Bank of America analysts say TSMC will reach its 2027 target for 2nm capacity this year, ahead of schedule, as orders from AI giants overflow the foundry's booking queues.

The call comes from the bank's semiconductor research team, as reported by BigGo Finance, and it marks a rare case of a major foundry hitting a capacity milestone early rather than late. Demand from large AI customers is the mechanism behind the shift.

Why is 2nm capacity arriving early?

The 2nm class node is the next full-generation step after TSMC's N3 family, and it is where the largest AI accelerator buyers want to be. According to the report, the biggest AI companies are scrambling to lock in wafer capacity, and their orders have overflowed what TSMC had planned to offer at this stage.

When a foundry sees more firm demand than its roadmap assumed, it pulls capital spending forward. That appears to be what Bank of America expects: the 2027 capacity target becomes achievable in the current year.

The takeaway in plain words: customer demand, not TSMC's own schedule, is now setting the pace of the 2nm buildout.

Who is driving the orders?

The report does not name individual customers, but it frames the buyers as "AI giants" — the hyperscalers and AI chip designers that already buy most of TSMC's leading-edge output. Their scramble for 2nm wafers suggests leading-edge supply remains tighter than demand.

For the foundry, overflow orders mean pricing power and fuller fabs. For customers, it means queue position at 2nm has become a competitive issue in itself.

The takeaway: whoever secures early 2nm wafer allocation gains a head start on the next generation of AI silicon.

What remains unconfirmed?

The claim rests on analyst estimates, not on a TSMC statement. Bank of America has not published the underlying capacity figures in the report summary, and TSMC has not confirmed the accelerated timeline.

Readers should therefore treat the 2027-target-this-year claim as a forecast from one bank's research desk. Confirmed numbers — wafer starts per month, capital expenditure, or node revenue — have not accompanied it.

The takeaway: the direction is credible, but the arithmetic behind it is not yet public.

What comes next?

If the forecast holds, 2nm capacity arrives roughly two years early, and the fight for allocation will define which AI chip families reach the market first at that node.