Heavy Asset Low Obsolescence (HALO)

Heavy Asset Low Obsolescence (HALO) is a simple idea: own the parts of the economy that artificial intelligence cannot copy or replace. Copper mines, freight railroads, billboards, parts warehouses. These are U.S.-listed companies whose value sits in physical assets and real-world operations, not in software that a model can reproduce.

CategoryThematic / AI-resilient real economy
Representative companies10
Related ETFRoundhill HALO ETF (LOHA)
Last updated2026-06-04
Key takeaways
  • HALO (Heavy Asset Low Obsolescence) groups U.S.-listed companies whose value sits in physical assets AI cannot easily copy or automate.
  • Representative names span mining (Southern Copper), freight and rail (J.B. Hunt, Canadian National), engines (Cummins), airlines (Delta), HVAC (Lennox), auto parts (Magna), materials (Cabot), billboards (Lamar), and industrial distribution (Grainger).
  • Academic research places physical, hands-on work at the bottom of AI task-exposure, which is the intellectual basis for the concept.
  • The Akros U.S. HALO Index is tracked by the Roundhill HALO ETF (LOHA); the holdings are cyclical, capital-heavy, and suited to long horizons.

What is Heavy Asset Low Obsolescence?

HALO starts from one question. What does a company own that an AI can’t copy? Software, text, code, images, a model can reproduce all of it. A copper mine, a rail line, a national network of parts warehouses, it cannot. HALO groups U.S.-listed companies whose value rests on that second kind of asset. “Heavy asset” means a large base of property, plant, and equipment. “Low obsolescence” means the asset keeps earning even as the technology around it changes.

Why HALO matters now

For two years the market has paid up for anything software can automate and marked down anything it might replace. HALO looks at the other side of that trade. Some businesses are valuable precisely because they are physical and slow to copy, and the numbers behind them are not small. Trucks still moved 72.7% of U.S. domestic freight by tonnage in 2024 (American Trucking Associations). A single electric car uses up to four times the copper of a gasoline one (IEA, Global EV Outlook 2025). Lamar owns more than 362,000 billboards on land it controls (Lamar 10-K). None of that gets cheaper because a model got better at writing.

Roundhill, which runs an ETF built on this theme, frames the bet directly:

[The companies in the strategy are] “uniquely positioned to withstand the persistent threat of AI disruption.”

— Roundhill Investments, HALO ETF (LOHA) objective

What does the research say?

The thesis has academic backing. In “GPTs are GPTs,” researchers at OpenAI and the University of Pennsylvania estimated that about 80% of U.S. workers have at least 10% of their tasks exposed to large language models, and roughly 19% have at least half their tasks exposed, with higher-paid, college-educated, information-processing jobs most affected (Eloundou et al., Science 2024). The work that scores low on exposure is the physical, hands-on kind: extraction, transport, installation, maintenance. That is the part of the economy HALO buys.

Supply tells a similar story for the materials these companies dig and move. The IEA expects announced copper projects to fall roughly 30% short of demand by 2035 (IEA, Global Critical Minerals Outlook 2025), as electric vehicles, the grid, and AI data centers all pull on the same metal (S&P Global, “Copper in the Age of AI”). Goldman Sachs sees the copper price climbing toward $15,000 a tonne by 2035 (Goldman Sachs Research). You cannot prompt your way out of a metal shortage.

Who should consider HALO?

This suits a long-horizon investor who wants tangible, hard-to-replace businesses as a hedge against AI-concentration risk, owning copper, railroads, billboards, and the warehouses that keep factories running instead of betting on the next model. The thesis has an academic floor: task-exposure research finds about 80% of U.S. workers have at least 10% of their tasks exposed to large language models and roughly 19% have at least half exposed, but physical, hands-on work, extraction, transport, installation, scores lowest (Eloundou et al., Science 2024). The demand under these assets is real and supply-led: the IEA expects announced copper projects to fall roughly 30% short of demand by 2035 (IEA, Global Critical Minerals Outlook 2025), and trucks still moved 72.7% of U.S. domestic freight by tonnage in 2024 (American Trucking Associations).

Size it as a value-tilted or real-asset satellite inside a diversified portfolio, not a parking spot for cash. The same physical assets that make these companies hard to copy also make them cyclical and capital-heavy: copper, freight, and advertising all track the economy, so the basket can fall as hard as anything else in a downturn. It is not for an investor who needs low volatility or a short holding period, and “AI can’t touch this” is a judgment rather than a guarantee, automation can still squeeze costs and headcount inside these industries even when it never replaces the asset. The holdings are liquid U.S. large caps, which keeps the position easy to model, but suitability always depends on horizon, risk tolerance, and existing exposures.

What makes these businesses hard to disrupt?

It comes down to replacement cost and physical work. Building a new copper mine or rail line costs far more than buying an existing one, so incumbents rarely face fresh competition. And the actual job is physical. Freight has to move, copper has to be dug, billboards sit on owned land. An AI can route the trucks better or trim the back office, but it cannot do the work itself. The moats here are old and unglamorous, which for durability is usually a feature rather than a flaw.

Which companies represent HALO?

CompanySectorWhat it does
Southern Copper (SCCO) Materials · Copper Mining Mines, smelts, and refines copper and other metals across Mexico, Peru, and the U.S., sitting on some of the largest copper reserves in the industry. You cannot prompt an AI to dig an orebody.
J.B. Hunt Transport Services (JBHT) Industrials · Trucking & Logistics One of the largest U.S. surface-transport and logistics operators, running intermodal, dedicated fleet, and trucking networks that physically move freight across the country.
Lamar Advertising (LAMR) Real Estate · Outdoor Advertising (REIT) Owns more than 362,000 billboards and outdoor displays across North America. The locations are physical real estate, which is exactly what an ad-generating model cannot reproduce.
W.W. Grainger (GWW) Industrials · Industrial Distribution Distributes maintenance, repair, and operating supplies through a North American network of branches, inventory, and logistics that customers depend on to keep plants running.
Canadian National Railway (CNI) Industrials · Rail Transportation Operates a transcontinental freight rail network spanning Canada and reaching the U.S. Midwest and Gulf Coast; the right-of-way and track base are irreplaceable physical assets.
Cummins (CMI) Industrials · Engines & Power Designs and manufactures diesel and alternative-power engines, components, and power systems; its plants and global service network are capital-intensive and hard to replicate.
Delta Air Lines (DAL) Industrials · Airlines Runs one of the world's largest airlines; aircraft, gates, hubs, and route authority form a capital-intensive physical network that software cannot reproduce.
Lennox International (LII) Industrials · HVAC Equipment Manufactures heating, ventilation, air-conditioning, and refrigeration equipment; factories, distribution, and a large installed base anchor its value in physical hardware and service.
Magna International (MGA) Consumer Discretionary · Auto Parts One of the world's largest automotive suppliers, making body, chassis, and powertrain systems and assembling complete vehicles from a global manufacturing footprint.
Cabot Corporation (CBT) Materials · Specialty Chemicals Produces carbon black, specialty chemicals, and performance materials from capital-intensive plants that supply tires, batteries, inks, and industrial products.

What are the risks?

HALO is not a hiding place. The same physical assets that make these companies hard to copy also make them cyclical and capital-hungry.

  • Cyclicality. Copper, freight, and advertising all track the economy. In a downturn these names can fall as hard as anything else.
  • Capital intensity. Mines, fleets, and distribution networks need constant reinvestment, which eats into free cash flow when demand softens.
  • The thesis can be wrong. “AI can’t touch this” is a judgment, not a fact. Automation can still cut costs and headcount inside these industries, even when it never replaces the asset.
  • Rates and regulation. Capital-heavy businesses feel interest rates, and miners and railroads carry permitting and environmental exposure.

Related concepts, securities & terms

Related indices & ETFs

Frequently asked questions about HALO

What is the HALO investment concept?

HALO (Heavy Asset Low Obsolescence) groups U.S.-listed companies whose value sits in physical assets and real-world operations, like mines, railroads, billboards, and industrial distribution, that AI cannot easily copy. The intellectual basis is task-exposure research: about 80% of U.S. workers have at least 10% of their tasks exposed to large language models, but physical, hands-on work scores lowest (Eloundou et al., Science 2024).

How does HALO relate to AI disruption?

HALO flips the AI question around: instead of asking which companies AI will help, it asks which own things AI cannot reproduce. Research finds roughly 80% of U.S. workers have at least 10% of tasks exposed to LLMs and about 19% have at least half exposed, with the least-exposed work being physical, like extraction and transport (Eloundou et al., Science 2024). A model can write code, but it cannot mine copper or move a freight container.

Which companies are representative of HALO?

Representative names include Southern Copper (SCCO), J.B. Hunt (JBHT), Lamar Advertising (LAMR), W.W. Grainger (GWW), Canadian National Railway (CNI), Cummins (CMI), Delta Air Lines (DAL), Lennox International (LII), Magna International (MGA), and Cabot (CBT). Their value is physical, for example Lamar owns more than 362,000 billboards on land it controls (Lamar 10-K), an asset an ad-generating model cannot reproduce.

What are the main risks of HALO?

These businesses are cyclical and capital-heavy, so they can fall sharply when the economy slows. Demand is also commodity-linked: copper relies on continued capex, and the IEA expects announced copper projects to fall roughly 30% short of demand by 2035 (IEA, Global Critical Minerals Outlook 2025). AI resilience is a thesis, not a guarantee, and automation can still squeeze costs and labor inside these industries.

Are there stocks that are safe from AI disruption?

No stock is fully safe, but some are far more resilient. In task-exposure research about 80% of U.S. workers have at least 10% of their tasks exposed to LLMs, with physical, hands-on work least affected (Eloundou et al., Science 2024). Companies that earn from physical assets, such as miners, railroads, and billboard owners, are much harder for AI to copy than software and content businesses.

How can I invest in companies AI cannot easily disrupt?

One approach is to focus on physical-economy sectors like mining, freight and logistics, industrial distribution, and infrastructure, where demand is tangible: trucks still moved 72.7% of U.S. domestic freight by tonnage in 2024 (American Trucking Associations). The HALO concept groups U.S.-listed examples; the Akros U.S. HALO Index packages them, and the Roundhill HALO ETF (LOHA, 0.35% expense ratio) tracks that index. Always check fees, holdings, and risk before investing.

Is HALO suitable for a fee-only fiduciary client?

It can work as a value-tilted or real-asset satellite for clients who want durable, physical-economy businesses and can accept commodity cyclicality over a multi-year horizon. The structural case is supply-led: the IEA expects announced copper projects to fall roughly 30% short of demand by 2035 (IEA, Global Critical Minerals Outlook 2025). Because the holdings are liquid U.S. large caps, it is straightforward to model, though suitability always depends on a client's goals, risk tolerance, and existing exposures.

Is HALO a good long-term buy for a DIY investor?

For a patient, long-horizon investor it offers tangible, hard-to-replace businesses such as mines, railroads, billboards, and industrial distribution rather than fast-moving tech. The demand backdrop is durable, for instance a single electric car uses up to four times the copper of a gasoline one (IEA, Global EV Outlook 2025). These are cyclical and can fall hard in a downturn, so it suits a hold-for-years approach rather than a quick trade.

How does HALO fit an institutional or thematic mandate?

HALO works as a real-asset or thematic-value sleeve inside a multi-asset mandate, because it packages an AI-resilience tilt with value and real-asset exposure from liquid, U.S.-listed large caps that are easy to model. The differentiator versus a generic value or industrials index is the structural, supply-led demand under the holdings: the IEA expects announced copper projects to fall roughly 30% short of demand by 2035 (IEA, Global Critical Minerals Outlook 2025), and AI itself is now a copper sink, with data-center demand for the metal projected to climb sharply through 2030 (S&P Global). The trade-off an allocator weighs is commodity beta and cyclicality, which drive tracking dispersion versus broad benchmarks, and an AI-immunity thesis that is hard to backtest cleanly.

Is HALO suitable for a conservative or income-focused investor?

Only partly. HALO is built for AI-resilience and real-asset exposure, not low volatility or yield, and its holdings are cyclical and capital-heavy, so they can fall as hard as anything in a downturn and are a poor fit for capital-preservation goals. There is some income inside the basket, since names like Lamar Advertising are REITs that own physical land, with more than 362,000 billboards under their control (Lamar 10-K), but the concept is a long-horizon growth-and-resilience tilt rather than a defensive income holding. A conservative investor should treat it as a small satellite, sized to risk tolerance, not a bond-like core.

Sources & references

  1. Akros U.S. Heavy Assets Low Obsolescence (HALO) Index · Methodology (EN) · Akros Technologies, Inc., 2026-04-09
  2. Akros Thematic Index Methodology Framework (EN) · Akros Technologies, Inc., 2025-11-01
  3. Lamar Advertising Company · 2025 Annual Report (Form 10-K) · Lamar Advertising Company, 2026-02-20
  4. American Trucking Trends 2025 (share of U.S. freight tonnage by mode) · American Trucking Associations, 2025-06-01
  5. Global EV Outlook 2025 (copper intensity of electric vehicles) · International Energy Agency, 2025-05-01
  6. Roundhill HALO ETF (LOHA) · fund objective · Roundhill Investments, 2026-05-14
  7. GPTs are GPTs: Labor market impact potential of LLMs · Eloundou, Manning, Mishkin & Rock · Science, 2024-06-21
  8. Global Critical Minerals Outlook 2025 (copper supply gap) · International Energy Agency, 2025-05-21
  9. Copper in the Age of AI · S&P Global, 2025-09-01
  10. Copper price forecast (long-run demand vs supply) · Goldman Sachs Research, 2026-01-15