TSMC can't make chips fast enough while CoreWeave drowns in debt — pair long TSMC against short CoreWeave

Verdict: the TSM-strong, CRWV-fragile spread is well-grounded — but no leg is live yet, so wait

The economics behind this pairs idea are genuinely strong: TSM's fiscal 2024 showed revenue up 33.9% to TWD 2.89T with a 45.7% operating margin (the 98.8th percentile of IT peers) and free cash flow up 197.9% to TWD 870.2B, while CoreWeave's Q1 2026 gross margin fell 2.1 points to 65.5%, net margin worsened 6.9 points to -35.6%, and shares outstanding grew 5.9% to 532M — a coherent quality-versus-fragility spread. The completed 9-month backtest returned 10.9% across 25 trades with a 48% win rate and an 11.8% maximum drawdown, but that is a single regime, and the 12-, 24-, and 60-month windows never ran because CRWV daily-bar coverage was incomplete — so there is no evidence across a full chip cycle. Nothing is live today: TSM's RSI (19.8) is a quarter of a point from the first gate but 25 points from the second, and price ($408.07) sits 7.2% below its 50-day average of $437.60. The deepest concern is that the "hedge" is really a bet against TSM's own customer — in a genuine AI capex bust, CRWV distress could lead TSM order cuts, and the 0.47 correlation between the two names could turn positive exactly when diversification matters most. Wait until the entry conditions fire, and size the position so a CRWV-style 66.9% drawdown on the short leg is survivable.

Trade now

**Nothing is live yet — both legs are waiting on TSM.** The setup's entry conditions require TSM's 14-day RSI to be above 20 and above 45, and TSM's close to be above its 50-day moving average. At the latest close, TSM trades at $408.07 versus a 50-day average of $437.60 — the price is below the average, and RSI is 19.8, a quarter of a point under the 20 threshold but a full 25 points away from the 45 requirement. The only condition already met is the CRWV half: CoreWeave closed at $64.01, comfortably below TSM's 50-day average, which is what the strategy needs to confirm the stressed borrower is lagging the healthy chipmaker. TSM's RSI crossing back above 20 is close — that condition is flagged as near — but the recovery above 45 and the reclaim of the $437.60 average are the real gates. **Risk framework is fixed.** When an entry triggers, the strategy takes profit at +5.2% and cuts losses at -2.6%, with hard structural exits at TSM's nearest resistance ($410) and support levels ($406.83 first, $400 second). Position sizing caps any single position at 25% of the portfolio with roughly 2.6% risk per trade, and a 90-day holding-period exit forces any stale position off the books. On the entry math, the +5.2% target against the -2.6% stop implies roughly 2:1 reward-to-risk on each triggered trade — but only once the entry fires. **What "wait" means concretely:** no leg gets opened today. The historical record for this frozen rule set is a completed 9-month backtest on the TSM long leg — 25 trades, a 48% win rate, a 10.9% total return, and an 11.8% maximum drawdown — and exits were filled on daily bars rather than intraday prices, so treat that win rate as coarse. No robust parameter setup was established because the strategy's market data could not be fully evaluated, so the frozen rules trade exactly as written above. The actionable signals to watch are TSM's RSI recovering through 20 (imminent), then 45, and price reclaiming $437.60.

A fundamentals-backed divergence: the foundry's cash machine vs. the borrower's burn

The core economic argument behind this pairs idea holds up in the numbers. TSM's fiscal 2024 results show a business at the top of its peer group: revenue of TWD 2.89T, up 33.9% year over year, a 56.1% gross margin, a 45.7% operating margin (the 98.8th percentile of 899 Information Technology peers), net income up 36% to TWD 1.16T, and free cash flow up 197.9% to TWD 870.2B. It also pays a growing dividend — trailing twelve months of $3.85 per share, with the latest quarterly payment raised to $1.107 (ex-date 2026-09-16). That is the 'healthy demand story' leg, and the recent CNBC report on the 'fierce' battle among Nvidia, Apple, and Broadcom for TSMC capacity, published alongside the record quarter, supports the demand premise directly. The short leg is equally well-supported. CoreWeave's trajectory is deteriorating on every margin measure in the latest quarter (Q1, ended 2026-03-31): gross margin fell 2.1 points to 65.5%, operating margin worsened to -6.9%, net margin worsened 6.9 points to -35.6%, and return on equity dropped to -15.5%. Free cash flow was -$7.25B for fiscal 2025 and a further -$4.71B in Q1 2026, against $3.13B of cash and $46B of total liabilities. Bloomberg's piece on CoreWeave sliding on debt the same day confirms the market is repricing exactly this leverage risk. The two legs are genuinely driven by different risks, which is what makes the hedge coherent. TSM's downside is cyclicality of demand; CRWV's downside is balance-sheet fragility. In a sector-wide rout, both can fall, but the thesis only needs the spread to widen — TSM outperforming a financially stressed borrower — and CoreWeave's negative free cash flow plus share count growth (502.4M to 532.0M shares in one quarter, a ~5.9% increase) point to continued dilution and financing dependence on the short side. The realized evidence is consistent with the thesis. On the evaluated daily-bar window (9 months, 187 bars), the traded pair returned 10.9% across 25 trades with a 48% win rate and a maximum drawdown of 11.8%. That is not a high hit-rate strategy — nearly half the trades lose — but the payoff asymmetry came…

Scores

• Conviction score breakdown: 56
• Thesis support: 80
• Trade readiness: 30
• Risk quality: 50
• Backtest evidence: 45
• Fundamentals trend: 75

Watch items

• TSM — RSI (14)
• TSM — RSI (14)
• TSM — Close vs EMA (50)
• CRWV — Close vs TSM EMA (50)
• CRWV — Net margin
• CRWV — Shares outstanding
• TSM — Close vs support