Is Apple Winning the AI Race?

Mon, 20 Jul 2026 11:33:19 +0800

The AI Race: From Data Centers to Devices

The public AI race took off when OpenAI released ChatGPT 3.5, stunning the world with its conversational abilities. Soon after, research showed that simply increasing model parameters often improved performance, igniting an arms race for model scale and data center capacity. While companies like Google, Meta, and OpenAI poured billions into infrastructure and ever-larger models, Apple seemed to lag behind.

The Truth About the AI Moat

Open-weight models like Kimi K3 are the reason why LLM companies don't have a moat.

Here’s the secret: the much-discussed “AI moat”—a supposed insurmountable advantage built on proprietary data and models—doesn’t really exist. Frontier companies scraped data from across the web and tried to gatekeep their advancements. But open-source communities quickly reverse-engineered these models and released competitive open-weight versions to the public. “Good enough” models proliferated, slashing the cost of entry. Today, with a powerful enough server—or even a high-end Mac Studio—almost anyone can run state-of-the-art models.

Open Weight Models on Tech Journeyman

Is Apple Winning the AI Race?

The AI Race: From Data Centers to Devices

The Truth About the AI Moat