To The Moon

Everyone knows what $100 oil does to airlines.

Everyone knows what it does at the petrol pump.

Everyone knows what it can do to inflation.

But I think Wall Street may be overlooking a much stranger potential casualty.

Artificial intelligence.

Not because data centres run on crude oil.

They don’t.

Because the AI boom increasingly runs on something else:

CAPITAL.

And the price of that capital is moving.

⸻

🛢️ THE OIL SHOCK DOESN’T HAVE TO TOUCH A DATA CENTRE TO HIT IT

The first-order trade is obvious.

Oil rises.

Energy companies benefit.

Transport costs rise.

Consumers feel it.

Inflation becomes harder to kill.

But follow the chain another few steps:

OIL ↑

⬇️

INFLATION PRESSURE ↑

⬇️

BOND YIELDS / RATE EXPECTATIONS ↑

⬇️

COST OF CAPITAL ↑

⬇️

AI INFRASTRUCTURE FINANCING GETS MORE EXPENSIVE

⬇️

THE RETURN REQUIRED FROM AI INVESTMENT RISES

That’s where this gets interesting.

Brent has been trading above $100 amid the current energy shock, while the U.S. 10-year Treasury yield recently breached 5%, its highest level since 2007. Reuters reported the sell-off alongside intensifying inflation concerns and expectations for tighter monetary policy.

The AI boom has never had to prove itself at this scale against this financing backdrop.

⸻

🤖 AI NOW HAS A HURDLE RATE

For the last few years, Wall Street’s AI questions were mostly technological.

Who has the GPUs?

Then:

Who has the HBM?

Then:

Who has the power?

Then:

Who has the data centres?

But the buildout has become so large that another constraint is emerging.

Who can finance all of it, and at what return?

S&P Global Ratings projects combined hyperscaler capital expenditure to exceed $1.3 TRILLION by 2027.

More importantly, it expects the six hyperscalers in its analysis to generate negative free operating cash flow in both 2026 and 2027, with recovery not projected until 2029.

S&P says debt, leases, guarantees, equity issuance, joint ventures and other financing arrangements are playing increasingly important roles in funding the AI infrastructure expansion.

This isn’t just a semiconductor story anymore.

It’s becoming a capital-markets story.

⸻

💰 WALL STREET IS ALREADY FUNDING THE MACHINE

Alphabet.

Amazon.

Meta.

Microsoft.

Oracle.

Together, those companies have issued approximately $220 BILLION of bonds over the past year as the data-centre and AI infrastructure expansion accelerates.

And the financing requirement extends far beyond hyperscalers.

Vantage Data Centers is reportedly seeking another $2 billion from institutional investors including Pimco and PGIM after borrowing roughly $48 billion since 2025 for hyperscale data-centre developments.

One reason?

Traditional lenders are approaching concentration limits on technology exposure.

Read that again.

The AI infrastructure boom isn’t merely consuming:

GPUs.

Memory.

Electricity.

Land.

Cooling.

Transformers.

Fibre.

It’s consuming balance-sheet capacity.

⸻

📈 NOW PUT A 5% TREASURY NEXT TO IT

This is where I think the $100-oil story becomes much bigger than oil.

A higher risk-free rate potentially squeezes the AI trade from two directions.

On one side:

Future AI earnings are discounted at a higher rate.

That puts pressure on valuations.

On the other:

The infrastructure required to generate those future earnings becomes more expensive to finance.

That puts pressure on project economics.

So the question changes.

It isn’t simply:

“Will AI generate revenue?”

We’re already seeing evidence that it can. S&P notes strengthening cloud expectations at Microsoft, Amazon and Google, even while heavy infrastructure spending pressures cash generation.

The harder question becomes:

WILL AI GENERATE ENOUGH RETURN TO BEAT ITS RISING COST OF CAPITAL?

That’s a very different hurdle.

⸻

⚠️ THIS DOES NOT MEAN THE AI BOOM IS DEAD

Quite the opposite.

The largest hyperscalers still possess enormous financial resources.

S&P estimated earlier this year that the five large cloud providers it rates could spend around $750 billion in 2026 alone, while noting that most currently retain balance-sheet capacity to manage the investment cycle.

AI demand remains powerful.

Cloud growth remains strong.

The infrastructure is still being built.

And there’s an important counterargument here.

Higher long-term yields aren’t necessarily only an AI negative.

JPMorgan Private Bank recently argued that part of the rise in bond yields may itself reflect expectations of stronger future productivity and economic growth generated by AI.

So this isn’t:

$100 oil → AI collapses.

That’s far too simplistic.

My thesis is different.

⸻

🔥 THE MARGIN FOR ERROR MAY BE SHRINKING

When capital is cheap, investors can tolerate long payback periods.

When capital becomes expensive, the question becomes:

Where are the returns?

That could eventually separate the AI ecosystem into two camps.

Companies capable of converting infrastructure spending into durable cash flow.

And companies whose economics only looked extraordinary when money was cheap.

That’s why I’m increasingly interested in ROIC, free cash flow, debt structure and utilisation, not simply GPU counts.

The next phase of the AI trade may reward the companies that can prove the economics behind the infrastructure.

Not merely build more of it.

⸻

🧠 THE SECOND-ORDER TRADE

Maybe Wall Street is still looking at the wrong bottleneck.

First it was:

CHIPS.

Then:

MEMORY.

Then:

POWER.

Then:

DATA CENTRES.

But if trillions of dollars must keep moving through the system to build the next generation of AI infrastructure…

perhaps the next scarce resource isn’t technological at all.

Perhaps it’s:

CHEAP CAPITAL.

And if that’s right, $100 oil isn’t merely an energy story.

It may be quietly raising the hurdle rate on the largest technology infrastructure buildout in history.

⸻

🐂 THE BULL CASE

AI monetisation accelerates fast enough to outrun higher financing costs.

Cloud revenues continue compounding.

Utilisation rises.

Productivity gains become visible.

Free cash flow recovers.

And today’s enormous capex ultimately looks justified.

🐻 THE BEAR CASE

Oil keeps inflation sticky.

Yields remain elevated.

Financing costs stay high.

AI capex continues outrunning cash generation.

Projects take longer to reach economic utilisation.

And investors begin demanding something much less exciting than another GPU announcement:

proof of return.

⸻

🎯 WHAT I’M WATCHING

Not whether AI survives.

I think that’s the wrong question.

I’m watching when the market changes the question from capacity to economics.

Because once that happens:

GPU COUNT → UTILISATION

CAPEX → ROIC

AI PROMISE → CASH FLOW

BUILD AT ANY COST → PROVE THE RETURN

And that could change who actually wins the next phase of AI.

⸻

🚨 FINAL THOUGHT

The market spent years asking whether the world had enough chips to build AI.

Then whether it had enough memory.

Then enough electricity.

Maybe the next question is:

DOES IT HAVE ENOUGH CHEAP MONEY?

Because if the AI buildout increasingly depends on capital markets while oil keeps inflation and yields elevated, the next AI bottleneck may not sit inside a server rack.

It may sit on the balance sheet.

And that leaves me with one question:

WHAT HAPPENS WHEN THE PRICE OF INTELLIGENCE MEETS THE PRICE OF MONEY?

$NVDA $GOOGL $AMZN $META $MSFT $ORCL

Personal market analysis only. Not financial advice. Always do your own research.

adz5150 🐯