HALO (physical economy) vs AI mega-caps
AI and Big-Tech stocks capture the upside of artificial intelligence; physical-economy (HALO) stocks own what AI cannot replace. This page compares the two side by side (what you own, AI exposure, valuation, and risk) and explains why many investors hold both rather than choosing one.
How do HALO stocks and AI mega-caps differ?
| Dimension | Physical-economy (HALO) | AI & Big-Tech mega-caps |
|---|---|---|
| What you own | Mines, railroads, billboards, distribution networks | Software, chips, platforms, and AI models |
| Representative tickers | SCCO, JBHT, LAMR, GWW, CNI, CMI, DAL | AAPL, MSFT, NVDA, AMZN, GOOGL, META, TSLA (the Magnificent Seven) |
| AI exposure | Low: physical, hands-on work is least exposed to LLMs | High: both the upside and the disruption risk |
| Style | Value and real-asset; often cheaper, inflation-aware | Growth; often richly valued |
| Earnings engine | Commodity- and freight-cycle linked; trucks still moved 72.7% of U.S. freight tonnage in 2024 (ATA) | Q4 2025 earnings grew 27.2% vs 9.8% for the other 493 S&P 500 firms; 23.5% expected for CY 2026 (FactSet) |
| Capex profile | Steady, maintenance-heavy reinvestment: U.S. freight railroads invest about $25 billion a year (AAR) | AI arms race: Amazon plans about $200 billion and Meta $115-135 billion of capex in 2026 (SEC 8-Ks) |
| Volatility profile | Cyclical drawdowns that track the real economy: copper, freight, and ad budgets | Narrative-driven repricings: Nvidia fell 17%, almost $600 billion, in one day on DeepSeek news (Jan 2025) |
| Index footprint | Small individual S&P 500 weights; adds breadth to a cap-weighted core | Magnificent Seven at about 34.8% of the S&P 500 (May 2026) |
| Main risk | Commodity cyclicality and capital intensity | Concentration, disruption, and multiple compression |
| Cash-flow driver | Replacement-cost moats and real-world demand | Network effects, scale, and R&D |
When do AI mega-caps make sense?
Big-Tech and AI mega-caps make sense for investors who want direct exposure to the AI build-out and can tolerate concentration and rich valuations. The concentration is not hypothetical: the Magnificent Seven (Apple, Microsoft, Nvidia, Amazon, Alphabet, Meta, and Tesla) accounted for about 34.8% of the S&P 500 as of May 2026, up from roughly 12.5% in 2016 (The Motley Fool). That means a standard index fund already tilts heavily toward a handful of AI-exposed names, so any AI setback hits both the active bet and the “passive” core.
The case for accepting that concentration is an earnings engine the rest of the market cannot match. The Magnificent Seven grew earnings 27.2% in the fourth quarter of 2025, against 9.8% for the other 493 companies in the S&P 500, and analysts expect the gap to persist: 23.5% growth for the group in calendar 2026 versus 11.8% for everyone else (FactSet). The premium multiple buys roughly twice the measured growth rate.
That growth is being purchased with some of the largest capital-spending programs in corporate history. Amazon expects to invest about $200 billion in capital expenditures in 2026 (Amazon Q4 2025 results, SEC Form 8-K). Meta guides to $115-135 billion for 2026, nearly double the $72.22 billion it spent in 2025 (Meta Q4 2025 results, SEC Form 8-K), and Alphabet purchased $91.4 billion of property and equipment in 2025 alone (Alphabet Q4 2025 results, SEC Form 8-K). A mega-cap shareholder owns both the AI revenue and the obligation to earn a return on that outlay.
With such strong demand for our existing offerings and seminal opportunities like AI, chips, robotics, and low earth orbit satellites, we expect to invest about $200 billion in capital expenditures across Amazon in 2026, and anticipate strong long-term return on invested capital.
— Andy Jassy, CEO, Amazon (Q4 2025 results, SEC Form 8-K)
When does HALO (physical economy) make sense?
Physical-economy (HALO) names make sense as a real-asset counterweight: their value rests on the least AI-exposed parts of the economy, since research puts physical, hands-on work at the bottom of the task-exposure scale (Eloundou et al., Science 2024). Their cash flows are anchored in real-world demand: the energy transition alone is set to strain hard-asset supply, with the IEA projecting announced copper projects falling roughly 30% short of demand by 2035 (IEA, Global Critical Minerals Outlook 2025).
About 80% of U.S. workers have at least 10% of their tasks exposed to large language models, with physical, hands-on work the least exposed.
The assets behind these cash flows are productive and expensive to replicate, which is the moat. U.S. freight railroads privately invest about $25 billion a year in their networks and reinvested roughly $840 billion of their own funds between 1980 and 2024, about six times more than the average U.S. manufacturer as a share of revenue (Association of American Railroads). The demand on top of those assets is stubbornly physical: trucks still moved 72.7% of U.S. domestic freight by tonnage in 2024 (American Trucking Associations), and a single electric car uses up to four times the copper of a gasoline one (IEA, Global EV Outlook 2025). None of that demand depends on AI staying expensive; if anything, the data-center build-out adds another buyer for the same copper and freight.
What does AI concentration risk look like in practice?
January 27, 2025 is the cleanest case study. After Chinese startup DeepSeek released an open-source model it said took two months and less than $6 million to train, Nvidia fell 17% and lost nearly $600 billion of market value in a single session, the largest one-day market-cap loss in U.S. history, with Broadcom dropping 17% alongside it (CNBC). One research claim repriced the entire AI trade in a day, and with the Magnificent Seven near 34.8% of the S&P 500, the move reached every index investor. A copper mine or a rail right-of-way is not repriced by a cheaper training run, which is the structural difference between the two sides of this comparison rather than a quirk of one bad day.
Which investors are better suited to Physical-economy (HALO) versus AI & Big-Tech mega-caps?
Big-Tech captures the AI upside but carries concentration and disruption risk; physical-economy (HALO) names own what AI cannot easily replace. The data cuts both ways: Magnificent Seven earnings are growing roughly twice as fast as the rest of the S&P 500, yet a single AI headline erased almost $600 billion of Nvidia’s value in one session. Many investors hold both: Big-Tech for growth, HALO as a real-asset, AI-resilient counterweight.
Related concepts & themes
- AI-Resilient Investing parent
- Heavy Asset Low Obsolescence (HALO) related
FAQ
Should I own AI stocks or AI-resilient stocks?
They are not mutually exclusive. AI and Big-Tech mega-caps offer exposure to the AI build-out, while physical-economy (HALO) names offer a counterweight built on assets AI cannot easily replace. Many portfolios hold both, sizing each to goals and risk tolerance. One reason to size deliberately: the Magnificent Seven made up about 34.8% of the S&P 500 as of May 2026, so a plain index fund already carries heavy AI mega-cap exposure.
Is HALO a hedge against an AI bubble?
It is better described as a counterweight than a hedge. HALO does not move opposite to AI stocks, but its value rests on physical assets and cash flows that do not depend on AI staying expensive, which can diversify a portfolio that is heavy in AI and Big-Tech.
How much are the AI mega-caps spending on AI infrastructure?
The disclosed plans are among the largest capital-spending programs in corporate history. Amazon expects to invest about $200 billion in capital expenditures in 2026 (Amazon Q4 2025 results, SEC Form 8-K), Meta guides to $115-135 billion for 2026, up from $72.22 billion in 2025 (Meta Q4 2025 results, SEC Form 8-K), and Alphabet spent $91.4 billion on property and equipment in 2025 alone (Alphabet Q4 2025 results, SEC Form 8-K). That spending is the bull case for AI suppliers and the open question for mega-cap shareholders, who need those dollars to earn a return.
Do AI mega-cap earnings justify their premium valuations?
The growth gap is real and measured: the Magnificent Seven grew earnings 27.2% in Q4 2025 versus 9.8% for the other 493 S&P 500 companies, and analysts expect 23.5% versus 11.8% growth for calendar 2026 (FactSet). Whether that justifies the premium depends on how durable AI demand proves; because the gap is already priced, any narrowing can compress the multiple at the same time it slows the earnings.
What does AI concentration risk look like in practice?
January 27, 2025 is the cleanest example. After Chinese startup DeepSeek released a model it said cost under $6 million to train, Nvidia fell 17% and shed nearly $600 billion of market value in a single session, the largest one-day market-cap loss in U.S. history (CNBC). With the Magnificent Seven at roughly 34.8% of the S&P 500 (The Motley Fool), a repricing like that reaches even plain index investors.
Which ETFs, funds, or listed securities provide access to Physical-economy (HALO) and AI & Big-Tech mega-caps?
The AI mega-cap side is the Magnificent Seven: Apple, Microsoft, Nvidia, Amazon, Alphabet, Meta, and Tesla, which a cap-weighted S&P 500 or Nasdaq-100 fund already holds in size. Representative physical-economy (HALO) names include Southern Copper (SCCO), J.B. Hunt (JBHT), Lamar Advertising (LAMR), W.W. Grainger (GWW), Canadian National Railway (CNI), and Cummins (CMI). The Roundhill HALO ETF (LOHA) packages the basket by tracking the Akros U.S. HALO Index, with a 0.35% expense ratio.
Sources & references
- GPTs are GPTs: Labor market impact potential of LLMs · Eloundou, Manning, Mishkin & Rock · Science, 2024-06-21
- The Magnificent Seven's Market Cap vs. the S&P 500 · The Motley Fool, 2026-05-12
- Global Critical Minerals Outlook 2025 (copper supply gap) · International Energy Agency, 2025-05-21
- Amazon.com Announces Fourth Quarter Results (Form 8-K, Exhibit 99.1) · Amazon.com, Inc., 2026-02-05
- Meta Reports Fourth Quarter and Full Year 2025 Results (Form 8-K, Exhibit 99.1) · Meta Platforms, Inc., 2026-01-28
- Alphabet Announces Fourth Quarter and Fiscal Year 2025 Results (Form 8-K, Exhibit 99.1) · Alphabet Inc., 2026-02-04
- "Magnificent 7" Companies Reported Earnings Growth Above 25% for Q4 · FactSet Research Systems, 2026-02-27
- Nvidia sheds almost $600 billion in market cap, biggest one-day loss in U.S. history · CNBC, 2025-01-27
- Freight Rail Investments (annual private network investment) · Association of American Railroads, 2025-06-01
- American Trucking Trends 2025 (share of U.S. freight tonnage by mode) · American Trucking Associations, 2025-06-01
- Global EV Outlook 2025 (copper intensity of electric vehicles) · International Energy Agency, 2025-05-01
- Roundhill HALO ETF (LOHA) · fund objective · Roundhill Investments, 2026-05-14