J.P. Morgan reportedly expects TSMC’s A14 process—the company’s 1.4nm-class technology—to enter mass production in the first half of 2028 and begin contributing revenue in the second half. That would move the expected production ramp earlier within 2028, but it remains an analyst forecast, not a confirmed change to TSMC’s schedule.
4
7
11
What the forecast changes—and what it doesn’t
Earlier expectations cited in coverage put A14 volume production in the second half of 2028. The J.P. Morgan forecast brings mass production into the first half, while placing initial revenue in the second half.
4
7
That timing is broadly consistent with TSMC’s stated goal of beginning A14 pre-production in 2027 and volume production in 2028. TSMC’s reported guidance has not confirmed the more specific first-half target.
9
14 In other words, the forecast refines the possible timing; it does not establish that TSMC has officially changed its target year.
Why an earlier ramp is considered plausible
AI demand is one reported driver. J.P. Morgan’s outlook links the faster commercialization timeline to demand for advanced chips and tight leading-edge wafer supply. One report describes supply constraints extending through 2027–2028, rather than establishing a precise shortage period limited to 2026–2027.
7
Construction and pilot plans could support a 2028 ramp. Reports say TSMC’s planned Central Taiwan site includes four fab buildings. The first building was reported as being on track for completion before April 2027; a separate report says pilot production there could begin as early as the third quarter of 2027 if construction, equipment installation and qualification proceed smoothly. These are reported milestones, not guarantees of commercial output.
10
22
Pilot timing also depends on which site is being discussed. Reports describe experimental A14 production at Hsinchu’s Fab 20 in early 2027 and trial production at Taichung’s Fab 25 later that year. Those accounts do not necessarily conflict, but they are distinct site-specific plans.
8
What the near-90% test results mean
TSMC has reported that an internal product-like test vehicle reached close to 90% of target device performance, while a 256-megabit SRAM test vehicle achieved close to 90% yield. Those results are encouraging signs of process development, not evidence that complete commercial processors already have a 90% manufacturing yield.
1
9
13
The distinction matters: SRAM test structures are not equivalent to large finished chips, and their yield figures cannot be directly applied to an Apple processor or an AI accelerator.
13
The cost estimate and possible Apple implications
One report estimates an A14 wafer at about $45,000, compared with roughly $30,000 for a 2nm wafer. If those estimates are comparable, that is a $15,000 increase, or about 50%, per wafer. They are estimates—not confirmed prices paid by customers.
4
A higher wafer cost does not, by itself, mean a finished chip or phone would cost 50% more. The per-chip economics also depend on factors such as die size, usable chips per wafer and yield.
Reports have speculated that Apple could use A14 for the A22 Pro in high-end iPhones expected in 2028. That remains a product rumor, not a confirmed Apple commitment.
7
34 Separate reporting has floated Intel as a possible additional foundry for some Apple chips, but that possibility is speculative and does not mean Intel would manufacture TSMC’s A14 process.
28
Bottom line
The available evidence supports a plausible case for TSMC’s A14 process reaching mass production in the first half of 2028, with revenue contributions following in the second half. AI demand, planned pilot activity, factory construction and development results help explain the forecast, but none independently confirms the exact ramp date. TSMC’s reported target remains volume production in 2028; the more precise schedule and possible Apple adoption are still unconfirmed.
7
9
22