H2 2026: I’m Staying Long AI Hardware – And I Think the Market Is Still Underestimating the Next Leg

Everyone is asking the same question heading into the second half of 2026:

“Has the AI trade peaked?”

After watching memory stocks deliver massive gains, Nvidia raising billions with ease, and several AI names suffering brutal corrections only to recover days later, it’s understandable why investors are nervous.

My view is different.

I think we are still in the middle innings of the AI infrastructure buildout, not the end of it.

Why I’m Staying Bullish

The market is treating AI as if it were a normal technology cycle.

It isn’t.

This feels much closer to previous mega-infrastructure booms:

* The internet buildout in the late 1990s.

* The smartphone ecosystem in the 2010s.

* Cloud computing over the last decade.

In every case, the market repeatedly thought demand had peaked, only to discover that the next wave of adoption was even larger.

AI may be following the same path.

⸻

The Biggest Misconception: “Everyone Has Already Bought AI”

No, they haven’t.

Most companies are still experimenting with AI rather than deploying it at scale.

Most governments are only beginning to invest in sovereign AI.

Most enterprises are still building their infrastructure.

Data centres globally are still racing to secure:

✅ GPUs

✅ High-bandwidth memory

✅ Networking equipment

✅ Storage

✅ Power capacity

This isn’t a one-quarter story.

This could be a multi-year capital expenditure cycle.

⸻

Why Memory Could Continue to Surprise

A year ago, few investors cared about memory companies.

Today, everyone suddenly understands that AI servers require significantly more DRAM and HBM than traditional servers.

But here’s the interesting part:

Even after the massive rally, demand visibility remains exceptionally strong.

AI model sizes continue increasing.

Inference demand is exploding.

Every new generation of AI models requires more computing power and more memory.

This creates a powerful flywheel:

More AI users → More inference → More servers → More memory demand.

That’s why I believe the market may still be underestimating the earnings power of memory companies over the next few years.

⸻

The Volatility Is a Feature, Not a Bug

Many investors were shaken out in H1.

Stocks fell 15%.

Recovered 20%.

Then corrected again.

This volatility made the rally feel unhealthy.

I disagree.

This is exactly how secular bull markets behave.

The biggest winners in history were never straight lines upward.

They climbed a wall of worry.

The volatility transfers shares from impatient investors to long-term holders.

⸻

But Doesn’t Valuation Matter?

Absolutely.

Not every company with “AI” in its presentation deserves a premium valuation.

There will be casualties.

There will be companies that overbuild.

There will be businesses that fail to monetise AI.

But the leaders supplying the infrastructure may continue benefiting because demand still appears to be outpacing supply in several critical areas.

The market often overestimates short-term growth and underestimates long-term structural change.

That could be happening again.

⸻

My H2 Positioning

I am not rotating aggressively into value.

I am not abandoning AI because stocks have already doubled or tripled.

Instead, I’m doing something simpler:

Stay invested in high-quality AI infrastructure companies and use volatility as an opportunity rather than a reason to panic.

Because if AI truly becomes as transformative as the internet or smartphones, then 2026 may eventually be remembered not as the peak…

…but as the period when the market finally realised that the AI infrastructure supercycle was only beginning.

The trend remains the same: AI demand is growing faster than the world’s ability to supply the infrastructure needed to support it.

And until that changes, I remain structurally bullish on AI hardware into H2 2026. 🚀📈