To The Moon

The TACO Trap: Retail Bets on Politics, Capital Bets on Physics

On March 11, crude oil collapsed from nearly $120 to around $90 per barrel. The market quickly labeled the move with a familiar acronym: TACO — Trump Always Chickens Out.

Traders started pricing in geopolitical de-escalation. Energy risk premium was dumped aggressively. Capital rotated back into high-valuation tech names, particularly AI application companies and SaaS platforms.

This is a retail meat grinder.

Political events can create price volatility. They cannot generate free cash flow. Policy signals on social media do not raise return on capital. They do not manufacture transformers or expand power grids.

Institutional capital cares about two things only:

• cash-flow certainty

• physical barriers to entry

The market’s most dangerous mistake right now is confusing macro sentiment with industrial reality.

Oil can drop $30 in a day. AI data-center construction does not stop for a second.

Power infrastructure upgrades continue. Server procurement continues. Semiconductor fabs keep expanding.

Political trading changes screen prices. It does not change capital allocation.

Meanwhile, the AI application layer is swimming naked.

Many SaaS companies now trade at Forward P/E ratios approaching or exceeding 80x, while their FCF yields remain fragile. Their real competitor is not another startup.

It is the foundation model itself.

Every generation leap in large models compresses the economic value of application software. Algorithm cycles now move monthly. Product life-cycles shrink to a few quarters.

Using traditional software valuation frameworks to price AI applications in 2026 is simply underwriting a bubble.

The CAPEX Reality: A $600 Billion Infrastructure War

Hyperscale data centers are the physical backbone of the AI industrial cycle.

The AI revolution is not about elegant code.

It is about industrial capacity.

Based on guidance released in Q1 earnings outlooks for 2026, North American technology giants are expected to deploy more than $600 billion in total CAPEX for the full year.

This number is historic.

More importantly, the destination of that capital is crystal clear.

The money is not flowing into fragile software layers. It is flowing into heavy infrastructure:

• hyperscale data-center construction

• power grid expansion

• semiconductor manufacturing capacity

• large-scale storage infrastructure

AI competition has already moved beyond model benchmarks.

It is now a brutal industrial supply race.

In this capital reallocation cycle, the real winners belong to a category we define as HALO assets:

Heavy Asset, Low Obsolescence.

Algorithms can depreciate in months. Physical infrastructure collects tolls for decades.

When capital cycles align with asset lifetimes, profit becomes inevitable.

Sector #1: Power Equipment — The First Toll Booth of AI

The first bottleneck of AI infrastructure is not GPUs.

It is electricity.

A large AI data center can demand 300 megawatts of power, roughly equivalent to the consumption of a mid-sized city.

Compute hardware is only the endpoint. The power grid is the gatekeeper.

The global power-equipment industry is now experiencing an unprecedented supply squeeze.

By early 2026, delivery lead times for 400kV ultra-high-voltage transformers have stretched toward 48 months.

This means grid infrastructure has become a hard constraint on AI deployment.

Industry structure is highly concentrated. Major players include Eaton, ABB, and $Schneider Electric SE(SBGSF)$ .

Order backlogs across the sector have effectively locked in revenue visibility for years.

In capital-market terms, backlog equals pre-confirmed cash flow.

Sector #2: Semiconductor Equipment — The Industrial Machines of Compute

If electricity is the entrance to compute, semiconductor equipment is the industrial womb of compute.

This sector has even higher concentration than power infrastructure.

The critical global players include $ASML Holding NV(ASML)$ , Lam Research, and Tokyo Electron.

The most extreme example of HALO capital intensity is the next generation High-NA EUV lithography machine.

These systems are almost entirely monopolized by $ASML Holding NV(ASML)$ .

Each unit now costs close to $400 million, with delivery and calibration cycles exceeding one year.

This is the true capital-devouring monster of the semiconductor industry.

When AI demand explodes, equipment supply cannot expand quickly. Chip designers may suffer through inventory cycles.

Equipment suppliers do not.

They monetize multi-year CAPEX waves.

That is the aesthetic of HALO monopolies.

Sector #3: Storage Infrastructure — The New Energy Arbitrage

AI is transitioning from training phase to inference phase.

Training demands compute. Inference demands data access speed.

With the rapid adoption of Retrieval-Augmented Generation (RAG) architectures, enterprise storage has become a key layer of AI infrastructure.

The newest PCIe Gen5 enterprise SSDs now reach capacities of 122.88TB.

Large AI systems increasingly deploy dense storage clusters to cache vector databases and reduce repeated GPU computation.

This is essentially an energy arbitrage strategy:

Replace expensive compute cycles with storage retrieval.

Key industry participants include $Western Digital(WDC)$ , $KIOXIA HLDGS CORP(KXIAY)$ , and Solidigm.

As AI architectures evolve, storage spending is steadily claiming a larger share of hyperscale CAPEX.

The Super-cycle Transmission: Inflation Through the Physical Chain

Another macro card is the commodity supercycle.

When oil remains trapped in the $90–$120 range, energy costs begin cascading through the industrial system.

History shows a clear transmission path:

Energy → Chemicals → Agriculture.

In the later stages of the cycle, agricultural commodities often surge.

Fertilizer costs rise. Transportation costs rise. Energy input costs rise.

Soybeans and corn futures typically benefit from this pressure.

Upstream fertilizer producers also gain pricing power.

The global fertilizer market is highly concentrated, dominated by companies such as Nutrien and CF Industries.

For macro portfolios, agriculture and fertilizer sectors serve as both inflation hedges and commodity-cycle leverage within the broader AI industrialization theme.

Global Long / Short Playbook

The source of alpha in 2026 is not individual stocks.

It is industrial structure.

Short: AI Application and SaaS Sector

Short companies trading at extreme valuation multiples while relying heavily on model APIs. Many operate with forward P/E ratios approaching 80x and weak free-cash-flow durability.

When foundation models absorb more functionality, their economic moats will evaporate.

Algorithmic depreciation leaves them defenseless.

Long: HALO Infrastructure and Commodity Transmission

Go long sectors controlling the physical entry points of AI:

• power equipment

• semiconductor manufacturing equipment

• enterprise storage infrastructure

Complement the portfolio with fertilizer and agricultural commodities as inflation hedges through futures or ETFs.

When $600 billion of annual CAPEX is being poured into physical infrastructure, the AI revolution has already crossed the line from software narrative to industrial reconstruction.

Politics fades. Algorithms iterate.

But physical infrastructure endures.

In 2026, code is cheap.

Silicon, steel, electricity, and grain are expensive.