I agree on this being a bubble that SaaS companies are only helping to inflate out of self-interest. But it's important to note that none of this is new. Software companies do two things- make software and sell software. They are always looking for a new marketing angle. Something like AI that you essentially create wrappers around the same few services is quickly commoditized so a company like Salesforce will spin it with their own angle ("ours will protect your privacy" or the more recent agent spin).

Of course it doesn't work as advertised, but that's not the intent. The intent is to market the software. It's an investment.

Where I completely agree, though is that AI is eventually going to be a black hole that in a decade will be looked back on as what put Microsoft, Amazon and especially Meta into years long holes financially given the size of the bets they're making. I'm not as bearish on SaaS companies due to this though because their investments are very small and they can quickly pivot to the next hot thing that no doubt emerges in a few years.

I do think the hyperscalers are united by a common thread in their pursuit of AI which is not SaaS companies but rather the CEO's like Zuckerberg and Dell having an axe to grind with Steve Jobs and Apple as well as Google and having this unending desire to beat them and create a new platform. Mark Zuckerberg is a smart dude. His instincts for running a social media company are second to none. But he has never successfully launched a consumer hardware product because doing so is extraordinarily difficult and I don't think is overcome simply by money and a lot of smart people.

If you follow the arc of the smartphone it goes all the way back to the telegraph and it all boils down to one thing- enhancing communication. Steve Jobs and Apple understood how to move forward the phone in a revolutionary fashion and the device is more or less now perfected. Interestingly, Apple hasn't made huge investments in AI. Nothing prevented them from spending tens of billions on data centers and Nvidia GPU's yet they didn't. Instead they outsourced the off-device LLM piece to OpenAI. I think that speaks volumes. I mention the arc because Apple's unveiling of the iPhone was a natural progression from what people had already demonstrated a preference for and were used to using. Steve Jobs himself said that the phone became popular not because it was inherently superior but because telegraphs required specialized knowledge to use so nobody owned them. The great leap forward came when phones were made easy enough to use that anyone could operate them. I don't see a natural progression from what a smartphone does to a headset.

Now Mark Zuckerberg seems hell-bent on creating this new platform but it is a near certainty he will fail simply because the iPhone is an extraordinarily high bar and people's preferences are well established. They love their iPhones and they love Instagram.

Hyperscalers are under enormous pressure from Wall Street to perform. If your largest competitor is investing tens of billions on this thing that everyone thinks will dwarf the internet, you'd better either follow suit or show that you're beating them without doing it.

So the whole thing becomes this positive feedback loop where Microsoft in an attempt to overtake Google and Apple goes all in on OpenAI. In response, Amazon, Oracle and Google respond by doing the same. Meta is kind of a special case because I truly believe Zuckerberg is still chasing the dream of creating a platform to destroy the iPhone. Then SaaS companies follow by showing that they're on the bleeding edge of this AI stuff and quickly incorporate it into their products and talk up its benefits ad nauseum.

To me its been self-evident for a long while that these tools just aren't that useful at scale. What will be the thing that pokes a hole in it is anybody's guess, it probably won't be one thing. It seems that cracks a...