The AI Trade Isn’t Dead. It’s Moving On.

For the last few years, the AI trade has been pretty straightforward. Buy the companies selling the picks and shovels. More GPUs. More data centers. More networking gear. More memory. More power. NVIDIA has obviously been the name at the center of it. Not the only name, but the one that defined the trade.

I think that part of the trade still matters, but I also think we are moving into the next phase. Meta’s new Muse AI agent is a good example of what I mean. Muse is not just another chatbot meant to answer questions. It is supposed to do things. Book appointments. Fill out forms. Move between apps. Work through basic tasks for the user. Meta is rolling it out through its own app and WhatsApp, and it has also pushed it further into the Apple ecosystem, where it can interact with files, messages, calendars, notes, and email. We have all heard enough AI hype by now to be skeptical of another product launch. I get that. But from an investment standpoint, this one is worth watching because it points to where the money may start to shift.

The big tech companies have already spent staggering amounts of money building the AI backbone. Meta, Microsoft, Amazon, Alphabet, Oracle, and others have poured hundreds of billions into chips, data centers, and power. Meta alone is expected to spend somewhere in the range of $130 billion to $145 billion in 2026. Across the industry, AI infrastructure spending could be above $800 billion this year and possibly over $1 trillion in 2027. That is a massive number. At some point investors are going to ask a simple question: where is the return? The market was willing to reward the buildout phase because the demand was obvious. Now the conversation is starting to change. These companies have to show that the spending can turn into revenue, margin, and cash flow.

Meta has something most AI startups do not have: distribution. It does not have to fight for attention in the same way because it already owns some of the most-used apps in the world. Facebook, Instagram, WhatsApp, and Messenger give Meta a built-in path to billions of users. If an AI agent can be dropped into tools people already open every day, that is a real advantage. Muse reportedly reached number two in Apple’s U.S. App Store rankings shortly after launch, behind only ChatGPT. That does not mean it is a business yet. Downloads are not revenue. But it does tell me people are at least willing to try it, and that is the first step toward monetization.

To be clear, I am not saying the infrastructure trade is over. NVIDIA, Micron, AMD, Broadcom, and the data center names still matter. The buildout is not stopping. But the eventually the buildout will slow down as we reach needed capacity. The next winners may not be the companies that simply provide the hardware. They may be the companies that turn that hardware into products people actually use and pay for. That is not so different from the late-1990s internet buildout. Cisco and Nortel benefited from the infrastructure cycle, but the real long-term wealth was created by companies built on top of the network: Google, Amazon, Netflix, Facebook. The same kind of shift may be starting in AI.

So the question is not whether AI is real. I think that debate is mostly over. The better question is who captures the economics. Meta is one answer, but not the only one. Microsoft has distribution through Office, Teams, Azure, and Windows. Apple has the device layer. Amazon has AWS and consumers. There may also be a company we are not talking about yet that ends up becoming one of the biggest beneficiaries. That is usually how these cycles work. Investors crowd into the obvious names first, and then the second wave shows up later.

The Trade

If I wanted to express this view through Meta, I would probably look at longer-dated calls or a defined-risk call spread rather than simply chasing the stock. A LEAP would give me time for the thesis to play out, but the premium matters. That is always the trade-off with buying calls outright. You get exposure to the upside, but you also need the move to happen before time decay eats away too much of the option value.

That is where a bull call spread could make sense. In simple terms, I would buy one call option at a strike price closer to where the stock is trading and then sell another call option at a higher strike price with the same expiration date. The call I sell helps pay for the call I buy, which lowers the total cost of the trade. The downside is that I also give up some upside beyond the higher strike price. So I am not betting on unlimited upside. I am betting that Meta moves higher over time, but within a reasonable range.

For example, if Meta were trading around a certain level, I might buy a call near the current price and sell a call a few strikes above it. My maximum loss would be the net premium I paid for the spread. My maximum gain would be the difference between the two strike prices, minus what I paid to enter the position. I know the risk going into the trade, I reduce the cost compared with buying a call outright, and I still have a way to participate if the stock rerates higher as investors start valuing Meta less like a social media company and more like a serious AI monetization platform.

I would not treat this as a short-term product-launch trade. Muse could flop, and Meta could still be fine. Muse could work, and the stock might not immediately care. The bigger idea is that AI may be moving from the buildout phase to the monetization phase. If that is right, the next leg of the AI trade may look very different from the last one. The market already knows these companies can spend. Now it wants to know who can turn the

spending into products, revenue, and earnings power. That is the part of the trade I want to watch most closely.