How would you allocate $10M? Here's what I'd do.

Recently a friend / Investor posed an interesting question:

How would you allocate $10M into a Consumer Products Holdco today?

Over the past few years I’ve witnessed the potential for consumer brands to run incredibly lean, high quality operations with proper AI implementation. In the beginning of May, I turned my attention back to Holstee after stepping back for a year, and in 3 short weeks I’ve done more than I would have in 3 months at a fraction of the cost. So this question was timely for me, though I hadn’t thought about it at this scale. After sitting with the question for a day, I started to see the picture more clearly.

If I had $10M to allocate, I would aim for a concentrated portfolio. I’d pick five to seven bets, weighted toward businesses where taste and brand are the moat and AI is operational leverage underneath.

Concentrated because the holdco’s real job is being useful to each operator, and that doesn’t scale across twenty bets. Five to seven is a number where you can actually know the brand, plus I wouldn’t want to go much higher at this budget. I’d be on the hunt for less mature versions of Aesop, Le Labo, or modern kitchen icons like Brightland, where the brand following is fiercely loyal, the founder’s taste is the product, and the aesthetic can’t be cloned by a competitor in a quarter.

My thesis is that AI has the ability to compress much of the operational middle (sourcing, support, copy, logistics coordination, even basic merchandising).

Take a brand like Fishwife. They revived a dusty category through brilliant art direction and a distinct cultural voice. That voice is their real moat. But underneath that beautiful packaging is a complex web of sourcing from micro-canneries globally, managing custom import logistics, and navigating intense inventory cycles.

That back-end complexity is exactly what used to require bloated operational teams. Today, it can be absorbed into a lean, shared software layer. I’m watching this compression happen at Holstee in small ways every week.

What that leaves is the part that was the real moat anyway. Taste built over years. A specific voice. A community that actually trusts the founder/ team behind it. AI can’t shortcut into any of that.

So the money would go toward founder-led brands where the person at the top has an edge when it comes to taste and is willing to run lean. My focus would be less on AI-first consumer products that are mostly wrappers and more on businesses that would have been excellent in 2010, run by someone willing to use 2026’s leverage.

In 2010 to 2020, scaling an e-commerce holdco meant hiring massive, centralized teams for performance marketing, supply chain management, and data analysis. In 2026, scaling a holdco means building a centralized infrastructure layer of bespoke AI agents that handle the rote execution.

Same logic applies to the unsexy stuff. Entity formation, equity plans, product development process. Each new bet starts with that already in place.

Where the holdco earns its keep is at the seams between brands. When one brand figures out a sourcing workflow or a customer-agent setup that works, every other brand has it within weeks. Taste-driven companies with the operational back-end of a conglomerate, without becoming one.