AI means we’re bundling. Again. Because -

Every business is cyclical in some way.

(pssst: You can substitute industry for business if you want to be more macro, and stick with business if you want to be more micro)

But, everything is cyclical.

And one good way to understand which direction the cycle is happening in, and where you are at any point in time, is to think in terms of bundling AND unbundling.

You might not remember the name Jim Barksdale, but he's the poet in this story. Towards the end of the Netscape IPO roadshow (remember them?), he said one of the most important lines I've ever heard: "There are only two ways to make money in business: bundling and unbundling." It hasn't stopped being true since.

A business can either be putting a bunch of stuff together, better and faster than their unbundled competition,

OR

It can be dividing stuff up into its component parts, better and faster than their bundled competition.

Note how the competition plays a role in each.

With new ideas, we typically see unbundled opportunities first. They disrupt current bundled players in some way. Startups often start here. A new tech is one idea that proliferates into related but disparate ideas, that grow like daisies through the concrete of the current "but it was made of concrete!" system until it starts disintegrating in front of our eyes.

And people marvel how the old system was so strong, and so permanent.

Phone companies were regional, built on wires and poles. They connected everything. An unbundled communication solution that rivaled rail and mail.

Then somebody said, "Wouldn't it be cool to connect one local network to a regional one?" The bundling begins.

Next, the regional becomes national, and long distance costs suck at first, but the explosion of consolidation goes farther than anybody could see. The end state - the AT&T/Bell system, fully networked and integrated - is as close to freshly poured concrete as it gets.

Then came the 1984 antitrust breakup. Seven Baby Bells were born. The unbundling was official, and you can be assured the investment bankers passed out cigars while shares were distributed. And those babies ran/grew up for roughly the next twenty years - maturing into regional fiefdoms, across fragmented maps, with no clear singular winner but lots of smaller and successful ventures.

The institution of a global phone network survived, but the “new” players were scattered.

Which made the concrete industry idea ripe for a new wave of daisies.

The wireless era was the next round of unbundlers. New infrastructure, new entrants, new chaos - the prior structure has to look overrun until some developer can buy it on the cheap and pave it over. And the bundling phase kicked back in, I’m sure you’ll recognize some of these names - Cingular absorbed AT&T Wireless, Verizon swallowed Alltel, Sprint merged with Nextel - collapsing dozens of regional players back into four national carriers.

It bears mentioning here, that I didn't have a phone bill until the cell phone phase. I lived this whole arc but wasn't meta-aware of it at the time. What woke me up to the Barksdale story was early in my financial advising days, when I'd regularly help clients assemble their AT&T stock cost basis from all the various Baby Bells. I was fascinated - and, candidly, annoyed, by how genuinely tedious the work of gathering these paper shares were - while also fascinated by how much the industry had evolved in just my lifetime alone.

The point is, this repeats in every business and every industry, over and over.

Most importantly, in a bundling phase, you want to think about the end state. The laid-in concrete, "we made it" moment.

What happened next after that moment? First it was policy pressure. Then the unbundling era. Then new entrants. Then re-bundling around scale.

This repeats. Every time.

AI is disrupting a number of areas, with software in the crosshairs for the past several years. Those companies had a concrete foothold based on coders and marketers and account men and, let's be honest, great long-term incentive plans that kept talent in place so a clear corporate ladder and career option was available to college grads for the past ten if not twenty years.

I was writing about the task-bundle version of this back in 2019. The robots-coming-for-your-job conversation was always really a bundling and unbundling conversation. What tasks can be automated, and how do you re-bundle what remains?

AI is unbundled. For now. The bundling is starting. The daisies are through the concrete, but if you haven't started connecting your LLMs via MCPs and APIs to all of your various softwares for tasks and workflows - if all of that acronmy-speak is meaningless to you - I say it only so we are clear: the bundling has started.

The end state looks like a handful of winner-take-all AT&Ts and Verizons. There will likely be a Google and an Apple in that world too. Someone will bundle all of this up. Our job is to look at the old concrete, look at the new daisies, and envision the next, impermanent but temporarily grandiose, concrete structure.

The modern companies are smart. They know ALL of this. They're racing for that position, and looking to bundle faster than they can be bundled. There will be winners and losers along the way. I don't know who they will be, and if somebody else tells you they do, they're only speculating (re: lying). This is part of how it goes, too.

So what do you do? What do WE do?

I can't tell you. Because I don't know. All I can say is to show you what I'm doing.

I've attempted to make a move on this theme, professionally. When I left the financial Walmart that is Bank of America Merrill Lynch - and you can see the bundling right there in the Elmer's glue and popsicle sticking of those iconic brand names - it was because the unbundling was already so underway that the opportunity only existed outside of the concrete platform, at least for me. I say this with love for my friends and former colleagues who are choosing to stick it out.

My bet was that a new bundling cycle was already in motion in this industry. New tools meant we could bundle better for the specific people we actually serve, instead of fitting a closed loop of products and services around some mass-market idea of what an ultra-high-net-worth client wanted. The old line applies: if you make something for everyone, you make something for no one. I felt that deeply, so my assistant Lisa and I took the plunge and joined Sunpointe in 2022.

And to think, I thought we were late(!). It's still early enough. We had no clue the AI cycle was coming right on the heels of our exit, about to change how many daisies were going to start springing up.

You too can be a part of the bundlers right now. You can play in this game. All you have to do is to look at who is new and disruptive, look at who is bundling, and keep your eye on the end state of where they're going.

Then, look at who is in concrete and at risk of crumbling. Realize they won't go away so much as be broken down or broken up in some way. Their importance is waning, that’s all. Don't be too tied into their future just because they felt permanent a short while ago.

Even daisies can bore through those cracks in concrete. Never forget that.

In your business, do business with the bundlers, and stay leery of the already bundled folks pretending nothing will change.

In your life, you have to play with the new tools, experiment with your own disruptions of prior habits and tasks, and embrace the reality that just like the world is changing, you get to change, too.

No one has the answer, but we know we're in a transition state right now.

Look around, and if you choose to deny it, good on you, but if you accept it even in the slightest, embrace the opportunity.

Oh, and above all, if you’re going to play, never forget this detail - it’s as important as unbundling and bundling itself: daisies are prettier than concrete.

I've been writing about bundling and unbundling since 2018. If you want the short version, here's the Lego take from 2024. If you want my original public thought, here's where it started.