Forbes AI 50 2026: OpenAI and Anthropic Lead, 20 New Companies Revealed

I’ll be honest—every year when the Forbes AI 50 list drops, I actually look forward to it. Not because the rankings are gospel, but because it’s like a mirror showing where this industry is actually heading.

The 2026 list just came out, and a few numbers caught my attention: OpenAI and Anthropic alone have raised a combined $242.6 billion. That’s roughly 1.66 trillion RMB. To put that in perspective, you’d need to bundle together the valuations of China’s top 10 AI companies and you still might not catch up.

What’s more interesting is that 20 companies are new to the list this year. Looking through the names, there’s a clear trend emerging—pure foundation model startups are becoming rare. Instead, AI application layers, vertical industry solutions, and AI infrastructure companies dominate.

This reminds me of a conversation I had recently with a friend in AI investing. He said the window for foundation model plays is essentially closed. Not because there’s no technical room to grow, but because capital isn’t willing to bet on “the next OpenAI” anymore. The reasons are straightforward: training costs are astronomical, monetization takes forever, and the head-start effect is already locked in.

One company on the list that stuck with me does AI-powered legal document analysis. Not because their tech is revolutionary, but because their customer retention rate is over 95%. In the AI industry, that’s basically unheard of—most AI products lose monthly active users faster than the stock market drops.

Another company doing manufacturing quality inspection AI has already hit $100 million in annual revenue. Their core strength isn’t having the best model—it’s their ability to squeeze AI algorithms into legacy factory equipment, making 20-year-old machines “intelligent.” This “backward compatibility” might be more valuable than pure algorithm accuracy.

Of course, it’s no surprise that OpenAI and Anthropic continue to dominate. But look closely at their valuation structures, and you’ll see investor expectations have shifted. It’s no longer just betting on whether AGI will happen—it’s about who can turn large models into actual infrastructure, like AWS did with servers.

After the list dropped, I saw comments in tech groups saying “same old faces, boring.” But my take is that the list’s value lies precisely in this “boredom.” When the AI industry moves from hype to actual implementation, the rankings naturally become less exciting—and that’s a healthy sign.

One final observation: there are fewer Chinese companies on the list this year. Some will interpret this as “China’s AI is falling behind,” but I think the more accurate explanation is that domestic AI companies operate under increasingly different funding and valuation logic than the US market. Forbes’ criteria may simply not fully apply.

The list is a reference, not the answer.