Taiwan Semiconductor Manufacturing Company Limited
TSM
BUY
Sep 11, 2026Neutral
Decision brief
TSM, the dominant leading-edge foundry in Technology/Semiconductors, closed at $433.24 on September 11, 2026, up 1.22%, recovering roughly half of the two-day pullback from the September 8 breakout high of $444.29 and moving back toward the $440 Street-low analyst target. The macro backdrop improved materially in one session: fell from 17.84 to 15.84 (-3.8% on the close-to-close print and down from the prior day's 8.4% spike), the broad tape rallied hard (S&P +0.86%, Nasdaq +0.96%, Dow +0.98%), and the 10-year yield held at 4.97% — a neutral regime tilting back toward risk-on, which directly relieves the multiple-compression pressure that hit this 31.93x TTM-multiple name on September 10. The technical repair is visible in the data: the histogram re-expanded to +1.75 from +1.74 after the breakout thrust, (14) recovered to 56.89 from 54.10, and snapped back above 100M to 106.93M after dipping to 96.46M — the retest of the $419-422 / confluence never even came close, with price finding buyers at $424-428 instead. Fundamentals remain exceptional and the news flow strengthened: TSMC posted record August revenue of NT$514.81B (~$16.3B), up 53.3% YoY, and the freshest coverage (September 11) reveals a new dynamic — TSMC may raise prices by as much as 10% next year, a move that weighs on customers like Nvidia and AMD but is a direct confirmation of foundry pricing power at the bottleneck of the AI supply chain. Fundamentals are outstanding: forward of 19.76x against 36% revenue growth and 77.4% growth ( 0.81), FCF yield of 32.52%, net cash covering 109.01% of market cap, 6-for-6 beats, and FY26 estimates up $1.31 over 90 days with 9 upward and zero downward full-year revisions. The principal counterweights: cash-flow fair-value models peg intrinsic value near $330 (up to 30% below price), the 10-year at 4.97% keeps pressure on growth multiples, scheduled N2/overseas-fab dilution guarantees Q3/Q4 gross margins 3-4 points below Q2's record 67.7%, and the October 15 earnings print (33 days away) carries documented 7-10% reaction risk against an inverted expectations function. We restore the 1-month stance to BUY from HOLD — the retest held, momentum re-expanded, retreated, and the tape recovered — while maintaining BUY for 6 months and STRONG BUY for 1 year on the durability of the process moat, the one-way revision cycle, and the High-NA roadmap. The overall BUY score rises to 82 from 80 on the successful defense of the breakout zone and the improving regime.
Decision
Live price
$433.24
Recommendation score
82%
82%
Confidence
86%
86%
Outlook by horizon
1M
BUY
Near
6M
BUY
Mid
1Y
STRONG BUY
Long
Primary catalyst
The September 11 article pairing record August sales (+53.3% YoY) with the report that TSMC may raise prices up to 10% next year is the highest-impact item: it confirms both the demand boom and the pricing-power mechanism, but it simultaneously triggered margin-anxiety across Nvidia and AMD — making TSMC's gain the AI complex's cost problem. This crystallizes the setup's central tension: operating momentum is flawless and accelerating in TSMC's favor, yet the market is discounting flawless execution at a rising discount rate, 33 days before an October 15 print with documented 7-10% downside reaction risk.
The regime improved back to the constructive side of neutral: fell from 17.84 to 15.84, the tape rallied broadly on September 11, and TSM recovered to $433.24 on above-average volume with the histogram re-expanding and reclaiming 100M — the breakout-zone retest held. Existing holders should maintain full core positions. New entries are appropriate at two triggers: a confirmed daily close above $436.04 (second breakout attempt) or a shallow pullback into $424-428; avoid initiating fresh positions above $440-444 ahead of the October 15 print without fresh fundamental confirmation. A daily close below $419 would be an early warning to cut exposure by a third; a close below $405 forces exit. Do not add leverage into the October 15 event window, and respect September seasonality despite the improved tape.
Thesis Snapshot
Why It Works
TSM trades at 19.76x forward earnings with 77.4% growth, a of 0.81, an FCF yield of 32.52%, and net cash equal to 109.01% of market cap — a valuation that assumes deceleration the data refuses to deliver, and the past week added both the hardest operating confirmation yet (record August revenue of NT$514.81B, +53.3% YoY) and a new pricing-power datapoint: reported price increases of up to 10% next year, converting the AI supply-chain bottleneck directly into TSMC margin expansion. Every available quarter has beaten consensus (6/6, with the last two surprising +11.2% and +10.9%), FY26 estimates rose $1.31 in 90 days with zero downward full-year revisions, and the moat just got structurally deeper via the TSMC-ASML High-NA initiative locking in leadership for 2030-2033. The technical repair is equally concrete: the September 10-11 pullback found buyers at $424-428 without ever testing the $419-422 confluence, snapped back to 106.9M on above-average volume, the histogram re-expanded, and retreated to 15.84 as the broad tape rallied — the setup now offers entry at $424-437 with a 3.9:1 risk/reward to the consensus-target exit, for a business whose order book, capacity commitments (equipment at ~1.9x year-end levels, $60-64B capex), and revenue prints keep exceeding every estimate.
What Breaks It
The market has twice punished flawless execution: a -7.3% collapse after a 77% profit jump in Q2, and a sell-the-news fade on record August revenue. When record results are the baseline, the only trade left is disappointment — and the new 10% price-increase story is a double-edged sword that the September 11 coverage itself shows 'weighed on semiconductor shares' by raising the specter of Nvidia and AMD margin compression, meaning TSMC's best news now actively threatens its largest customers' economics and could trigger a complex-wide de-rating. Three separate cash-flow models put fair value at $313-$330, roughly 30% below the $433.24 close, and the 10-year at 4.97% keeps the discount rate moving against a 31.93x TTM multiple 16.4% above its 200-day . The chart shows a specific unresolved failure: the September 8 thrust to $444.29 pierced the $440 Street-low target and was rejected within hours, and the September 11 recovery stalled at $435.37 — just under the $436.04 resistance — leaving the two-session upper wick of overhead supply intact until a daily close clears $436. The margin math remains scheduled, not hypothetical: N2 and overseas fabs guarantee Q3/Q4 gross margins 3-4 points below Q2's record 67.7%, the Fed and September seasonality linger, short interest at 0.62% of float means no squeeze cushion in a shock, and 33 days remain until a print the market has shown it will sell on good news. Buying here means paying ~30% above modeled intrinsic value, 33 days before an event with documented 7-10% downside reaction risk.
A daily close below $405.15 (August 24 low) on above-average volume would confirm the failed breakout has become distribution and demand immediate position reduction; an earlier warning is a close below $419.82 () with making new lows for the move. Fundamentally, an October 15 report showing revenue below the raised full-year trajectory, gross margin below ~63% (worse than the guided dilution), or flat-to-down 2027 capex commentary would signal the demand/price-power thesis is cracking.
Expectations inversion with a new transmission channel: with a documented -7.3% reaction to a 77% profit jump, a sell-the-news fade on record August revenue, and cash-flow fair-value estimates of $314-$330 (up to ~30% below price), an October 15 print that beats but doesn't re-accelerate could trigger a 10-15% de-rating — and the 10% price-increase story now risks spilling margin anxiety from Nvidia/AMD across the whole AI complex.
Rising-rate multiple compression: the 10-year at 4.97% with beta of 1.25, TTM of 31.93x, and price 16.4% above the 200-day means further yield spikes mechanically compress the multiple regardless of TSMC's order book — September seasonality compounds the timing risk into the October 15 event.
Scheduled margin compression: the detailed dilution schedule — 2-3 points from overseas fabs initially (3-4 later) plus 3-4 points from the N2 ramp in H2 2026 — means Q3/Q4 gross margins will print below Q2's record 67.7% by design, tightening the error margin if utilization or pricing power slips even slightly.
Execution friction and second-order AI risk: the 1.9x equipment expansion carries named risks of supplier bottlenecks, long installation cycles, and customer-site delays; capital-efficient open-weight Chinese models threaten US AI profitability (the capex narrative beneath TSMC's demand); short interest at just 0.62% of float means no squeeze protection in a shock — while the geopolitical concentration in Taiwan fabs remains the tail risk beneath all of it.
Watch Next
Next earnings: Oct 15, 2026
The September 11 article pairing record August sales (+53.3% YoY) with the report that TSMC may raise prices up to 10% next year is the highest-impact item: it confirms both the demand boom and the pricing-power mechanism, but it simultaneously triggered margin-anxiety across Nvidia and AMD — making TSMC's gain the AI complex's cost problem. This crystallizes the setup's central tension: operating momentum is flawless and accelerating in TSMC's favor, yet the market is discounting flawless execution at a rising discount rate, 33 days before an October 15 print with documented 7-10% downside reaction risk.
Signal Dashboard
Decision inputs
6 signals
Fundamental
27.2%
Upside to analyst mean target
TTM of 31.93x is optically rich, but forward of 19.76x against 36% revenue growth and 77.4% growth produces a of 0.81 — the earnings math says the market pays far less than full price for this growth. of 4.85x and an FCF yield of 32.52% are extraordinary for a 64.23% gross-margin business with net cash of $2.45T (109.01% of market cap); price-to-sales of 0.51x reflects the massive revenue base at current FX conversion, and price-to-book of 88.66x reflects the depreciation-inflated capex base and should not be read conventionally. The absolute-valuation counterargument stands: cash-flow models (GuruFocus GF Value ~$329.58) peg fair value roughly 30% below the $433.24 close, a gap that widened as the stock rallied into the September 8 high — the rally itself manufactures discount-rate-sensitive valuation risk with the 10-year at 4.97%. The target-range structure remains fully supportive: the $440 Street low sits just 1.6% above price, the $551.26 mean implies ~27.2% upside, and the $700 high implies ~61.5%, with the $444.29 September 8 high having already tested the low end of the distribution. Synthesis: cheap on earnings and growth-adjusted multiples, expensive on DCF models embedding a geopolitical discount — the gap is a bet on AI-driven earnings durability, which record August revenue (+53.3% YoY), six straight beats, the raised FY26 guide, and now prospective 10% price increases all support.
Technical
Bullish
Trend quality
The retest held without even reaching the line: after the September 8 breakout to $444.29 (piercing the $440 Street-low target) and two sessions of profit-taking to $428.03, the September 11 session rallied +1.22% to $433.24 with an intraday high of $435.37 — buyers defended the $424-428 zone and price never approached the $419-422 / confluence, which now sits at $421.71/$419.82 below a rising market. Price remains above the and , ~16.4% above the rising 200-day ($372.16), with above , so the golden-cross uptrend is fully intact. The structure reads as a constructive pullback-and-recovery inside a breakout: $436.04-$444.29 remains the resistance band (the September 11 high of $435.37 stalled just under it), $421.71/$419.82 is the dynamic support the market did not need to test, and $405.15 (August 24 low) is the structural floor. A daily close below $419 would revive the failed-breakout scenario; below $405 negates the entire August-September base.
Earnings
33d
Accelerating
Q2 2026 of $4.31 beat consensus of $3.89 by 10.9% with revenue of NT$1.27T and a record 67.7% gross margin, yet ADRs fell 7.3% the next day in the largest single-day TAIEX point decline on record — a textbook illustration of the expectations bar, echoed again when the stock slipped on record August revenue.
News
Mixed-Bullish
Headline tone
The September 11 article pairing record August sales (+53.3% YoY) with the report that TSMC may raise prices up to 10% next year is the highest-impact item: it confirms both the demand boom and the pricing-power mechanism, but it simultaneously triggered margin-anxiety across Nvidia and AMD — making TSMC's gain the AI complex's cost problem. This crystallizes the setup's central tension: operating momentum is flawless and accelerating in TSMC's favor, yet the market is discounting flawless execution at a rising discount rate, 33 days before an October 15 print with documented 7-10% downside reaction risk.
Macro
Neutral
4.97% 10Y / 15.8 VIX
fell from 17.84 to 15.84 and equities rallied broadly (S&P +0.86%, Nasdaq +0.96%) on September 11, moving the tape from cautious back toward the top of the neutral band — a constructive shift for high-beta growth names like TSM, though the 10-year at 4.97% still mechanically caps multiple expansion for a stock trading 16.4% above its 200-day at 31.93x TTM earnings.
Stability
High-Beta Growth
Core Growth
TSM is the indispensable leading-edge foundry at the center of the AI buildout, with a 64.23% gross margin, 39.97% , and net cash of $2.45T equal to 109.01% of market cap — monopoly-like economics on a fortress balance sheet. The evidence base keeps compounding: record August revenue of NT$514.81B (+53.3% YoY), a September 7 TSMC-ASML initiative on 12-inch High-NA EUV photomasks locking in node leadership through 2030-2033, six consecutive beats, and FY26 estimates up $1.31 over 90 days with zero downward full-year revisions. However, beta of 1.25, 20-day realized volatility of 26.95%, near 2.6% of price, and a 4.97% 10-year yield mean investors must tolerate 10-15% drawdowns when the expectations bar is tested — as the documented -7.3% post-Q2 reaction and the ~1% fade on record August revenue both proved. This is a core growth sleeve holding, not a defensive anchor, and should be sized accordingly.
Analyst Spread
Mean Target
$551.26
High / Low
$440 / $700
Median Target
$536.00
Consensus range
$440 - $700
— analysts
Mean
$440$551$700
Consensus
The mean target of $551.26 implies ~27.2% upside from $433.24, the median of $536.00 implies ~23.7%, and the $700 high target implies ~61.5%. The wide $440-$700 spread reflects genuine disagreement about how long the AI capex cycle runs, but the structural fact is that the entire target range sits at or above market price — and the September 8 session actually traded through the Street low of $440 (intraday high $444.29) before being rejected, so the most conservative estimate has been tested and is now acting as the next reference level just 1.6% overhead. With 6 Strong Buys, 14 Buys, 1 Hold, and 0 Sells, analyst positioning is unambiguously bullish and still being revised upward; the market is trading between the Street's low end and its mean.
Positioning Pulse
Insider Signal
Moderate Buy
Net Bias
Buy-heavy
Insider Own.
0.0%
Institutional Own.
15.5%
Interpretation
The complete absence of insider sales combined with open-market purchases by a director and an officer during the consolidation signals management confidence, though the dollar amounts are modest relative to the company's scale and should be weighted lightly.
Notable activity
Director Ursula Burns acquired 1,000 shares worth approximately $322,050 in the most recent filing, and officer Bor-Zen Tien made a series of small open-market purchases (20 shares/$7,760, 17/$6,718, 10/$4,050, 15/$5,850) across recent sessions. No insider sales appear in the latest Form-4 data.
Market Context
Industry
Semiconductors
Bullish
Sentiment
53
Tracked names
13
Grouping
industry
Semiconductors industry is currently Bullish with an average recommendation score near 70. capital and conviction are improving across the group, and leadership is concentrated in NVDA, AVGO, MU.
Rotation Drivers
Average recommendation deltas are trending higher across Semiconductors industry, which suggests improving risk appetite inside the group.
12 names are rated bullish versus 0 bearish names, keeping the group tone bullish.
Leadership is concentrated in NVDA, AVGO, MU, so those names are the cleanest read on group-level capital rotation.
TSM is the critical bottleneck in the AI infrastructure buildout, fabricating leading-edge logic for Nvidia, AMD, Broadcom, and custom-silicon programs from the hyperscalers. This week's evidence hardened the theme on fundamentals: record August revenue of NT$514.81B (+53.3% YoY), the High-NA EUV roadmap from 2030, and now reported price increases of up to 10% next year — the toll-road owner converting demand scarcity into pricing power. But the theme is maturing on valuation and margin transmission: the same pricing-power story that is bullish for TSM is being read as a margin threat for Nvidia and AMD, semiconductor ETFs are up sharply YTD, and the 10-year at 4.97% caps multiple expansion. The theme is strengthening on cash flows and fraying at the expectations margin — exactly the environment where the balance-sheet fortress outlasts the more speculative names.
The September 11 combination of record August revenue, the reported 10% TSMC price increase, a retreat to 15.84, and a +1.22% recovery to $433.24 crystallized both the demand confirmation and the margin-transmission anxiety within the AI trade on the same session.
Pricing power is migrating up the supply chain: the September 11 coverage explicitly frames foundry, HBM, and advanced-packaging suppliers as gaining bargaining power over Nvidia and AMD — capital is beginning to reward the bottleneck owners (TSM, ASML, HBM suppliers) rather than only the accelerator designers.
ETF-driven structural demand persists: TSM remains a top holding in the major semiconductor ETFs, so passive flows keep absorbing supply even on red days — while short interest at just 0.62% of float means the marginal seller in a risk-off tape is a long holder, amplifying drawdowns when expectations are tested.
High-NA EUV Lithography Race
Score: 70
Bullish
The September 7 TSMC-ASML initiative to establish a 12-inch High-NA EUV photomask pilot line by 2031 — enabling production systems by 2033 — formalizes the next lithography generation, with TSMC committing to High-NA high-volume manufacturing from 2030 as AI transistor architectures demand more High-NA layers. Intel Foundry is the early mover (first EXE system installed 2024, over one million wafers processed, High-NA used in Panther Lake volume production), while Samsung targets High-NA for advanced DRAM by 2028 — making this a three-way competitive race where lithography access and mask infrastructure determine who owns leading-edge logic economics in the 2030s. For TSM, the initiative is a moat-extension: lower chipmaking costs and eliminated stitching constraints on the nodes that will fabricate the next AI generations, deepening the very monopoly the market is paying up for — and now reinforcing with 10% price increases.
The industrywide 12-inch mask initiative was announced September 7 ahead of the SPIE BACUS conference, with Intel simultaneously disclosing High-NA volume production on Panther Lake — the first concrete timeline (2030-2033) for the technology that will define the next foundry cycle.
Equipment-partner co-investment signals conviction: ASML is building a 350,000 sqm second Brainport complex (first phase 2029) and expanding its Japan workforce ~40% for TSMC's second Kumamoto fab — suppliers rarely commit multi-year capacity without contracted demand visibility from their anchor customer.
Competitive urgency is public and quantified: Intel's disclosure of 1M+ High-NA wafers and Samsung's 2028 DRAM target turn lithography leadership into a measurable race, supporting the premium the market assigns to TSMC's node roadmap (A14 tape-outs ahead of schedule, High-NA from 2030).
Key Stats
23 metrics total
Market Cap
$2.25T
P/E (TTM)
31.9
Forward P/E
19.8
Revenue Growth
36.0%
ROE
40.0%
Beta
1.25
Full Metric Set
17 additional stats
Valuation Stack
Multiples and cash-flow framing.
Price / Sales
0.5
Price / Book
88.7
EV / EBITDA
4.8
Quality & Balance Sheet
Margins, leverage, and liquidity.
Gross Margin
64.2%
Operating Margin
60.3%
Debt / Equity
16.5
Current Ratio
2.46x
Quick Ratio
2.13x
Positioning & Range
Short interest and 52-week location.
Short % Float
0.6%
Short Ratio
2.96x
52W High
$479.00
52W Low
$257.98
From 52W High
-9.6%
From 52W Low
67.9%
Company Classification
Sector and industry context.
Sector
Technology
Industry
Semiconductors
Additional Signals
Free Cash Flow
$730826014720.00
Detailed Analysis
Valuation
TTM of 31.93x is optically rich, but forward of 19.76x against 36% revenue growth and 77.4% growth produces a of 0.81 — the earnings math says the market pays far less than full price for this growth. of 4.85x and an FCF yield of 32.52% are extraordinary for a 64.23% gross-margin business with net cash of $2.45T (109.01% of market cap); price-to-sales of 0.51x reflects the massive revenue base at current FX conversion, and price-to-book of 88.66x reflects the depreciation-inflated capex base and should not be read conventionally. The absolute-valuation counterargument stands: cash-flow models (GuruFocus GF Value ~$329.58) peg fair value roughly 30% below the $433.24 close, a gap that widened as the stock rallied into the September 8 high — the rally itself manufactures discount-rate-sensitive valuation risk with the 10-year at 4.97%. The target-range structure remains fully supportive: the $440 Street low sits just 1.6% above price, the $551.26 mean implies ~27.2% upside, and the $700 high implies ~61.5%, with the $444.29 September 8 high having already tested the low end of the distribution. Synthesis: cheap on earnings and growth-adjusted multiples, expensive on DCF models embedding a geopolitical discount — the gap is a bet on AI-driven earnings durability, which record August revenue (+53.3% YoY), six straight beats, the raised FY26 guide, and now prospective 10% price increases all support.
Balance Sheet
of 16.5%, cash-to-debt of 3.29x, current ratio of 2.46, and quick ratio of 2.13 define a fortress balance sheet. Operating cash flow of $2.63T covers total debt 2.47x, and net cash of $2.45T equals 109.01% of market cap — TSM faces zero refinancing stress and can self-fund the $60-64B 2026 capex program, the $265B Arizona buildout, the second Kumamoto fab, and the German ESMC facility without accessing capital markets, even at a 4.97% 10-year yield.
Growth
Revenue growth is +36% YoY and growth +77.4% YoY, with Q2 2026 delivering NT$1.27T in revenue (~$40.2B) and record 67.7% gross margin, 60.3% operating margin, and 49.9% net margin. The freshest datapoint remains record August revenue of NT$514.81B (~$16.3B), up 53.3% YoY — a single month equal to roughly 40% of the entire prior quarter's sales, hard operating confirmation that AI chip demand is not slowing. The September 11 coverage adds a critical new dimension: TSMC is reportedly positioned to raise prices by as much as 10% next year, and suppliers across the foundry/HBM/packaging chain are gaining bargaining power over Nvidia and AMD — for TSM this converts the demand bottleneck into margin expansion rather than just volume. Structural moat-extension continues via the September 7 TSMC-ASML High-NA EUV initiative (high-volume manufacturing from 2030, 12-inch photomask pilot line by 2031, production systems by 2033), with 2nm already monetizing, A14 tape-outs ahead of schedule, and equipment requirements at ~1.9x year-end levels confirming a steepening demand curve. The known cost: N2 ramp and overseas fabs dilute gross margin 3-4 points in H2 2026, so sequential margin prints will sit below Q2's record by design.
Business Quality
Gross margin of 64.23%, operating margin of 60.34%, margin of 71.33%, of 39.97%, and of 19.0% reflect monopoly-like pricing power in leading-edge foundry — no competitor can manufacture 2nm/A14 at scale, and the prospect of 10% price increases next year demonstrates the market-power side of that monopoly, not just its technical side. The High-NA EUV commitment through 2030-2033 extends leadership a full node-generation beyond the current cycle. The near-term cost of the moat is quantified and scheduled: overseas fab ramps dilute gross margin 2-3 points initially (3-4 later) and the N2 ramp dilutes 3-4 points in H2 2026. At 19.76x forward with a 0.81 , the earnings math does not fully price this quality; DCF models demanding near-flawless cash conversion at elevated discount rates show the expectations bar is demanding.
Capital Allocation
TSM funds leading-edge expansion entirely from internal cash flow: net cash of $2.45T covers 109.01% of market cap, and the 2026 program runs $60-64B with equipment requirements at ~1.9x year-end levels — on top of the $265B Arizona commitment, 13 new leading-edge and packaging fabs in Taiwan, JASM in Japan, the ~$4.06B German ESMC facility, and High-NA EUV tool deployments from 2030. There is no dilution pressure, the payout ratio is a modest 25.74% with a 0.95% dividend yield, and every deployment is directed at sustaining process leadership, packaging capacity, and geographic security rather than financial engineering. The scale embeds near-perfect demand visibility as an assumption — never tested at a downturn — but at a 32.52% FCF yield the optionality is enormous.
Estimate Revisions
FY26 consensus has risen from $15.626 to $16.934 over 90 days ($16.93 vs $16.90 7 days ago) and FY27 from $19.536 to $21.925, with 9 upward and 0 downward revisions in the last 30 days for both fiscal years and 1 upward revision in the last 7 days for every period. The only blemish is a single downward revision on the current quarter over 30 days (4 up, 1 down), with the 0q estimate edging up to $4.46. The recommendation mix improved again to 6 Strong Buy / 14 Buy / 1 Hold / 0 Sell from 5/12/2 three months ago, and the action tape confirms it: Stifel initiated at Buy/$515 on September 2, Bernstein raised from $430 to $554 on August 11, Needham to $530, and DA Davidson to $500. The revision cycle is still climbing into the October 15 print — the single most important support for the thesis.