The trillion-dollar market capitalizations achieved by Samsung, SK Hynix, and Micron Technology (NASDAQ: MU) are fueled by the 'HALO' investment logic—betting on Heavy Assets and Low Obsolescence—amid the AI frenzy.

Yet, a fierce debate rages over whether this surge marks the start of a supercycle or a pre-bubble peak, centering on valuation metrics and the potential of long-term agreements to defy historical cyclicality.

The core question remains: can these memory giants retain their trillion-dollar valuations when speculative capital inevitably recedes?

The HALO strategy, targeting companies with significant capital investments and durable product relevance, has become the year's hottest trade.

Investors globally are pouring money into manufacturers across sectors from automotive to PCs, betting they will be prime beneficiaries of trillion-dollar AI infrastructure spending.

However, concerns are mounting that this rally may be more speculative mania than sustainable trend, with the parabolic rise of the Philadelphia Semiconductor Index drawing comparisons to the dot-com bubble era.

Evidence suggests trend-following strategies have amplified gains for the biggest AI winners, raising questions if the "picks and shovels" narrative is merely a story investors tell to justify risky bets.

This tension is most evident in the three dominant DRAM manufacturers: Samsung Electronics, SK Hynix, and Micron Technology.

Retail momentum chasing has undoubtedly been a key driver propelling all three into the trillion-dollar club this year, but these hardware firms are also beginning to generate impressive profits.

The critical valuation debate hinges on whether to price these chipmakers on a price-to-book (P/B) or price-to-earnings (P/E) basis.

Following historic rallies, their stocks appear expensive measured against net asset value, yet seem cheap relative to optimistic analyst earnings forecasts.

The memory chip industry is notoriously cyclical.

During boom periods when profits soar, P/E ratios can appear deceptively low, but the metric becomes meaningless or negative during downturns when net income turns to loss.

Traditionally, investors have favored P/B ratios as a more reliable gauge.

Proponents of the HALO trade, however, argue for a paradigm shift, championing "LTA" or Long-Term Agreements as a game-changer.

Unlike past cycles driven by PC or smartphone demand, AI hyperscaler customers are reportedly willing to sign multi-year, irrevocable contracts to secure chip supply for massive infrastructure builds, some with substantial upfront payments.

UBS estimates 20-30% of current DRAM output may be covered by such agreements, leading some analysts to claim unprecedented earnings visibility through the end of the decade.

UBS analyst Timothy Arcuri projects Micron could sustain EPS above $100 from 2027-2029, generating over $400 billion in free cash flow during that period—a stark contrast to the $16 billion generated in the past decade.

He recently raised Micron's price target to $1,625, implying 67% upside from recent levels.

By the same logic, the historic surge in Return on Equity (ROE) for Samsung and SK Hynix could also prove sustainable, challenging Korea's historical reputation for low capital productivity and its market's traditional P/B valuation lens.

Skepticism persists, however.

First, details of LTAs are scarce due to non-disclosure agreements, leaving analysts with rough estimates and anecdotal evidence.

Second, the memory industry's history of severe boom-bust cycles makes it difficult for veteran investors to envision a decade-long, AI-driven supercycle.

For instance, SK Hynix's ROE peaked at 19% in mid-2022 during the pandemic-driven PC/server boom, only to crash to -18% a year later as the economy reopened.

Furthermore, the current optimism is fundamentally tied to hyperscaler spending plans.

Major US tech firms recently raised their 2024 AI capital expenditure guidance to as high as $725 billion, a 90% increase from 2025 plans.

A pullback in this spending would unravel the bullish thesis.

We are only in the second year of this AI-driven expansion.

While the scale of investment is extraordinary, it is likely too early to declare a definitive break from historical cyclical patterns or a final verdict on the enduring power of the HALO strategy.