Semiconductors (SMH) vs 10Y Treasury Yield
SMH (VanEck Semiconductor ETF) tracks 25 of the largest US-listed semiconductor companies. Top holdings: NVDA approximately 20 percent, TSMC ~13 percent, AVGO ~8 percent, AMD ~5 percent.
Also known as: Semiconductors (SMH) (ETF_SMH, semiconductors, semis, chips) · 10Y Treasury Yield (10Y yield, 10 year treasury, TNX)
Why This Comparison Matters
SMH (VanEck Semiconductor ETF) tracks 25 of the largest US-listed semiconductor companies. Top holdings: NVDA approximately 20 percent, TSMC ~13 percent, AVGO ~8 percent, AMD ~5 percent. SMH price approximately $482 (April 2026, near recent highs). The 10-year Treasury yield (FRED DGS10) sits at 4.31 percent. Semiconductors are theoretically longest-duration tech sub-segment due to AI capex growth narratives with long payoff horizons. Rising 10Y yields should pressure SMH through duration framework. The 2024-2026 era has paradoxically broken duration framework: SMH +3x since November 2022 ChatGPT moment despite elevated 10Y range 3.6-5.0 percent. AI capex narrative ($400B+ hyperscaler annual run-rate) overwhelmed duration drag.
The April 2026 Configuration
SMH price approximately $482 (April 2026). 10Y Treasury yield 4.31 percent. NVDA ~20 percent of SMH, TSMC ~13 percent, AVGO ~8 percent, AMD ~5 percent. Combined top 4 ~46 percent of SMH (high concentration).
SMH has rallied substantially in 2024-2026. SMH +3x since November 2022 ChatGPT moment. AI capex narrative drove NVDA +12x peak-to-peak (2023-2025) plus Broadcom, AMD, ARM, ASML, Lam Research, Applied Materials.
NVDA Q3 fiscal 2026 results (released February 2026): revenue +62 percent YoY. Hyperscaler capex on AI infrastructure $400B+ annual run-rate (Microsoft, Meta, Alphabet, Amazon combined).
The combined April 2026 reading: SMH near recent highs despite elevated 10Y. Configuration confirms duration framework reasserts only in absence of major capex narrative. AI capex era (2024-2026) has overwhelmed duration sensitivity.
Forward-looking: NVDA next earnings May 2026 will signal AI capex sustainability. Hyperscaler Q1 2026 earnings (April-May 2026) will signal capex guidance. Sustained $400B+ run-rate supports continued SMH outperformance vs 10Y.
Why Semis Are Most Rate-Sensitive Tech Subsector
Semiconductors are theoretically the longest-duration equity sub-segment within tech. Drivers.
Growth profile: AI capex cycle has 5-15 year payoff horizon. Data center buildout (NVDA, AVGO, AMD) requires multi-year capacity expansion. Foundry capex (TSMC) requires 3-5 year construction timelines.
Multiple expansion: SMH forward P/E approximately 30x (April 2026) vs SPY 22x. Higher multiple reflects longer-duration cash flows.
Capital intensity: semis require substantial capex for fab construction. Higher discount rates compress NPV of future capex returns more than for software companies.
Empirical sensitivity (pre-2024): 100 basis point rise in 10Y yield typically associated with 8-15 percent SMH decline (over 60-90 day windows). Compare to XLK 5-10 percent decline; SPY 3-5 percent decline. Semis 1.5-2x duration sensitivity vs XLK.
2022 hiking cycle illustrated. 10Y rose 1.5 percent to 5.0 percent peak. SMH fell 38 percent peak-to-trough October 2022. Within semis, longest-duration names hit hardest: NVDA -65 percent peak-to-trough; AMD -55 percent. Established memory/legacy names (Intel, Texas Instruments) -30 to -40 percent.
2024-2026 era anomaly: AI capex narrative override. NVDA earnings revisions catalyzed SMH despite elevated 10Y.
The AI Capex Boom
AI capex represents largest tech investment cycle since dot-com era. Hyperscaler capex (top 4 cloud providers).
Microsoft: $80B+ FY26 capex run-rate (vs $50B FY24). AI infrastructure dominant. Meta: $60-65B 2026 capex (vs $40B 2024). Llama models + ad targeting AI. Alphabet: $75-80B 2026 capex (vs $50B 2024). Gemini + Cloud expansion. Amazon: $100B+ 2026 capex (vs $48B 2023). AWS Bedrock + Anthropic investments.
Combined hyperscaler capex $315-325B run-rate (2026), up from $200B (2022). Plus Tesla, Apple, Oracle adding additional $80B+ infrastructure capex. Total US AI capex approaching $400B+ annual run-rate.
Semiconductor beneficiaries. NVDA: H100/H200/Blackwell GPUs dominate AI training. Revenue +62 percent YoY Q3 fiscal 2026. AVGO: networking + custom AI chips for hyperscalers. AMD: MI300 GPUs gaining traction. TSMC: foundry to all advanced AI chips. ASML: EUV lithography monopoly ($200-400M per machine). Applied Materials, Lam Research: fab equipment. Memory: Micron, SK Hynix HBM (high-bandwidth memory) for AI accelerators.
The practical implication: AI capex provides earnings revisions that overwhelm rate sensitivity. SMH rallied through 2024-2026 despite 10Y at 4-5 percent.
How the Pair Performs Through Rate Cycles
Three rate-cycle examples.
2018-2019: 10Y rose 2.4 percent to 3.2 percent (Q4 2018 peak), then fell to 1.5 percent (mid-2019). SMH fell 25 percent peak-to-trough Q4 2018, then rallied 60 percent through 2019. Pattern: classical duration framework worked.
2020-2021 COVID: 10Y fell 1.9 percent to 0.5 percent then rose to 1.7 percent. SMH fell 38 percent peak-to-trough COVID flash crash; recovered + rallied 100 percent through 2021. Strong pattern.
2022 hiking: 10Y rose 1.5 percent to 5.0 percent peak. SMH fell 38 percent peak-to-trough October 2022. Classical duration framework worked. Within semis: NVDA -65%, AMD -55%, Intel -50%, TI -40%.
2023 ChatGPT moment + AI awakening: November 2022 ChatGPT release catalyzed AI narrative. NVDA earnings revisions began Q1 2023. SMH rallied 80 percent in 2023.
2024-2025 AI bull: NVDA +12x peak-to-peak. SMH +100 percent additional rally. AI capex narrative dominated.
2026: SMH near $482 (recent highs). 10Y stable 4.31 percent. AI narrative continues.
The pattern: duration framework worked 2018, 2020 COVID, 2022. AI capex narrative override emerged 2023-2026. Configuration shifts every cycle based on dominant capex narrative.
How the Pair Performs in Stress
Stress history.
2008-09 GFC: SMH -53 percent peak-to-trough. 10Y fell 4.5 percent to 2.0 percent. Pure liquidity crisis hit semis.
2018 Q4 Fed pivot: SMH -25 percent. Classical duration framework.
2020 COVID flash crash: SMH -38 percent peak-to-trough. Recovered + rallied 100 percent through 2021.
2022 hiking + China + auto demand: SMH -38 percent peak-to-trough October 2022. Multiple negatives: rate rise, China lockdown, auto demand softness, post-COVID PC/smartphone demand collapse.
2023 SVB: SMH -8 percent briefly. Pre-AI awakening still in process.
2024-2026 AI era: SMH +3x peak-to-peak. NVDA +12x. AI capex catalyzed unprecedented rally.
2026 Iran war: SMH held near highs. AI capex narrative immune to geopolitical shock.
The pattern: pre-2024, duration framework dominated. 2024-2026 AI capex catalyst providing earnings revisions overrode rate sensitivity. Forward stress catalysts: AI capex disappointment (most concerning); China-Taiwan geopolitical shock (TSMC concentration risk); 10Y above 5.5 percent (would test duration override).
Volatility and Trading
SMH realized volatility approximately 28-40 percent annualized. Higher than XLK (18-25 percent), SPY (13-18 percent). Beta to SPY approximately 1.5-1.8 over 2020-2026.
60-day rolling correlation between SMH and 10Y: -0.30 to -0.50 in 2024-2026 (negative as expected from duration but weakened by AI narrative). Pre-2024 correlation -0.50 to -0.70 stronger.
For pair-trade implementation: SMH exposure through SMH ETF (most liquid semi ETF, AUM approximately $25 billion) or SOXX (iShares Semiconductor, similar exposure). 10Y yield exposure through TLT or 10Y futures (TY).
The pair has produced cyclical returns. 2024-2026 long SMH / short TLT gained substantially (SMH +3x, TLT relatively flat). 2022 hiking long TLT / short SMH gained substantially (SMH -38%, TLT -50% peak-to-trough then recovered). 2008-09 GFC long TLT / short SMH gained dramatically (SMH -53%, TLT +37%).
Most actionable: monitor AI capex signals. NVDA quarterly earnings (Feb, May, Aug, Nov releases). Hyperscaler capex guidance (MSFT, META, GOOGL, AMZN quarterly). TSMC monthly revenue (early month release). ASML quarterly orders.
Reading the Pair as a Trading Tool
For macro allocators, SMH-vs-10Y provides semi cycle classification.
10Y rising + SMH rallying: AI capex narrative dominant (current 2024-2026). Long SMH benefits.
10Y rising + SMH falling: classical duration framework. Pre-2024 typical (2018, 2022). Long short-duration positioning.
10Y falling + SMH rallying: confirmed bull. Long SMH + long TLT.
10Y falling + SMH falling: rare. Often signals demand weakness (2008 GFC, 2020 COVID flash crash).
April 2026 setup: 10Y 4.31 percent + SMH $482 (near recent highs). AI capex narrative dominant. Long SMH continues.
Key watches. NVDA earnings (May 2026 next release). Hyperscaler capex guidance (Microsoft, Meta, Alphabet, Amazon Q1 2026 earnings April-May). TSMC monthly revenue. ASML quarterly orders. China-Taiwan geopolitical developments. 10Y trajectory (sustained above 5 percent would test duration reassertion).
For positioning: long SMH through SMH ETF or individual names (NVDA, AVGO, AMD, ASML). Hedge through long TLT for downside duration protection. Watch AI capex sustainability signals for any inflection.
How SMH-vs-10Y Compares to Other Sector-vs-Rates Pairs
SMH/10Y captures longest-duration tech sub-segment. Compared to other sector-vs-rates pairs.
Vs XLK/10Y: XLK broader tech (AAPL, MSFT, NVDA, AVGO, ORCL ~65 percent of XLK). SMH narrower semis-only. SMH higher beta to AI capex narrative; XLK has broader tech mix.
Vs XLF/10Y: XLF positive rate sensitivity (NIM expansion). Opposite directional preference.
Vs XLU/10Y: XLU bond proxy + AI data center demand (utility benefits from data center electricity). XLU + SMH both benefit from AI capex but through different channels.
Vs XLRE/10Y: XLRE REITs duration-sensitive. Data center REITs (EQIX, DLR) overlap with AI capex narrative but limited to ~18 percent of XLRE.
For allocator monitoring, SMH/10Y is foundational AI capex cycle indicator. April 2026 reading: 10Y 4.31 percent + SMH $482 (AI capex dominant). Pair complements XLK/10Y (broader tech), XLF/10Y (NIM beneficiary), XLU/10Y (data center utility), XLRE/10Y (data center REIT) for comprehensive AI capex theme read.
Forward View: Watch NVDA Earnings and Hyperscaler Capex
SMH ~$482 (April 2026). 10Y yield 4.31 percent. NVDA ~20 percent of SMH (revenue +62 percent YoY Q3 fiscal 2026). Hyperscaler AI capex $400B+ annual run-rate (Microsoft, Meta, Alphabet, Amazon). NVDA +12x peak-to-peak (2023-2025).
Forward-looking through 2026: NVDA next earnings (May 2026). Hyperscaler Q1 2026 earnings (April-May 2026) will signal capex sustainability. Sustained $400B+ run-rate supports continued SMH outperformance vs 10Y.
Key watches: NVDA quarterly results (May 2026 next release). Hyperscaler Q1 2026 earnings. TSMC monthly revenue. ASML quarterly orders + book-to-bill. China export restriction developments.
Key risks: AI capex disappointment from hyperscalers (most concerning - would compress SMH); China-Taiwan geopolitical shock (TSMC concentration); 10Y above 5.5 percent (would test duration override); demand weakness in PC/smartphone affecting Memory + Mobile semis; competitive intensity (Anthropic/OpenAI alternatives to NVDA).
Expected SMH range $450-$510 absent major catalyst. 10Y range 4.0-4.5 percent. Configuration suggests continued AI capex narrative supporting SMH absent capex disappointment or rate surge.
The Semiconductor Cycle Mechanics
Semis are inherently cyclical. Three-year cycle pattern typically: capacity expansion → oversupply → underutilization → demand recovery → tight supply → price acceleration → capacity expansion (repeat).
2024-2026 AI cycle has lengthened pattern. Pre-AI semi cycles: 24-36 months peak-to-peak. AI cycle has multi-year buildout phase (3-5 year duration). Hyperscaler capex multi-year commitment.
Key indicators. Inventory levels: semi inventory typically 80-120 days. Above 120 days = excess; below 80 = tight. Book-to-bill ratio: above 1.0 = orders exceeding shipments (positive); below 1.0 = orders lagging (negative). ASML book-to-bill 1.2-1.4 in 2024-2026 (very strong). Foundry utilization: TSMC utilization 90+ percent in 2024-2026 (near full). Customer capex announcements: hyperscaler quarterly guidance.
April 2026 setup: AI cycle in expansion phase. Foundry utilization near full. Inventory typical. Book-to-bill above 1.0. Hyperscaler capex strong. Configuration suggests continued cycle expansion.
Watch for cycle inflection signals: hyperscaler capex moderation; foundry utilization decline below 80 percent; book-to-bill below 1.0; inventory above 120 days. April 2026 no inflection signals.
Conditional Forward Response (Tail Events)
How 10Y Treasury Yield has historically behaved in the 5 sessions following a top-decile or bottom-decile daily move in Semiconductors (SMH). Computed from 1,242 aligned daily observations ending .
Following these triggers, 10Y Treasury Yield rises 0.58% on average over the next 5 sessions, versus an unconditional baseline of +0.50%. 124 qualifying events; 10Y Treasury Yield closed positive in 52% of them.
Following these triggers, 10Y Treasury Yield rises 1.27% on average over the next 5 sessions, versus an unconditional baseline of +0.50%. 124 qualifying events; 10Y Treasury Yield closed positive in 57% of them.
Past behavior in the tails is descriptive, not predictive. Mean response is the simple arithmetic mean of compounded 5-day forward returns following each trigger event; baseline is the unconditional mean across the full sample window. Edge measures the gap between the two.
90-Day Statistics
Explore Each Metric
Related Scenarios & Forecasts
Get daily macro analysis comparing key metrics delivered to your inbox. Stay ahead of market-moving divergences.
Frequently Asked Questions
What are SMH and the 10Y Treasury yield?+
SMH (VanEck Semiconductor ETF) tracks 25 largest US-listed semiconductor companies. Top holdings: NVDA ~20%, TSMC ~13%, AVGO ~8%, AMD ~5% (top 4 ~46% high concentration). SMH ~$482 April 2026 (near recent highs). 10Y Treasury yield 4.31% April 2026. SMH +3x since November 2022 ChatGPT moment. AI capex narrative drove NVDA +12x peak-to-peak (2023-2025). Hyperscaler capex on AI infrastructure $400B+ annual run-rate (Microsoft, Meta, Alphabet, Amazon). NVDA Q3 fiscal 2026 revenue +62% YoY. SMH AUM ~$25B. SMH realized volatility ~28-40% annualized (vs XLK 18-25%, SPY 13-18%).
Why are semis most rate-sensitive tech subsector?+
Three drivers. Growth profile: AI capex 5-15 year payoff horizon. Data center buildout (NVDA, AVGO, AMD) multi-year. Foundry capex (TSMC) 3-5 year construction. Multiple expansion: SMH forward P/E ~30x (April 2026) vs SPY 22x. Higher multiple reflects longer-duration cash flows. Capital intensity: semis require substantial capex for fab construction. Higher discount rates compress NPV more than software. Pre-2024 empirical: 100bp 10Y rise = 8-15% SMH decline (60-90 day windows). XLK 5-10%; SPY 3-5%. Semis 1.5-2x duration sensitivity vs XLK. 2022 hiking 10Y 1.5% to 5.0%: SMH -38% (NVDA -65%, AMD -55%). 2024-2026 anomaly: AI capex override.
What is the AI capex boom?+
Largest tech investment cycle since dot-com era. Hyperscaler capex (top 4): Microsoft $80B+ FY26 (vs $50B FY24); Meta $60-65B 2026 (vs $40B 2024); Alphabet $75-80B 2026 (vs $50B 2024); Amazon $100B+ 2026 (vs $48B 2023). Combined ~$315-325B run-rate, up from $200B (2022). Plus Tesla, Apple, Oracle add ~$80B+. Total US AI capex ~$400B+ annual run-rate. Semi beneficiaries: NVDA H100/H200/Blackwell GPUs dominate AI training (revenue +62% YoY); AVGO networking + custom AI chips; AMD MI300 GPUs; TSMC foundry to all advanced AI chips; ASML EUV lithography monopoly ($200-400M per machine); Applied Materials, Lam Research fab equipment; Micron, SK Hynix HBM for AI accelerators.
How does the pair perform through rate cycles?+
2018-2019: 10Y 2.4% to 3.2% (Q4 2018) then 1.5% (mid-2019). SMH -25% Q4 2018 then +60% through 2019. Classical duration framework. 2020-2021 COVID: 10Y 1.9% to 0.5% then 1.7%. SMH -38% peak-to-trough COVID then +100% through 2021. Strong pattern. 2022 hiking: 10Y 1.5% to 5.0% peak. SMH -38% peak-to-trough October 2022. Classical duration. NVDA -65%, AMD -55%, Intel -50%, TI -40%. 2023 ChatGPT moment + AI awakening: November 2022 release catalyzed narrative. SMH +80% in 2023. 2024-2025 AI bull: NVDA +12x. SMH +100% additional. 2026 SMH ~$482. AI narrative continues. Pattern: duration worked 2018, 2020 COVID, 2022. AI capex override 2023-2026.
How does the pair perform in stress?+
2008-09 GFC: SMH -53% peak-to-trough. 10Y 4.5% to 2.0%. Pure liquidity crisis. 2018 Q4 Fed pivot: SMH -25%. Classical duration. 2020 COVID flash crash: SMH -38% peak-to-trough then +100% through 2021. 2022 hiking + China + auto demand: SMH -38% peak-to-trough October 2022. Multiple negatives: rate rise, China lockdown, auto demand softness, post-COVID PC/smartphone collapse. 2023 SVB: SMH -8% briefly. 2024-2026 AI era: SMH +3x peak-to-peak; NVDA +12x. AI capex catalyzed unprecedented rally. 2026 Iran war: SMH near highs (AI immune to geopolitical). Pattern: pre-2024 duration dominated; 2024-2026 AI override. Forward stress catalysts: AI capex disappointment, China-Taiwan, 10Y above 5.5%.
How is the pair traded?+
SMH realized volatility ~28-40% annualized (vs XLK 18-25%, SPY 13-18%). Beta to SPY 1.5-1.8 over 2020-2026. 60-day correlation SMH-10Y: -0.30 to -0.50 in 2024-2026 (negative but weakened by AI narrative); pre-2024 -0.50 to -0.70. SMH exposure: SMH ETF (AUM ~$25B) or SOXX (similar). 10Y exposure: TLT or 10Y futures (TY). 2024-2026 long SMH / short TLT gained substantially (SMH +3x, TLT relatively flat). 2022 hiking long TLT / short SMH gained (SMH -38%, TLT -50% peak-to-trough). 2008-09 GFC long TLT / short SMH gained dramatically (SMH -53%, TLT +37%). Most actionable: NVDA quarterly (Feb, May, Aug, Nov), hyperscaler capex guidance, TSMC monthly revenue, ASML quarterly orders.
How is the pair used for trading?+
10Y rising + SMH rallying: AI capex narrative dominant (current 2024-2026). Long SMH benefits. 10Y rising + SMH falling: classical duration framework. Pre-2024 (2018, 2022). Long short-duration. 10Y falling + SMH rallying: confirmed bull. Long SMH + long TLT. 10Y falling + SMH falling: rare. Demand weakness (2008 GFC, 2020 COVID flash crash). April 2026: 10Y 4.31% + SMH $482 (near recent highs). AI capex dominant. Long SMH continues. Watch NVDA May 2026 earnings, hyperscaler Q1 2026 earnings (April-May), TSMC monthly revenue, ASML orders, China-Taiwan, 10Y above 5%. Long SMH (NVDA, AVGO, AMD, ASML) + hedge through TLT for downside protection.
How does SMH compare to other rate-sensitive sectors?+
Vs XLK/10Y: XLK broader tech (AAPL, MSFT, NVDA, AVGO, ORCL ~65% of XLK). SMH narrower semis-only. SMH higher beta to AI capex narrative; XLK has broader tech mix. Vs XLF/10Y: XLF positive rate sensitivity (NIM expansion). Opposite directional preference. Vs XLU/10Y: XLU bond proxy + AI data center demand (utility benefits from data center electricity). XLU + SMH both benefit from AI capex but through different channels. Vs XLRE/10Y: XLRE REITs duration-sensitive. Data center REITs (EQIX, DLR) overlap with AI capex narrative but limited to ~18% of XLRE. April 2026 reading: 10Y 4.31% + SMH $482 (AI capex dominant). Foundational AI capex cycle indicator.
Related Comparisons
Explore Across Convex
Data sourced from FRED, CoinGecko, CBOE, and other providers. This page is for informational purposes only and does not constitute financial advice. Past performance does not guarantee future results.