Legacy vs AI-native companies

There’s a thought experiment I like to return to. Imagine we’re in Three Body Problem. An alien civilization threatens to freeze our technology. Models stop improving. The AI we have today is the AI we’ll have forever. Would the world still transform?

The answer is yes. Profoundly so.

Claude Code and the rest of the AIs tools we’re experiencing are incredible. And yet, I think they are at maybe 10% of what they could be with a mature platform. The integration into workflows is primitive, the surrounding tooling is still nascent. We could improve infrastructure for years without touching the models and still completely change the world.

But the models aren’t freezing, they’re accelerating. Every 3-4 months, another leap. So we have two exponentials. Model capability + Infrastructure. And they don’t add, they multiply.

This creates a shift that most people haven’t grasped yet. The old question was: “How can AI help me do this task?” The new question is: “What should I be doing?”

I have a friend who sells jewelry online. He asked me how to make TikTok scripts. The right question was: “What should I do with my business?” Let AI figure out that growth is the answer. LEt AI identify TikTok is the best growth channel. Let AI write the scripts. Let AI test which ones work. The paradigm shift is from “AI helps me execute” to “AI helps me think”. Hell, don’t think. Let AI think and just review.

This feels uncomfortable. It should. It’s the same discomfort our grandparents felt when we told them to “just Google it” instead of remembering facts. Our generation grew up knowing information is free. The next generation will grow up knowing reasoning is free, and they will have the capacity to manage one, ten, a hundred companies at the same time.

Dario Amodei predicts a single-employee billion-dollar company by 2026. That person isn’t hypothetical. They’re probably 25 years old right now, building in their apartment, with AI doing the work of a hundred engineers. The future arrives faster than we can think about it.

When dropshipping went mainstream around 2015, traditional retailers didn’t see it coming. They were optimizing inventory management and negotiating with suppliers while teenagers in their bedrooms were launching 50 stores a week on Shopify.

Traditional retailers were caught off guard. They’d spent decades building relationships with suppliers, optimizing warehouses, training staff. Meanwhile, Shopify acquired Oberlo in 2017 and suddenly anyone with a laptop could import products from AliExpress and start selling within hours.

What happened to the incumbents? The market split in two. Mass-market products became commodities. Profit margins dropped from 30-50% to 10-20%. The middle got crushed between dropshippers racing to the bottom and luxury brands holding the line.

Something similar is about to happen with AI.

AI cuts time to first prototype from weeks to days. Tools like Lovable let solo entrepreneurs create functional products in hours. OpenClaw allows anyone to build a full company of agents running your business 24/7 for you.

When you can iterate 100x faster than competitors, you win. A fashion AI-native could test 100 designs per day while a traditional company agonizes over one collection per season. And there will be a thousand of these new AI companies, even if only a niche of users can access them.

And legacy companies can’t just “adopt AI” to catch up. Netflix took 7 years to complete its cloud migration. The average enterprise digital transformation takes 3-5 years. The math is simple. AI-native companies iterate in days when legacy companies transform in years. If you’re too slow, you die.

There’s a chance they adapt quickly. Employees didn’t wait for permission before using chatGPT to perform better. One can hope the pattern continues with the new AI capabilities.

So what do legacy companies actually have? Two things. Brand and manufacturing.

Brand provides time. Wide moat brands can maintain competitive advantage for 20+ years. Coca-Cola has raised prices through wars, recessions, and inflationary cycles for a hundred years. But brand is defensive, it only buys time until your competitors build theirs.

Manufacturing is different. Until robotics catches up, physical production is the bottleneck. The AI-native startup can generate 1,000 product designs in a day. They cannot manufacture 1,000 products without factories, suppliers, and logistics infrastructure.

This is exactly what happened in the dropshipping era. China became king precisely because they had the manufacturing capacity. Their production now exceeds that of the nine next largest manufacturers combined. The China factories that powered dropshipping were the strategic asset. Not the Shopify stores. Not the Facebook ads.

If you’re a legacy company, your factories are your most strategic asset now. Not your ideas, not your marketing, not even your current products.