To The Moon

Today was a macro-driven selloff, not the end of the AI trade.

The Dow fell 2.2%, the S&P 500 lost 1.5%, and the Nasdaq dropped 1.7%. Oil surged back toward $90, the Fed delivered a hawkish hold, and crowded semiconductor positions faced another round of forced selling.

But after the close, Microsoft and Meta confirmed something far more important: the AI investment cycle is not slowing down. If anything, it is still accelerating.

The Fed Held Rates, but the Message Was Hawkish

The Federal Reserve kept rates unchanged at 3.50%–3.75%, but the vote was 9–3, with three policymakers calling for an immediate 25-basis-point hike.

That was not a normal pause.

The Fed is still worried about inflation, while the market has started pricing in a real possibility of another hike. At the same time, AI capital expenditure and rising prices for memory, logic chips, electricity, and data-center infrastructure have entered the inflation discussion.

At first glance, that sounds bearish for AI stocks.

Stronger AI demand pushes infrastructure prices higher. Higher prices create inflation pressure. Inflation keeps rates elevated, and higher rates compress technology valuations.

However, I do not think this weakens the long-term AI thesis.

In fact, it shows just how large the AI investment cycle has become. AI demand is no longer a small technology trend. It is now strong enough to affect chip prices, power demand, corporate capital spending, and even the Fed’s inflation debate.

Higher rates can temporarily pressure stock prices, but they do not automatically stop Microsoft, Meta, Google, and Amazon from building data centers. These companies have enormous cash flow and are not relying on cheap financing to continue the AI arms race.

Trump’s Comments Turned Oil Into the Final Trigger

The market was already fragile before the Fed decision.

After renewed attacks in the Middle East and Trump’s warning that the United States would retaliate strongly against Iran, Brent crude jumped roughly 7%–8% and returned to around $90.

The timing could not have been worse.

The market immediately began trading the following chain:

Higher oil prices → higher inflation → fewer rate cuts or another rate hike → higher Treasury yields → lower technology valuations.

That is why today’s decline spread so quickly.

$Micron Technology(MU)$ fell around 9%, $SanDisk Corp.(SNDK)$ dropped approximately 7.6%, SOXX lost roughly 4.5%, and $Vertiv Holdings LLC(VRT)$plunged more than 17%.

But $Seagate Technology PLC(STX)$still managed to rise after delivering strong earnings.

I think that detail matters.

If investors truly believed AI and data-center demand were collapsing, Seagate would not have remained strong while the rest of the sector sold off. The market was not abandoning AI. It was reducing leverage, taking profits, and punishing companies that failed to clear extremely high expectations.

The fundamentals did not collapse today. Positioning did.

Meta’s Selloff Does Not Mean Its AI Strategy Failed

$Meta Platforms, Inc.(META)$reported second-quarter revenue of $60.8 billion, up 28% year over year. Ad impressions increased 14%, while the average price per ad rose 12%.

The core business is still extremely strong.

The negative side was mainly profit and spending. Total costs increased 55%, net income declined 14%, quarterly capital expenditure reached $31.1 billion, and free cash flow fell to only $784 million.

Meta also raised the lower end of its full-year capital-expenditure guidance from $125 billion to $130 billion, while keeping the upper end at $145 billion. The stock fell around 4%–5% after hours as investors focused on lower profits, rising costs, and weaker free cash flow.

I understand that reaction, but I do not see it as evidence that the AI cycle is ending.

Meta is not increasing spending because AI demand has disappointed. It is spending more because AI is improving advertising, recommendation systems, engagement, model development, and its long-term computing strategy.

The company’s advertising engine is already benefiting from AI. The problem is simply that infrastructure spending is arriving faster than the profits generated by that infrastructure.

That creates short-term pressure on the stock, but it remains long-term demand for GPUs, networking equipment, servers, power systems, and storage.

For the AI supply chain, Meta raising its capital-expenditure floor is not bearish. It is another confirmation that hyperscalers are still fighting for compute capacity.

Microsoft Delivered the Strongest AI Signal

$Microsoft(MSFT)$ gave the market a much clearer answer.

Quarterly revenue reached $90 billion, up 18% year over year, while Azure grew 43%, comfortably beating expectations of roughly 40%. Microsoft’s contracted cloud backlog also increased to $678 billion, and Microsoft 365 Copilot now has more than 30 million paid users.

Meanwhile, quarterly capital expenditure reached approximately $41 billion, up more than 70% from a year earlier.

The important point is that Microsoft is spending more while Azure growth is accelerating.

This is exactly what the AI bull case needed to see.

If AI demand were mostly hype, Azure would not be growing 43% while Microsoft continued to face computing-capacity constraints. Customers would not be signing hundreds of billions of dollars in future cloud contracts, and Copilot adoption would not continue rising.

Microsoft’s results show that AI capital expenditure is not disappearing into a black hole. It is already producing cloud revenue, software subscriptions, and a larger backlog.

That is why Microsoft rose after hours while Meta fell.

The difference is not that Microsoft believes in AI and Meta does not. Both companies are spending aggressively. Microsoft is simply further ahead in turning that spending into visible revenue.

Why I Remain Bullish on AI and Storage

Today’s selloff has not changed my view.

The Fed can pressure valuations. Trump’s comments can push oil higher. Treasury yields can trigger another round of deleveraging. None of those things prove that AI demand is weakening.

Microsoft’s Azure growth, its expanding backlog, Meta’s higher capital-expenditure floor, and continued computing shortages all point in the opposite direction.

The AI investment cycle is still expanding faster than the existing infrastructure can support.

That remains bullish for GPUs, networking, data centers, power equipment, HDDs, DRAM, HBM, and enterprise storage.

Storage stocks may remain volatile because expectations became too aggressive. SK Hynix already showed that record earnings are sometimes not enough when the market expects perfection. One weaker number can trigger a violent correction even when the industry is still growing.

But there is a major difference between falling expectations and falling demand.

Right now, expectations are being reset. Demand is not collapsing.

I therefore see this correction as a painful shakeout inside a long-term AI bull market, not the beginning of the end. The market may become more selective, but that is normal after such a crowded rally.

The strongest AI companies will continue spending because they cannot afford to fall behind. Every hyperscaler understands that insufficient computing capacity today could mean losing the next generation of cloud, advertising, search, and software revenue.

Today the Fed hit valuations, oil hit sentiment, and leverage hit the semiconductor sector.

But Microsoft and Meta delivered the message that matters most:

The AI arms race is still getting bigger.