You’ve probably felt it: that creeping dread every time you open LinkedIn and see another CEO swearing they’ll invest $100 billion in AI this quarter. Or that moment you realize your company’s AI strategy is just ‘copy what the big guys are doing.’
Here’s the truth nobody wants to say out loud: The current AI arms race isn’t just wasteful—it’s actively accelerating the death of the very companies running it.
Let me show you what I mean. Remember when everyone said retail was ‘transform or die’? Suning and Gome chose to transform. They poured billions into ads and stores. JD.com poured money into R&D. Who’s still standing? History doesn’t repeat, but it rhymes—and right now, the rhyme is playing on a loop for every tech giant that thinks throwing cash at AI will save them.
Baidu is the perfect case study. For years, Baidu Search was unassailable. Then DeepSeek, Doubao, and dozens of AI chatbots came along. Now your dad asks his phone a question instead of typing into a search bar. That’s not a competitive loss—that’s a dimensional shift. You can’t outrun a paradigm by running faster in the old direction.
And Baidu is just the first domino. Tencent? Safer, because social networks are moats AI can’t easily cross. But even the Moat Kings have to ask: what happens when communication moves from ‘you send me a message’ to ‘your AI sends my AI a message’? The answer is terrifying if you’re a company built on eyeballs and clicks.
Here’s the paradox that’s eating the smartest people in the room: Every giant knows it must invest in AI to avoid waiting to die. But the way they’re investing is actually making them die faster.
Think about it. The AI arms race is a war of attrition. Giants compete on who can spend more on R&D, who can offer free tiers longer, who can hire the most PhDs. All of that is paid for by the core business—the one that’s already being disrupted by AI. It’s like a patient paying for chemotherapy by selling their own organs. Eventually, there’s nothing left to sell.
I saw this firsthand when I watched Kuaishou spin off its AI video model, Kling. Kuaishou didn’t fund Kling internally—it raised $3 billion from outside investors, including BAT and top global VCs. The valuation? $18 billion. That’s a company that can stand on its own. Meanwhile, Baidu’s Kunlun chip is spinning off for an IPO reportedly worth more than Baidu itself. When a spin-off is worth more than the parent, the parent has a strategy problem, not a technology problem.
This is the playbook that works: incubate AI ventures inside the mothership, let them prove their tech, then spin them out with independent funding and independent management. The giant becomes a venture capitalist, not an operator. The core business provides cash flow—not a life support system for a dying model.
And yes, I know what you’re thinking: ‘But ByteDance is charging users for Doubao! They’re making AI pay for itself!’ That’s a bet. And it’s a risky one. Because the moment you charge users, you’re competing in a market where the best product might be free tomorrow. Independence is safer. Capital markets are patient; quarterly earnings reports are not.
But here’s the twist that changes everything: Current AI is not creating new wealth. It’s redistributing old wealth.
Every AI that replaces a call center agent, every model that optimizes a supply chain, every tool that automates a spreadsheet—these are efficiency gains. They take money from one pocket and put it in another. They don’t create new industries, new demand, or new sources of growth. The Industrial Revolution created railways, cars, and electricity grids. AI has created chatbots and image generators. The difference is not subtle.
We’re in the early steam-engine phase. The first steam engine took 15 years to go from patent to commercial use. The first car took 10 years from prototype to production, and another 22 to scale. AI is at that same awkward stage: powerful enough to be dangerous, not yet powerful enough to be transformative in a wealth-creating sense.
Which means the giants that survive this era won’t be the ones that win the R&D spending war. They’ll be the ones that stay alive long enough to see the real transformation—and then buy their way in.
You don’t have to be the inventor of the internal combustion engine to profit from the automobile. BMW, Toyota, and Mercedes didn’t invent the engine. They just kept their options open, invested wisely, and stayed on the board. The same applies to AI. You can miss the first wave. You can miss the second. But you can’t miss the table entirely.
So here’s what I’m really saying: Stop trying to win the AI race. Start trying to survive the AI winter that’s coming.
Spin off your AI projects. Let them raise their own money. Use your core business to generate cash, not to subsidize a losing war. Invest in the startups that will eat your lunch—because if you don’t, someone else will. And when the real AI revolution finally arrives—the one that creates new trillion-dollar industries, not just more efficient versions of old ones—you’ll be there, with a seat at the table and a checkbook in hand.
Because the only thing worse than being Kodak is being the company that saw Kodak coming and still did nothing.
FAQ
Q: Isn't this just a fancy way of saying 'big companies should do nothing'?
A: No. It's saying stop doing the wrong thing. The 'do nothing' approach is waiting to die. The 'spin off and invest' approach is a proactive capital allocation strategy. It's the difference between burning cash on a losing war and buying land in the country that will win the next war.
Q: What if a giant's AI spin-off fails? That's a huge loss.
A: It's a smaller loss than betting the entire company on a moonshot that fails. Spinning off limits downside: the parent company has a minority stake, the spin-off has its own funding, and if it fails, it doesn't take the mothership down. The real risk is letting a failed internal AI project drain the core business for years.
Q: But what about companies like Google and Microsoft that are integrating AI deeply into their products?
A: They're the exception, not the rule. They have the cash, talent, and market position to make integration work. But for 90% of tech giants, integration is a trap. They're not Google. They're Suning trying to copy JD.com. The playbook is different when you're not the market leader—and pretending otherwise is how you become a footnote.