Run light with AI, build a deep moat with real assets
Originally written in Chinese. Translated with AI, reviewed by Gary. Read the Chinese original →
Will the future bring lots of "one-person unicorns"?
One person, one laptop, plus AI coding, vibe coding and all kinds of AI agents, building a company valued at a hundred million US dollars or even more.
It's an attractive idea, of course.
AI really has lowered many barriers to starting a business. Things that used to need a whole team, one person can now build as an MVP, test in the market, and maybe even start making money from.
The HALO effect
But today I came across a very interesting report. Goldman Sachs has put forward a concept called the HALO effect.
HALO = Heavy Assets, Low Obsolescence
Simply put: heavy assets that don't go out of date easily.
The concept gave me a lot to think about.
For more than a decade, everyone has loved asset-light companies. Software, platforms, SaaS and internet companies, because they scale fast, have high margins and don't need many assets.
But as writing code, design, content generation, customer service and data analysis gradually become tools everyone can use, what's truly scarce may turn out to be what AI can't copy right away.
Things like supply chains, warehouses, logistics, offline channels, distribution networks, customer trust, and real operating ability.
AI can make one person very strong. But AI can't instantly build you a real fulfilment system.
AI can generate content for you. But AI can't instantly handle your inventory, returns, broken sizes, clearance, and customers' real buying emotions.
That's not to say heavy assets are automatically good.
A mall with no demand is a burden. Inventory that won't sell is a burden too.
What's truly valuable are assets that will still be needed in ten or twenty years and are hard to copy.
People still need to eat and to dress, goods still need to move, brands still need to sell, inventory still needs someone to handle it, and supply chains still need someone to connect them.
Putting AI into a real business system
Future opportunities for founders don't come in just one shape.
It's not only one person using AI coding to build a very light company that counts as the future.
Another future might be:
You have real assets, real networks, a real supply chain and real distribution, and you use AI to make them faster, smarter and more efficient.
This is what I've been thinking about lately with AI retailing.
AI itself isn't the destination. AI is a tool.
What's really valuable is AI placed inside a real business system.
Take an apparel business. From the outside, it looks like just selling clothes. But once you actually do it, you find a lot of very heavy things inside.
Which styles will sell, which sizes will run out, when to restock, when to clear, how to sell in the livestream, how to push through affiliates, why customers buy, and why they don't.
None of this can be solved with a single prompt.
But AI can make these decisions faster.
Where we used to rely on experience, we can now add data. Where we used to rely on gut feel, we can now add signals.
The strongest companies of the future won't necessarily be the lightest.
Perhaps they'll be the ones that know how to use AI to make operations light, while having real assets and real networks underneath.
Light in operations. Heavy in the moat.
Is your heaviness a burden or a moat?
If you run a very traditional, very hard, very heavy business, it doesn't mean you have no future.
The key is: is your "heaviness" a burden, or a moat?
If your heavy assets are just old systems, old inventory and old thinking, they'll drag you down.
But if your heavy assets are supply chain, channels, customer trust, distribution networks, fulfilment capability and real data, the AI era may actually amplify them again.
People used to say: don't go heavy on assets, it's too slow.
Now I think that line needs a tweak.
It's not "avoid heavy assets". It's "avoid heavy assets with no strategic value".
The really good companies of the future may be neither purely light nor purely heavy.
Instead:
Light operations, powered by AI. A deep moat, built on real assets.
The world will certainly see lots of one-person companies.
But at the same time, it will also see many new "heavy-asset tech companies".
They may not be sexy.
They might be warehouses, logistics, supply chains, retail networks, physical outlets, or a distribution system that looks very boring.
But these boring things, once AI is added, may be the future that's truly hard to copy.