Your Capex is My Opportunity: The AI Capex Cycle Has Replaced the Old Economy Business Cycle
Your Capex is My Opportunity: The AI Capex Cycle Has Replaced the Old Economy Business Cycle
YouTube43 min 48 sec
Watch on YouTube
Note: AI-generated summary based on third-party content. Not financial advice. Read more.
Quick Insights

Investors should prioritize hardware and infrastructure over traditional software, specifically targeting Micron (MU) which is undervalued at a 5x P/E due to critical memory shortages. To capture the "Five-Layer Cake" of AI, look toward specialized chemical providers like Entegris (ENTG) and Chemours (CC), the latter of which has the potential to double this year. Industrial giant Caterpillar (CAT) serves as a high-conviction play on data center power needs, backed by a record $62 billion backlog and a projected tripling of power gen sales by 2030. In the digital asset space, monitor Ethereum (ETH) for two consecutive closes above $2,456 as a definitive breakout signal for a year-end rally. To hedge against rising inflation and power demands, maintain exposure to energy leaders Chevron (CVX) and Exxon (XOM), which are currently lagging behind oil price gains.

Detailed Analysis

AI Infrastructure & Semiconductors

The market is currently driven by a massive CapEx (Capital Expenditure) boom in AI, which has replaced the traditional business cycle. This cycle is characterized by "Your CapEx is my opportunity," where the massive spending of "Hyperscalers" becomes the revenue of the infrastructure providers.

  • Trillions in Backlog: The three major hyperscalers (Amazon, Google, Microsoft) have a combined cloud backlog of $1.3 trillion.
  • The "Five-Layer Cake" Framework: Investment should be viewed through five layers of AI infrastructure:
    1. Advanced Packaging/Semis: (Early cycle - hoarding phase).
    2. Optical Fiber: Essential for data center connectivity.
    3. Chemicals: Critical for advanced packaging and fiber tubing.
    4. Power/Grid: The base layer; includes power semiconductors and generation.
    5. Humanoids/Agents: The late-cycle phase where AI starts solving physical world problems.
  • Semiconductors as the New Beta: Semis are now the largest sector weighting in the S&P at $10 trillion.
  • NVIDIA (NVDA): Interestingly, while the stock is high, its forward P/E is at 10-year lows because earnings are growing faster than the price.
  • Micron (MU): Highlighted as "too cheap" at a 5x P/E off next year's earnings due to severe DRAM/memory shortages.

Takeaways

  • Shift from Software to Hardware: Traditional software companies (Salesforce, Adobe) are facing margin pressure as they must spend more on AI to stay competitive. Investors should consider underweighting "old" software in favor of hardware/infrastructure.
  • Focus on Power Semis: Look for companies involved in thermal management and power efficiency for AI servers.
  • Ignore "Bubble" Talk: The current rally is driven by earnings revisions and massive contracted backlogs, not just speculation.

Industrial & Capital Goods

The AI boom is spilling over into the "old economy" because data centers require massive physical construction and power generation.

  • Caterpillar (CAT): A key indicator of the AI build-out. Management expects sales of power generation equipment for data centers to triple by 2030.
  • Backlog: CAT has a record $62 billion backlog, providing high visibility into future demand.
  • Valuation: Trading at 36x forward earnings, which is high historically, but justified by the 2030 demand outlook.

Takeaways

  • Industrial AI Play: Treat heavy machinery and power equipment companies as indirect AI plays.
  • Monitor Backlogs: As long as backlogs for power generation remain at record highs, the industrial side of the AI trade remains intact.

Chemicals & Energy

These sectors represent the "mid-to-late cycle" opportunities that are currently under-owned compared to semiconductors.

  • Chemicals: Essential for the polymers and tubing used in optical fiber and advanced chip packaging.
  • Integris (ENTG): Mentioned as a specific name the analyst purchased following an earnings dip.
  • Chemours (CC): Predicted to potentially double this year due to its role in the AI supply chain.
  • Energy Stocks (CVX/XOM): Chevron and Exxon are viewed as lower-beta ways to play the power needs of AI. Energy stocks have lagged the recent move in oil prices, creating a potential catch-up opportunity.

Takeaways

  • Chemical Layer: Look for specialized chemical companies involved in semiconductor manufacturing materials.
  • Energy as a Hedge: Energy stocks provide a "late cycle" cushion and benefit from the rising "nominal GDP" environment.

Bitcoin (BTC) & Ethereum (ETH)

The crypto market is currently in a "boring" phase as retail interest has shifted temporarily to AI stocks, but the structural setup remains bullish.

  • Bitcoin (BTC): Currently finding support. A key macro trigger will be when CPI inflation moves above 3-month Treasury yields, creating a "run it hot" environment favorable for BTC.
  • Ethereum (ETH): Viewed as the more important story for the remainder of the year.
    • Valuation: Unlike BTC, ETH can be valued via discounted cash flow based on network usage.
    • Catalyst: The "Clarity Act" and developments in stablecoins/programmable money are major tailwinds.
  • MicroStrategy (MSTR): The analyst mentioned buying long-dated calls (end of year) rather than short-term trades.

Takeaways

  • Portfolio Allocation: BTC and ETH should be treated as essential "growth asset" allocations for diversified portfolios.
  • Watch the 200-Week MA: For Ethereum, look for two consecutive closes above $2,456 as a definitive breakout signal.

Macro Risks & Warning Signs

Despite the bullish AI theme, several "turbulence signals" are emerging.

  • Inflation/Rates: Gas and diesel prices are trending higher, suggesting CPI could stay above 4%. This may force the Fed to keep rates higher for longer or even consider hikes.
  • Market Breadth: The market is "thin." The median S&P stock is still 13% below its 52-week high, meaning the rally is concentrated in just a few names.
  • Airlines: Identified as a high-risk sector due to rising fuel costs and slowing air travel demand.

Takeaways

  • Benchmark Arbitrage: Traditional benchmarks (S&P 500) are weighted toward "yesterday's winners" (Software/Consumer). Investors may need to manually tilt portfolios toward the "Five-Layer Cake" of AI to outperform.
  • Watch 10-Year Yields: If yields break above 5%, it could trigger a broader market deleveraging event.
Ask about this postAnswers are grounded in this post's content.
Video Description
The market continues to rally despite the Iran War and the look of an “AI bubble” because the underlying earnings story is still accelerating. AI is no longer just lifting a handful of tech stocks; it is driving GDP growth, EPS revisions, margins, semiconductor demand, industrial backlogs, hyperscaler cloud commitments, and the next leg of capital spending. The key point is that the old consumer/industrial business cycle framework is no longer the right lens. Instead, the market is moving through a new AI business cycle: early cycle is semiconductors, advanced packaging, optical fiber, and rack infrastructure; midcycle is power, data centers, chemicals, energy, and physical buildout; late cycle is applications, agents, humanoids, and broader productivity gains. At the same time, the the rally is not risk-free and is sending some warning signals to watch. Inflation signals are rising through oil, gasoline, diesel, crop prices, prices paid indexes, inflation swaps, and global bond yields. Market breadth is weak, financials remain under pressure, software continues to lag, and the turbulence model has issued its first warning sign, though not yet a full trigger. The portfolio implication is “benchmark arbitrage”: traditional benchmarks are still overweight the winners of the last cycle, especially software and consumer-led growth, while underweight the physical AI buildout winners in semis, power, chemicals, optics, and infrastructure. Bitcoin and Ethereum are framed as additional growth assets that may benefit if inflation runs hot, the Fed stays constrained, and programmable money gains traction. Timestamps: (00:00–03:13) Setup: markets rally despite "AI bubble" talk in LA/DC; AI driving earnings, revisions, margins, and GDP higher but warning signs emerging (turbulence model's first cross-asset signal, financials still below 200DMA, software/PE near YTD lows). Benchmarks built for the old cycle are wrong for the AI world. (03:13–06:30) S&P, NASDAQ, and small caps at all-time highs, one of the best months in 25 years; forward EPS accelerating without a prior draw-down (unique vs. mid-90s and dot-com); analysts revising estimates up after 15 years of averaging -2%; capital goods exploding year-over-year, PMI close to 60 in regional surveys. (06:30–10:40) Caterpillar's $62B backlog and 3x power-generation sales forecast through 2030; PE at 36x next year while Nvidia trades at a 10-year low on forward PE. (10:40–15:50) Inflation warning signs: unleaded gas futures break out and back months rise (inflation persistence); September gas at $87; Goldman models show global oil inventories hitting all-time lows by mid-year if the Strait stays closed (15:50–19:30) China exporters raising prices, crop/fertilizer breakouts, El Niño forecast, the 1970s analog (nominal GDP up, inflation up, no jobs); inflation swaps rising, 30Y UK gilts and 10Y JGB yields at new highs; breath narrowing, median S&P stock 13% below 52-week high, 50% down YTD, financials still below 200DMA. (19:30–23:00) Semis are now the largest Level-2 GICS in the S&P at $10T (Nvidia ~half and a relative laggard); "your margin is my opportunity" of the last 15 years has been replaced by "your capex is my opportunity"; $90T five-layer cake; the AI business cycle has new phases, early (semis), mid (power, data centers), late (humanoids). (23:00–26:30) Five-layer cake breakdown: rack / advanced packaging / optical fiber (early); chemicals (mid); power (late). (26:30–30:30) Compute shortage is real: Anthropic at a $30B run rate forecasting $50–100B by year-end; OpenAI says there isn't enough compute in the world; White House officials worried (30:30–35:00) Three end-results for advanced semis (Intel, Samsung, TSMC); Elon's Samsung buyout and terafab build because catching up isn't possible; White House Section 303 Defense Production Act for the grid; US power plant equipment spend forecast to triple through 2030; power semis broke out to a 5-year high; chemicals overlaid with Texas Instruments show the same pattern. (35:00–39:10) Chemical layer report — 17 names with exposure scoring across advanced packaging, fiber tubing, and polymers; Integris bought after earnings drawdown; Camores up ~50%. Adam Parker on Compound & Friends: software sequence is multiples compress → earnings miss → sales miss; analysts still assume 80% gross margins through 2028–29, too static for an AI-disrupted world. Naval Ravikant says the same about Apple's terminal value. (39:10–43:38) Bitcoin MACD weekly trying to break out — up 33% from the $60K lows to $80K but losing relative interest as Korea/Asia trades AI; trigger remains CPI above 3-month yields; MicroStrategy calls bought into year-end. Ethereum is the main story for the rest of the year if the Clarity Act passes — programmable money, the Door Dash stablecoin decision, Ethereum's 200-week MA at 2456 is the level (need two closes above).
About Jordi Visser
Jordi Visser

Jordi Visser

By @jordivisserlabs

Empowering seasoned professionals to navigate the future of finance, technology, and AI. What We Offer: - Cutting-edge ...