Why Big Tech makes more sense to me today than chasing the semiconductor boom

Something very interesting is happening in the market today.

Many investors are extremely bullish on semiconductors, memory stocks, GPU suppliers, AI networking and the whole semiconductor supply chain. Nvidia, AMD, Broadcom, Marvell, Micron, TSMC, ASML and other companies are clear AI winners for many people.

At the same time, many of those same people claim that Big Tech is “dead”, that Microsoft, Meta, Amazon or Google have nowhere left to grow, that their AI capex is a problem and that they are spending too much money.

And that doesn’t sit right with me.

Because if you’re extremely bullish on semiconductors, memory stocks and AI infrastructure, you should first ask yourself who is financing this entire boom.

And the answer is simple.

A large part of today’s AI infrastructure boom is being financed by the hyperscalers — Microsoft, Meta, Amazon, Google, Oracle, OpenAI and other big players.

In other words: many people today don’t want to own the customers who are paying for this boom, they want to own their suppliers.

And that, to me, is an important distinction.

One more thing: if you’re bullish on semis and memory, you should logically be bullish on Meta, Amazon, Microsoft and Google too. Because without the hyperscalers, that growth simply won’t happen.

Semis are not an independent bet. They are a bet on Big Tech capex

When someone buys Nvidia, Micron, Marvell, Broadcom or other AI infra companies, they aren’t just buying “AI winners.”

They are buying the view that Big Tech will continue to spend huge sums on AI infrastructure.

They are buying the view that these expenditures will continue.

And most importantly, they are buying the view that these expenditures will pay off for the hyperscalers.

Because if it turned out that AI capex has a weak return, what happens?

Hyperscalers will start to slow their spending.

When they slow down, demand for GPUs, memory, networking, servers, data centers and the whole infrastructure will fall.

And then the ones who might feel the biggest pain wouldn’t necessarily be Big Tech itself, but their suppliers.

Because Microsoft, Meta, Amazon or Google still have huge existing businesses, customers, distribution, advertising, cloud, software, ecosystems and cashflow.

But many semiconductor and memory companies today have a large portion of their growth tied directly to the AI infrastructure boom.

Just look at the numbers

In its latest Q1 FY2027 report Nvidia reported record revenue of $81.6 billion, of which Data Center revenue was $75.2 billion. That means Nvidia today is, to a large extent, a data center / AI infrastructure company, not just the original “gaming GPU” company. (NVIDIA Investor)

AMD reported total revenue of $10.3 billion in Q1 2026, with the Data Center segment at $5.8 billion and growing 57% year-over-year. Lisa Su herself said that Data Center is now the primary driver of AMD’s revenue and profits. (Advanced Micro Devices, Inc.)

Marvell had as much as 74% of revenue from the data center segment in fiscal 2026. At the same time, its top 10 customers accounted for 82% of total revenue. That is a huge customer concentration and clear evidence that the business is tightly tied to large infrastructure customers. (Marvell Technology, Inc.)

Broadcom said in Q2 FY2026 that its AI semiconductor revenue was $10.8 billion, up 143% year-over-year. The company also expected AI semiconductor revenue to grow to $16 billion in the next quarter. (Broadcom Inc.)

Micron is another example. The memory business looks fantastic today precisely because AI infrastructure needs massive amounts of memory. Reuters recently wrote that Micron is benefiting from strong demand for AI-related memory chips, including HBM for Nvidia AI processors, and that the company announced customer commitments worth $22 billion. (Reuters)

These numbers are strong. I don’t dispute that.

But they also show one thing:

If you’re bullish on these companies, you are to a large extent bullish on the continuation of Big Tech AI capex.

That’s why it doesn’t make sense to be bearish on Big Tech and bullish on semis

This is the main point.

If someone says Microsoft, Meta, Amazon or Google are “dead money,” but at the same time is wildly buying Nvidia, Micron, Marvell or Broadcom, I think they should rethink their stance.

Because where does a large part of the growth for these semiconductor companies come from?

From the hyperscalers.

From firms that are spending hundreds of billions of dollars on AI infrastructure.

For example, Meta reported $56.3 billion in revenue in Q1 2026, +33% year-over-year, operating margin 41%, cash flow from operations $32.2 billion, and expects capex for 2026 in the range of $125–145 billion. (investor.atmeta.com)

Microsoft reported Microsoft Cloud revenue of $54.5 billion in Q3 FY2026, +29%, Azure and other cloud services grew 40%, and commercial remaining performance obligation increased 99% to $627 billion. (Microsoft)

Amazon reported AWS revenue of $37.6 billion in Q1 2026, +28%, and AWS operating income of $14.2 billion. Free cash flow fell mainly due to higher investments in property and equipment, which the company attributes largely to AI investments. (ir.aboutamazon.com)

So on one hand the market says:

“Big Tech is spending too much, AI capex is a problem, returns are uncertain.”

But on the other hand it says:

“The suppliers of that capex are great investments.”

And that strikes me as an odd mismatch.

We can’t pretend the semiconductor boom exists in a vacuum.

It doesn’t.

It’s tied to the hyperscalers’ budgets.

My view: I’d rather own the hyperscalers themselves

To be clear, I am strongly invested in Big Tech today.

To me, companies like Meta, Microsoft and Amazon are very attractive buys at today’s prices and valuations.

I see significant upside potential precisely because sentiment, in my view, is too negative.

The market today treats these companies as if their best years are behind them. As if AI capex is purely a problem. As if it’s throwing money away with no return.

I see it differently.

Meta has a huge advertising business, billions of users, strong cashflow, algorithms, engagement, Reels, WhatsApp, Instagram, an AI assistant and, in my view, a huge opportunity in AI glasses and a personal AI assistant.

Microsoft has Azure, Office, Copilot, GitHub, enterprise distribution, cloud, security and one of the best positions in B2B software in the world.

Amazon has AWS, retail, logistics, advertising, proprietary chips, a massive customer base and multiple businesses that can benefit from AI.

These companies don’t have only one way to make money from AI.

They have multiple monetization paths.

They can improve advertising.

They can increase efficiency.

They can sell AI services to customers.

They can improve the cloud.

They can automate internal processes.

They can create new products.

And most importantly — they already have customers and distribution today.

Semis have more leverage, but also more risk

I’m not saying semiconductor companies are bad investments.

Not at all.

Nvidia is an extremely high-quality company. Broadcom has a fantastic business. AMD has an interesting position as the second big player in AI accelerators. TSMC is one of the most important companies in the world. ASML has an almost irreplaceable position in lithography. Micron can benefit from a huge memory cycle.

But one thing must be acknowledged.

Semis today are often a more leveraged bet on the AI infra boom.

If the boom continues, their numbers can look fantastic.

When customers increase orders, revenues grow, margins expand, guidance rises and stocks soar.

But if a slowdown comes, the effect can be very harsh.

And especially for memory companies, the cyclicality is brutal. When there’s a shortage of memory, prices rise and profits explode. When there’s an oversupply, prices fall and sentiment can reverse very quickly.

So I don’t think it’s enough to look at current results.

An investor must ask:

Where do these results come from?

Who is the customer?

Is the demand sustainable?

Will Big Tech continue to spend?

And most importantly — will this capex generate sufficient returns?

The whole question rests on the return of AI capex

This, to me, is the most important question for the coming years.

Not whether AI will be important.

I believe it will be.

The question is:

Who will profit from it?

How much will they profit?

And will the massive hundreds of billions of dollars invested in infrastructure deliver sufficient returns?

If yes, Big Tech could be undervalued today and semis could continue to rally.

If not, there will be a re-evaluation of the entire AI chain.

But in that scenario I don’t think it’s logical to believe Big Tech will suffer while semis are fine.

Quite the opposite.

Big Tech has multiple sources of profit and can adjust capex.

Infrastructure suppliers are more dependent on continued orders.

Conclusion

My view is simple.

If you believe in semis, you should at least partially believe in Big Tech as well.

These two things go hand in hand.

The semiconductor boom is not separate from the hyperscalers. It is financed by them.

And that’s why it seems odd to me when someone says Big Tech is dead money but is extremely bullish on companies whose growth largely depends on Big Tech capex.

Personally, I prefer to own the hyperscalers themselves today.

Meta, Microsoft and Amazon make a lot of sense to me at today’s prices.

They have cashflow, customers, distribution, ecosystems and multiple ways to monetize AI.

Semis can have greater leverage to the boom.

But Big Tech, in my view, has a better balance between quality, risk, valuation and potential return.

That’s why I think today’s negative sentiment on Big Tech could be one of the most interesting opportunities in the market.

Not investment advice. This is my market view and the way I’m thinking about AI, semiconductors and Big Tech today.