JPMorgan raises TSMC's target price to NT$3200, mainly driven by improved visibility on AI demand and accelerated expansion plans.
AI summary card
JPMorgan raises TSMC's target price to NT$3200, mainly driven by improved visibility on AI demand and accelerated expansion plans.
The report maintains an Overweight rating on TSMC, believing data center AI demand can be sustained through 2029-30, while expansion in N2, N3, and CoWoS will support upward revisions to revenue and earnings.
- JPMorgan raises its FY26/27/28 EPS forecasts by 4%/2%/6%, respectively, to reflect stronger AI demand and faster capacity growth.
- 2026 capex guidance is raised to US$60-64bn, while JPMorgan further forecasts FY26/27/28 capex at US$62bn/US$81bn/US$90bn.
- Management indicated AI demand visibility extends to 2029-30, and JPMorgan expects data center AI revenue CAGR in 2024-29E to reach 69%.
- The report expects TSMC to maintain 95%+ share in the first two waves of N2 demand, with A14 expected to contribute revenue in 2029 and preserve its lead over Intel and Samsung Foundry.
- Short-term gross margin is affected by N2 ramp-up and dilution from overseas fabs, but the report forecasts 2026/27/28 gross margin at 66.9%/67.5%/68.7%, with a long-term upward trend.
Report interpretation
Overview
This is a JPMorgan company research report following TSMC's 2Q26 earnings. The report's core conclusion is that improved visibility on TSMC's AI data center demand, advanced process expansion, and competitive barriers is sufficient to offset short-term concerns about gross margin dilution, and therefore it maintains an Overweight rating and raises the target price.
Core views
The report believes TSMC's AI demand growth is more sustainable, with deployment timelines and product roadmaps from customers and end customers extending demand visibility to 2029-30. TSMC is expanding N2, N3, and advanced packaging capacity through higher capex to narrow the supply-demand gap in leading-edge nodes and limit room for new competitors. On competition, the report emphasizes that TSMC still has a hard-to-replicate moat in technology, capacity, IP, customer trust, and long-term collaboration. On gross margin, short-term N2 ramp-up and overseas fab expansion will create dilution, but price increases, cross-fab synergies, node transitions, and continued high utilization will support a gradual upward trend in medium- to long-term gross margin.
Analysis framework
The report uses an earnings review, forecast revisions, capex and capacity path analysis, process competition comparison, and a relative valuation framework. The target price is based on about 20x 12-month forward P/E, combined with judgments on AI demand, advanced process capacity expansion, gross margin trends, and competitive risks.
Methodology notes
Derives the Jun-27 target price of NT$3200 based on about 20x 12-month forward earnings.
The report states that the target multiple is above TSMC's five-year historical average, mainly reflecting AI-driven revenue growth and faster capacity expansion.
Raises FY26/27/28 EPS forecasts to reflect stronger AI demand.
JPMorgan raises its FY26/27/28 EPS forecasts by 4%/2%/6%, respectively, and expects adjusted FY26/27/28 EPS of NT$108.15/NT$141.37/NT$180.01.
Assesses the ability to fulfill AI demand through expansion in N2, N3, and CoWoS.
The report expects leading-edge process capacity to grow at a 24% CAGR in 2026-28, with N2 capacity reaching 60k/118k/177k wfpm by the end of 2026/27/28, respectively, and N3 capacity reaching 178k/223k/243k wfpm, respectively.
Compares TSMC with Intel, Samsung Foundry, and competing solutions such as EMIB-T.
The report believes TSMC still leads in leading-edge process nodes, IP, capacity, customer trust, and the maturity of 3D packaging such as SoIC, while EMIB-T is more of a packaging competition rather than a substantive breakthrough in front-end wafer foundry.
Asset mapping & comparison
Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).
- 2330.TWCore covered name, directly driven by AI data center demand, advanced process expansion, and the target price increase.
- Strengths
- AI demand visibility extends to 2029-30; N2, N3, and CoWoS expansion is clear; leading-edge process share and customer relationships are solid; gross margin has long-term upside potential.
- Weaknesses
- Short-term N2 ramp-up and overseas fab construction will dilute gross margin; capex intensity is rising significantly; investors remain concerned about the sustainability of the AI investment cycle.
- Comparison
- The report believes TSMC still leads Intel and Samsung Foundry in leading-edge process nodes, A14 progress, and the maturity of the SoIC platform, while EMIB-T is not yet seen as a major threat to front-end foundry business.
- Risks
- A slowdown in AI capex, weaker-than-expected execution of N2/N3 expansion, greater-than-expected gross margin dilution, customer concentration, intensifying competition, and uncertain returns on overseas investment.
- IntelPotential competitor in leading-edge process nodes and advanced packaging.
- Strengths
- Packaging solutions such as EMIB-T may gain adoption in some projects.
- Weaknesses
- The report believes its Foveros Direct is still in the early ramp-up stage, and penetrating TSMC's core customers in front-end process nodes remains difficult.
- Comparison
- TSMC is expected to maintain 95%+ share in the first two waves of N2 demand, and A14 progress is also viewed as ahead of Intel.
- Risks
- If Intel's packaging and process progress exceeds expectations, it could weaken TSMC's competitive narrative.
- Samsung FoundryCompetitor in leading-edge foundry process nodes.
- Strengths
- Has a presence in advanced foundry process nodes.
- Weaknesses
- The report believes TSMC has stronger advantages in technology, capacity, IP, customer trust, and product roadmap coordination.
- Comparison
- TSMC's A14 is expected to contribute revenue in 2029, and the report believes it can maintain its lead over Samsung Foundry.
- Risks
- If Samsung Foundry makes breakthroughs in yield, customer adoption, or pricing strategy, competition could intensify.
- OSATs(ASE、Amkor)Supplementary providers of advanced packaging capacity, related to TSMC's partial outsourcing trend.
- Strengths
- Can take on packaging demand such as CoWoS-R and CoWoS-L with smaller reticle sizes, easing advanced packaging tightness.
- Weaknesses
- May still be less mature than TSMC SoIC in the migration toward high-end 3D packaging.
- Comparison
- The report believes OSAT solutions are more of an enabling complement to TSMC's leading-edge wafer sales rather than a substitute for TSMC's core position.
- Risks
- If OSAT solutions mature rapidly, the profit pool allocation in advanced packaging could change.
Key data
- RatingOverweightThe report maintains a positive rating.
- Target priceNT$3200.0Jun-27 target price, previously NT$3100.0.
- Current priceNT$2470.0As of 16 Jul 26.
- Implied upsideApproximately 29.6%Estimated by comparing the NT$3200 target price with the NT$2470 current price.
- FY26/27/28 EPS revision4%/2%/6%Reflects faster revenue growth, stronger AI demand, and capacity expansion.
- FY26/27/28 capex forecastUS$62bn/US$81bn/US$90bnThe report raises capex forecasts, with focus on N2, N3, and advanced packaging.
- TSMC 2026 capex guidanceUS$60-64bnRaised from the previous US$52-56bn.
- Data center AI revenue CAGR forecastJPMe 69%(2024-29E)The report expects TSMC may raise its data center AI CAGR guidance over the next few quarters.
- Leading-edge process capacity growth24% CAGR in 2026-28Covers N5 and below.
- N2 year-end capacity forecast2026/27/28 at 60k/118k/177k wfpmDriven by the launch of 9 fab phases over the next three years.
- N3 year-end capacity forecast2026/27/28 at 178k/223k/243k wfpmIncludes new N3 fabs in Taiwan, Japan, and the United States, as well as N5-to-N3 conversions.
- CoWoS capacity2027 year-over-year growth of 60%+, annual capacity of about 2M wafersSupported by AI accelerators and CPU demand driven by agentic AI.
- 2Q26 revenueUS$40.2bnUp 12% quarter-over-quarter in US dollar terms.
- 2Q26 gross margin67.7%Above the previous guidance upper end of 67.5%.
- 3Q26 revenue guidanceUS$44.6-45.8bnMidpoint implies approximately 37% year-over-year growth and about 12% quarter-over-quarter growth.
- 3Q26 gross margin guidance65-67%Midpoint down about 170bps quarter-over-quarter, mainly due to N2 ramp-up.
- 2026/27/28 gross margin forecast66.9%/67.5%/68.7%Affected by dilution in the short term, but with an upward trend in the medium to long term.
Impact & implications
For investors, the report positions TSMC as a core beneficiary of the AI compute capex cycle. If major hyperscalers further confirm 2027 capex growth during the upcoming 2Q earnings season, this would be a stock price catalyst. Higher capex suppresses free cash flow and gross margin flexibility in the short term, but it also strengthens TSMC's ability to capture AI demand and defend against competition.
Risks
- The duration of the AI investment cycle may fall short of expectations, leading to a slowdown in data center AI demand or hyperscaler capex.
- Execution of N2, N3, and advanced packaging expansion may fall short of expectations, preventing the planned narrowing of the supply-demand gap.
- Gross margin dilution caused by N2 ramp-up and overseas fab expansion may exceed expectations.
- Competing solutions from Intel, Samsung Foundry, or EMIB-T may progress beyond expectations, weakening TSMC's leading-edge process or packaging advantages.
- Customer concentration risk may rise; if orders from major AI or HPC customers change, revenue volatility could increase.
- Demand for mature process nodes may be affected by weak consumer electronics demand, macro uncertainty, and rising component prices.
- High capex brings pressure on free cash flow, depreciation, and investment returns.
- Exchange rate fluctuations may affect gross margin and NT$ reported results.
What to watch
- Guidance from major US hyperscalers during the 2Q earnings season regarding AI capex growth in 2027.
- Whether TSMC raises its data center AI revenue CAGR guidance over the next few quarters.
- N2 ramp-up progress, yield maturity, and the actual degree of gross margin dilution in 2H26.
- Execution of new N3 fabs and the conversion from N5 to N3.
- Expansion of CoWoS and SoIC capacity, the proportion outsourced to OSATs, and the adoption scope of EMIB-T.
- A14 tape-out, pilot production, and the 2028 mass production schedule.
- Whether advanced process price increases can be implemented, especially the expected 8-10% price increase for N2/N3 in 2027.
- Whether 2026/27/28 capex remains on the US$62bn/US$81bn/US$90bn path.