By Maksym Misichenko · Yahoo Finance ·
What AI agents think about this news
The panel's discussion on HALO stocks reveals a nuanced view, with concerns about capital intensity, regulatory risks, and potential overvaluation outweighing the defensive appeal of these 'AI-immune' plays. While some panelists see potential upside from asset optimization and operational maturity, the consensus leans bearish due to the significant risks and crowded trade dynamics.
Risk: The single biggest risk flagged is the potential for heavy debt loads and regulatory hurdles to erode free cash flow, turning these supposed safe havens into yield traps that underperform during market volatility.
Opportunity: The single biggest opportunity flagged is the potential for AI-driven asset optimization to lift utilization and ROIC in HALO firms, not just pressure them.
This analysis is generated by the StockScreener pipeline — four leading LLMs (Claude, GPT, Gemini, Grok) receive identical prompts with built-in anti-hallucination guards. Read methodology →
Concerned about an AI bubble? Sign up for The Daily Upside for smart and actionable market news, built for investors.
In Renaissance art, halos are indicators of divinity and holiness, or at least spiritual enlightenment.
In financial markets, they have a more earthbound association. HALO, an acronym for heavy assets, low obsolescence, refers to a trade that confers a degree of immunity. Immunity from the potentially corrosive effects of AI on some assets, that is.
As the artificial intelligence bull run that’s lifted markets threatens formerly high-growth sectors like software with painful disruption, traders are seeking investments that can arguably sidestep or withstand AI and, in some cases, even benefit from it.
“HALO stocks are immune to Claude Code,” Josh Brown, the Ritholtz Wealth Management CEO, explained when he coined the term back in February. “Anytime people are freaking out about Claude Code and ripping capital out of the shares of its perceived gallery of victims, these are the types of stocks that are being bought instead.”
Apart from their perceived resilience to AI disruption, HALO companies often have little in common with one another, making them an unusual category, he added.
Heavy assets generally mean capital intensive businesses that need money up front, that use or manufacture machinery and that require physical property. For example, Enterprise Products Partners, up 19% this year, operates 50,000 miles of pipelines. Water utility H2O America, up 18%, operates 407,000 water and wastewater service connections.
Low obsolescence means something unlikely to be replaced by AI any time soon, like electric grids, air conditioners or Coca-Cola.
The opposite of this, what’s not HALO, are companies that are capital light, having minimal physical inventory or physical assets. Capital light firms, especially in technology and software, thrived in the era of zero interest rates and widespread liquidity in the decade following the Great Recession because they could raise money on the promise of rapid growth with little startup or maintenance costs.
Sign up for The Daily Upside at no cost for premium analysis on all your favorite stocks.
READ ALSO: Move Over, SpaceX: Popular Companies from Lime to Oura Line Up for Market Debuts and World Cup Offers Record Windfall for FIFA, Dubious Payoffs for Host Cities
The Costs of Going Light
“Technology companies and other capital light industries, which also took advantage of the expansion in the digital economy and smartphone usage, commanded persistent valuation premia,” wrote Goldman Sachs analysts. “By the early 2020s, this regime had pushed the MSCI Growth Index to more than twice the valuation of the MSCI Value Index.”
But recent years, even before the term HALO was coined, served as a reminder of just how valuable physical assets are. The post-pandemic inflation shock, oil market disruptions related to conflicts in Ukraine and Iran, and the Trump administration’s aggressive tariff regime pushed up the cost of capital, highlighting the importance of access to energy, supply chains and infrastructure.
Meanwhile, AI’s promise has added pressure on capital light businesses in technology and software. That’s because advanced agentic AI is poised to automate more workflows and coding, potentially eroding revenue models across entire industries such as software as a service (SaaS).
These factors have created conditions more favorable for capital intensive investments.
“The [return on equity] gap between capital light and capital intensive remains wide, but consensus forecasts now expect flat ROE for capital light companies, while capital intensive ROE is projected to improve,” Goldman’s analysts wrote.
Ritholtz’s Brown, in an effort to identify prominent HALO stocks, highlighted energy giant ExxonMobil (barrels of crude are “not going to end up in the crosshairs of DeepSeek”), retailer Walmart (“they have the best logistics network on earth, parking lots, stores, shelves, warehouses, inventory”), McDonald’s (you can’t code a Big Mac), and building materials producer Martin Marietta (“They make concrete. ChatGPT doesn’t and won’t.”).
There are also Hershey, Delta, NextEra Energy, Ventas and Caterpillar. All are HALO, one can argue, because even if an unprecedented AI model came out and proved the Riemann Hypothesis tomorrow, it wouldn’t replace Reese’s Peanut Butter Cups, Boeing 787 Dreamliners, the US power grid, retirement homes or bulldozers.
The most HALO-friendly sectors, Goldman said, include utilities, energy, telecom, airlines, and automobiles and auto parts.
Another beneficiary is real estate, according to analysts at Hazelview Investments, who noted real estate investment trusts outperformed the broader market in the first quarter.
“I’ve been in this business for over 25 years and, for many years, warehouses were just storage space; you’d go into a facility and see boxes,” says Rick Schaupp, a managing director at Clarion Partners, which oversees a $72.6 billion real estate portfolio spanning industrial, logistics, retail and apartments.
“Now, they’re the critical backbone behind the e-commerce supply chain and even supporting data center infrastructure and construction. Their strength lies in the fact that they are flexible, that layouts and design can change over time and reflect the different demand uses that we’re seeing and be part of the global supply chain network.”
Best Defense Is a Good Offense
While the HALO trade attempts to address the impact of AI on markets, it doesn’t quite fit the classic definition of defensive stockpicking, according to Roundhill Investments CEO Dave Mazza.
“Traditional defensives are often about stable demand and low earnings variability,” he wrote in March. “HALO is about something slightly different. It is about whether the core economics of the business are likely to be structurally disrupted by AI, or whether the company owns assets and networks that remain essential regardless of the next technology wave.”
That means HALO stocks don’t necessarily have to have classically defensive characteristics like steady earnings or dividends.
Mazza pointed to freight rail, which is cyclical due to its exposure to business cycles but is also virtually impossible to replace, as a sector with the requisite HALO qualities but not a traditional defensive profile.
Another quality that sets the HALO trade apart from traditional defensive positioning is that there’s an offensive element, too. Since the development and scaling of AI will require massive real-world assets and investments, companies that specialize in making and building them stand to gain.
There have been plenty of investor rotations into companies that stand to benefit. Shares in Vertiv, which makes cooling systems used by AI data centers, are up 86% this year. Comfort Systems, which specializes in heating, ventilation, and air conditioning installation and repair for server farms, is up 99%.
“The result is a combination of downside resilience and targeted offensive opportunity,” Peter Vanderlee, a portfolio manager at ClearBridge Investments, wrote of HALO stocks that “offer meaningful participation in structural growth themes.”
With AI hyperscalers set to deploy some $765 billion in capital spending this year and $1.6 trillion by 2031, according to Goldman estimates, that’s a lot of offensive opportunity for companies that specialize in real-world infrastructure to capture.
It’s also why memory and data storage giants like Micron Technology, Samsung Electronics and SK Hynix are considered HALO candidates. Or chipmaker Taiwan Semiconductor Manufacturing Corp. and photolithography machine manufacturer ASML.
All of them have sophisticated, advanced infrastructure that is unlikely to be replicated, and their products are crucial to scaling AI.
Old School Assets, Old School Problems
The virtue of HALO companies also explains their potential downsides. Heavy assets mean operations that require significant, ongoing capital expenditures and the maintenance and upgrading of infrastructure.
Real-world assets mean real-world problems: Earnings can take a hit from higher construction costs resulting from inflation, disruptions to supply chains, labor shortages and time-consuming regulatory approvals that hold back projects.
Many companies also rely on debt to fund and sustain their real-world footprint, which can make them sensitive to interest rates and fluctuations in credit markets.
Ultimately, that means the HALO trade will require thinking beyond sectors. Instead, investors will have to identify the companies that are the best at executing plans.
This post first appeared on The Daily Upside. To receive razor sharp analysis and perspective on all things finance, economics, and markets, subscribe to our free The Daily Upside newsletter.
Four leading AI models discuss this article
"HALO immunity to AI is not guaranteed; macro, financing, and maintenance costs can drive underperformance even for asset-heavy names."
HALO stocks are pitched as immune to AI disruption due to heavy assets and essential networks, but the safety net may be thinner than it looks. Capital-intensive models carry persistent capex, debt service and maintenance costs that amplify sensitivity to higher rates and inflation. Real-world assets can become stranded if demand shifts or if regulatory hurdles slow new projects; and a recession can crush ROEs in cyclicals like autos, airlines, and construction materials even if assets remain physically intact. The article glosses over dispersion within HALO—some names will fail the structural test, others will not. Valuations may already bake in resilience that doesn’t survive a macro shock.
The strongest counterargument is that HALO's fortress may simply reflect a risk premium: in a prolonged high-rate or recessionary environment, demand for heavy assets can crumble, debt-heavy halos amplify losses, and AI-driven productivity gains may still erode the fundamental need for certain assets.
"The HALO trade is a necessary hedge against software commoditization, but it risks becoming a value trap if investors ignore the interest rate sensitivity inherent in capital-intensive balance sheets."
The HALO thesis is essentially a pivot toward 'moat-by-physics.' By prioritizing companies like Caterpillar or NextEra Energy, investors are hedging against the deflationary pressure AI exerts on software margins. However, the market is conflating 'defensive' with 'capital-intensive.' While physical assets are harder to disrupt, they are also highly sensitive to the cost of debt and regulatory friction. We are seeing a valuation rotation where the 'AI-immune' trade is becoming crowded. If interest rates remain 'higher for longer,' the heavy debt loads required to maintain these assets will erode free cash flow, turning these supposed safe havens into yield traps that underperform during the very volatility they are meant to mitigate.
The strongest case against this is that AI-driven productivity gains will eventually lower the operational costs for these capital-intensive firms, potentially leading to a massive margin expansion that the market is currently ignoring.
"The HALO trade is a valid thesis about structural shifts in valuations, but most HALO candidates have already priced in mean reversion, and the remaining upside is modest relative to execution risk and sector-specific headwinds."
The HALO framework is intellectually coherent but dangerously backward-looking. Yes, capital-intensive assets have underperformed for 15 years and valuations have compressed—but the article conflates 'AI won't replace a bulldozer' with 'bulldozers are a good investment now.' The real risk: HALO stocks are priced for mean reversion that may already be baked in. Utilities, energy, and industrials have already rallied hard. More critically, the article ignores that many 'HALO' plays (ExxonMobil, Walmart, Delta) face structural headwinds—energy transition, labor costs, fuel prices—orthogonal to AI. The offensive case (Vertiv, ASML, Micron) is stronger but already reflected in 86-99% YTD gains. Chasing momentum into crowded trades disguised as defensive positioning is how you buy at peaks.
If AI truly automates white-collar work and erodes SaaS economics, capital-intensive businesses with pricing power and durable moats (utilities, energy infrastructure) may genuinely re-rate higher for a decade, making current valuations cheap on a 10-year view.
"HALO’s capital-intensity premium is overstated once ongoing capex and financing risks are properly priced."
The HALO rotation into names like ExxonMobil, NextEra Energy, Vertiv and Martin Marietta frames heavy assets as AI-proof, yet the same capital intensity that shields them from software disruption exposes them to sustained high capex, supply-chain inflation and rate sensitivity. Goldman’s projected ROE improvement for capital-intensive firms hinges on stable financing costs that could reverse if tariffs or credit spreads widen. Meanwhile, the $765 billion AI-related spend cited may favor only the highest-execution infrastructure players, leaving broader HALO laggards vulnerable to earnings compression.
Massive hyperscaler capex could overwhelm rate and cost headwinds, delivering structural revenue growth to precisely the asset owners the article highlights.
"AI-enabled asset optimization can lift ROIC and sustain HALO re-rating even if sentiment stays weak."
Responding to Claude: I think the 'mean reversion' critique misses that AI-driven asset optimization can lift utilization and ROIC in HALO firms, not just pressure them. If the ongoing capex cycle in utilities, energy infrastructure, and industrials remains durable and financing costs stabilize, HALO names could re-rate on operating leverage even if general sentiment darkens. The risk isn't only 'headwinds' but a potential upside surprise in cash flow quality.
"Legacy industrial firms lack the software maturity to convert AI integration into actual margin expansion, making current valuation premiums unsustainable."
Claude is right about the crowded trade, but ChatGPT’s 'asset optimization' thesis ignores the execution gap. Most legacy industrials lack the digital maturity to integrate AI-driven efficiency gains without ballooning IT overhead. We aren't seeing margin expansion; we are seeing a 'capex-to-nowhere' risk. If these firms can't prove ROIC on their digital transformation, the valuation premium evaporates. The real danger is the market pricing in AI-driven productivity that these firms lack the software DNA to actually capture.
"Not all HALO capex is speculative; some is regulatory-driven and decoupled from execution risk on AI-driven productivity."
Gemini's 'capex-to-nowhere' critique is sharp, but it conflates two different problems. Legacy industrials *do* have execution risk on digital transformation—true. But utilities and energy infrastructure are already operationally mature; their capex is mandated by grid modernization and decarbonization, not discretionary AI bets. NextEra's capex ROI doesn't depend on AI adoption; it depends on regulatory returns and demand. That's a different risk profile than, say, CAT trying to retrofit autonomous dozers.
"Regulatory allowed returns for utilities face compression from sustained high rates, undermining HALO resilience claims."
Claude distinguishes utilities' mandated capex from industrials' discretionary bets, yet overlooks that allowed ROEs for NextEra and peers are set against prevailing interest rates. If the Fed holds rates higher, those regulated returns get squeezed even as debt costs rise, eroding the very cash-flow stability the HALO thesis assumes. This regulatory lag creates a hidden duration mismatch that neither Gemini's execution concerns nor ChatGPT's optimization upside fully capture.
The panel's discussion on HALO stocks reveals a nuanced view, with concerns about capital intensity, regulatory risks, and potential overvaluation outweighing the defensive appeal of these 'AI-immune' plays. While some panelists see potential upside from asset optimization and operational maturity, the consensus leans bearish due to the significant risks and crowded trade dynamics.
The single biggest opportunity flagged is the potential for AI-driven asset optimization to lift utilization and ROIC in HALO firms, not just pressure them.
The single biggest risk flagged is the potential for heavy debt loads and regulatory hurdles to erode free cash flow, turning these supposed safe havens into yield traps that underperform during market volatility.