The AI bubble is bursting. At least, that’s what half of tech Twitter wants you to believe right now.
I own shares in Nvidia, Microsoft, and Meta. And while analysts are frantically publishing pieces about circular AI deals and bubble warnings, I’m doing something that might surprise you: absolutely nothing.
Not because I’m oblivious to the risks. Not because I think these companies are invincible. But because after reviewing AI toolkits for the past two years, I’ve seen something the bubble-watchers haven’t: actual utility.
The Circular Deal Panic
Recent coverage has focused on what critics call “circular AI deals” – tech giants investing in AI startups that then spend that money on cloud services from those same tech giants. Noah Smith’s Substack piece raised concerns about whether this creates artificial demand. Derek Thompson went further, outlining how he thinks the AI bubble will pop.
These are legitimate questions. But they miss a critical point.
I test AI tools every week. I’ve watched GitHub Copilot evolve from a glorified autocomplete into something that genuinely saves me hours. I’ve seen Claude go from interesting experiment to indispensable research assistant. I’ve watched Llama models get deployed in production environments where they’re actually making money.
This isn’t tulip mania. People are building real things with these tools.
What The Toolkit Reviews Tell Me
When I review AI toolkits on agntbox.com, I’m not looking at hype. I’m looking at whether something actually works. Whether it saves time. Whether it’s worth the money.
And increasingly, the answer is yes.
Microsoft’s Azure AI services are being integrated into enterprise workflows that would have required entire teams a few years ago. Meta’s Llama models are powering chatbots that handle real customer service queries. Nvidia’s chips are the only thing making any of this possible at scale.
Could these companies be overvalued? Sure. Could there be a correction? Absolutely. But calling this a bubble implies there’s no substance underneath. That’s not what I’m seeing in the trenches.
Why I’m Not Selling Nvidia
Nvidia is the obvious target for bubble accusations. The stock has had an absurd run. Everyone knows it. Everyone’s talking about it.
But every AI toolkit I review needs compute. Every model training run needs GPUs. Every inference deployment at scale needs their chips. Until someone builds a legitimate alternative – and AMD is trying, but they’re not there yet – Nvidia has a moat.
Am I buying more at these prices? No. But I’m not selling either.
Microsoft’s Quiet Dominance
Microsoft doesn’t get enough credit for how well they’ve executed on AI integration. GitHub Copilot alone has changed how millions of developers work. Azure’s AI services are becoming infrastructure.
The circular deal criticism applies here – Microsoft invests in OpenAI, OpenAI uses Azure. But I’ve watched enterprises adopt these tools because they actually solve problems, not because of some financial engineering scheme.
Microsoft’s position feels sustainable to me. They’re not betting everything on AI hype. They’re integrating it into products people already pay for.
Meta’s Underrated Play
Meta is the most interesting of the three. They’re giving away Llama models while competitors charge premium prices. This seems insane until you realize what they’re doing: commoditizing the complement.
If open-source models become good enough for most use cases, the value shifts to data, distribution, and integration. Meta has all three. They’re playing a longer game than most people realize.
Plus, their AI tools for advertisers are genuinely effective. I’ve tested them. They work.
What I’m Actually Doing
So what am I doing with these three positions? I’m holding. I’m not adding at current valuations, but I’m not running for the exits either.
Could there be a correction? Yes. Will these stocks be volatile? Absolutely. But the underlying technology is real, the adoption is real, and the utility is real.
I’ve reviewed enough vaporware AI tools to know the difference between hype and substance. These companies are building substance.
The bubble warnings might be right about timing. They might be right about valuations. But they’re wrong about the fundamental premise: that AI is all smoke and mirrors.
From where I sit, testing these tools every day, the AI revolution is just getting started. And these three companies are positioned to benefit regardless of short-term market drama.
So I’m holding. And I’m sleeping fine.
🕒 Published: