TSMC smashes records, ASML raises forecasts — buy the AI hardware boom

Verdict: Strong AI fundamentals, but the trigger hasn't fired — wait

The thesis is well-supported at the company level: ASML grew revenue 15.6% to $32.7B with net income up 26.9% and a 49% return on equity in FY2025, while TSMC's FY2024 revenue rose 33.9% with net income up 36.0% — and per Reuters (July 16) TSMC's Q2 profit jumped 77% to a record, with ASML raising guidance again per the WSJ (July 15). The completed 12-month backtest on ASML daily bars is a genuine traded run — 5 trades with an 80% win rate and a 59.9% return — but the 96.9% peak-to-trough drawdown means the strategy gave back nearly everything in a bad stretch, and the 24- and 60-month windows never triggered at all, so the sample is thin. Nothing is live today: ASML's 14-day RSI sits at 50.2 versus the 40 entry gate, TSM is at 65.5, and ASML's close of 1,729.52 is below its lower Bollinger Band, failing the reclaim-from-weakness condition. The parameter-sensitivity run exceeded its time budget, so no robust nearby setup was established — you trade the thresholds as written. This argues for patience: alerts on RSI and the 1,717/1,702 support levels, reassessment after each daily close, and no pre-positioning.

Trade now: armed, but the dip-buying trigger hasn't fired

Nothing is live today. The strategy buys ASML, TSM, or SMH after a one-day sector scare, but it requires several conditions on the same bar: a close below the 20-day EMA, a hold above the lower Bollinger Band, a 14-day RSI below 40 that has crossed back above 35, a bullish MACD crossover, a one-day move worse than -1%, and a low that tags the nearest support level before closing back above it. Right now none of the three trade symbols qualify. ASML's RSI is 50.2 — about 10 points above the 40 gate — and TSM at 65.5 and SMH at 59.7 are even further away. ASML did close below its 20-day EMA at 1,732.75, but at 1,729.52 it sits 27.6 points below its lower Bollinger Band, which fails the 'reclaim from weakness' requirement. "Wait" means: no position, alerts set on RSI and the support levels, and reassessment after each daily close. When a trigger does fire, the plan is mechanical: enter long at the close that satisfies all conditions, take profit at an 8% unrealized gain, cut at -7%, or exit after 45 trading days, whichever comes first. Position risk is sized at roughly 2.4% of capital per trade with a cap near 16.7% per position. The completed 12-month backtest on ASML daily bars produced 5 trades with an 80% win rate, a total return of about 0.6%, and a maximum drawdown under 1% — thin sample evidence, but it supports the immediate decision to let the rules fire rather than pre-position. Note that no robust parameter setup was established in the sensitivity run, so the thresholds as written are the thresholds you trade. For context on why patience is cheap here: the fundamentals remain strong, with TSM's latest annual revenue up 33.9% and net income up 36.0%, and ASML revenue up 15.6% with a 52.8% gross margin. The idea's dip-buying edge depends on panic days — those arrive on their own schedule, and the strategy is built to catch them, not chase them.

The fundamentals back the dip-buying thesis

The thesis — that TSMC and ASML are fundamental AI winners whose dips are buyable — has real balance-sheet muscle behind it. On the latest fiscal year, ASML grew revenue 15.6% to $32.7B and net income 26.9% to $9.6B, with free cash flow up 21.8% to $11.1B and a return on equity of 49%. TSMC's FY2024 revenue rose 33.9% and net income 36.0%, with operating cash flow up 47% and free cash flow nearly tripling. These are not defensive stories; they are the two companies the AI hardware buildout runs through, and the cited news reinforces it: per Reuters (July 16), TSMC's Q2 profit jumped 77% to a record, and per the WSJ (July 15), ASML raised guidance again on what it called unrelenting AI demand. The quality metrics are as strong as the growth. ASML's 52.8% gross margin sits in the 86th percentile of its sector and its 34.6% operating margin in the 95th percentile; TSMC's 45.7% operating margin is in the 98th percentile. Both generate free cash flow in the top 0.1% of their peer sets. Crucially for a dip-buying thesis, these are businesses that can absorb a drawdown without the thesis breaking — ASML cut its debt-to-equity 30.6% in the latest year while shrinking share count by 2%, and both pay growing dividends (ASML's trailing distributions are up 26.8% annually, TSMC's 31.9%). The ownership picture is not hostile to the idea. The most recent TSMC filings show net open-market insider buying of roughly $370,000 as of the June 30, 2026 reporting period — insiders putting their own money in alongside the thesis — and 20 holders reporting over 4.2 million shares. ASML's institutional coverage is thinner (9 holders reporting), but shows no signs of coordinated exit. On the strategy itself, the backtested evidence is a completed, traded run: over the trailing 12 months on daily bars, the ASML…

Scores

• Conviction score breakdown: 50
• Thesis support: 75
• Trade readiness: 30
• Risk quality: 30
• Backtest evidence: 35
• Fundamentals trend: 80

Watch items

• ASML — RSI (14)
• TSM — RSI (14)
• SMH — One-day rate of change
• ASML — Nearest support level
• ASML — Second support level
• ASML — Lower Bollinger Band (20)