This week the AI frontier looked less like a single race and more like a map being redrawn. For the first time, some of the world’s most capable models are coming out of China — and being given away for anyone to run. Moonshot’s Kimi K3 and Alibaba’s Qwen3.8-Max both landed within days, pushing the trillion-parameter frontier into the open and reigniting the debate over whether U.S. dominance is slipping. But the picture is split: the models are globalizing while the chips underneath them stay overwhelmingly American, with Nvidia still holding roughly 80% of the market. And a fresh look at AI-driven layoffs is a reminder that raw capability isn’t the same as real-world impact — the human element is proving stubbornly hard to automate away.

Moonshot’s Kimi K3 Becomes the Largest Open-Weight Model Yet

A Chinese lab, Moonshot AI, released Kimi K3 in mid-July — described as the largest open-weight model ever released, its trained parameters public for anyone to download and run. The 2.8-trillion-parameter multimodal system landed hard enough to move markets.

Alibaba Previews a 2.4-Trillion-Parameter Qwen Model

Three days later, Alibaba took the stage at the World AI Conference in Shanghai and previewed Qwen3.8-Max — a 2.4-trillion-parameter model, and the first in its Qwen family above a trillion parameters to handle text, images, video, and documents together.

Are Chinese Models Something to Fear?

A widely-read Stratechery essay, “Who’s afraid of Chinese models?”, argued the panic may be overblown — while an OpenAI executive voiced the opposite worry, that free, highly capable Chinese models are something for-profit U.S. firms may struggle to compete against.

Nvidia Still Owns the Chips Underneath It All

Even as the models increasingly come from China, the chips that train and run them stay overwhelmingly American. Nvidia still holds an estimated 80 to 81% of the data-center AI GPU market, and its next-generation Rubin platform entered full production this year.

The AI-Layoff ‘Boomerang’

A Forbes analysis describes an AI-layoff “boomerang” — companies that cut staff citing AI are now rehiring those same kinds of workers six to twelve months later, finding AI handles roughly 60% of a role’s duties but struggles with the rest.