a larger share of its chip production going forward.