TSMC is reportedly aiming to scale up production of its 2nm (N2) chip node dramatically, targeting about 100,000 wafers per month by the end of 2026. According to TechPowerUp, the company recently reached roughly 20,000 N2 wafers per month, meaning it would need to increase output fivefold in a matter of months to hit the new goal.
Currently, the N2 node makes up only about 3% of TSMC's revenue, while the older 3nm (N3) node accounts for 30% and the 5nm node leads with 33%, according to TechPowerUp. Achieving the fivefold wafer increase would push N2's revenue share above 10%, even accounting for continued demand in N3 and N5.
Demand for TSMC's 3nm N3 family remains strong. Both TechPowerUp and Wccftech report that customers including NVIDIA, AMD, and Broadcom have driven monthly N3 wafer output to about 180,000 units, surpassing TSMC's internal targets and arriving ahead of the company's original Q4 2026 timeline. Wccftech notes that Economic Daily News reported this figure, up from a previously mentioned target of about 175,000 wafers, prompting TSMC to consider rapid expansion plans for the node.
Despite N3's strength, TSMC says N2 is not underperforming. Both outlets report that the N2 node has seen four times as many customer tape-outs compared to N3 at a similar stage. TechPowerUp attributes this to N2's gate-all-around (GAAFET) transistor technology, which delivers up to a 15% performance increase at the same power level, or up to a 30% power reduction at the same speed, compared to N3. Customers doing N2 tape-outs include AMD, with its EPYC "Venice" server CPUs, and Apple, with its A20 Pro chips for the iPhone 18 Pro series.
Wccftech adds context on demand drivers, noting that AI customers are expected to increasingly shift orders to the 2nm process to gain higher inference performance while using less power and reducing datacenter footprint. The outlet also notes that Apple's A20 Pro chip is expected to launch in September in the iPhone 18 Pro and iPhone 18 Pro Max, with Qualcomm and MediaTek expected to follow with their own 2nm chips for 2027 flagship devices, and that AMD's MI455X GPUs are positioned to compete with NVIDIA's Rubin Ultra. Wccftech also reports that Apple may not use the 2nm process for more than two generations before moving to a 1.4nm node, though it does not specify a reason tied directly to supply figures beyond noting potential capacity constraints as datacenter deployments scale.
Wccftech cautions that its figures, sourced from Economic Daily News, should be treated with some caution, though it says TSMC could still achieve the milestone through an aggressive expansion plan.