本周AI数据：GB200部署不足5%，35B模型匹敌万亿参数

Hi all,

Here’s our short Monday roundup of data signals across AI, energy and markets:

• A GPU wave is ahead of us. More than 95% of the Grace-Blackwell GPUs have not yet been deployed, even though the chip has been shipping since December 2024.1 (h/t )
• Freelance hunting. Fable 5 now completes 16% of real freelance projects at a quality just as good as that of human professionals, according to the Remote Labor Index. This is about double the previous best2.
• David vs Goliath. A 35-billion-parameter model, trained differently3, now matches 1-trillion-parameter models on some long-horizon benchmarks.
• Chinese pharma boost. Chinese biotech companies’ licensing deals are up 87% in the first five months of the year compared to the same period last year.
• Wet-lab assistance. In a near-autonomous loop, GPT-5.4 helped Molecule.one’s lab run 10,080 reactions and increased the average yield in the Chan-Lam process4 by around 50%.
• Holding swap. Central banks worldwide now hold more gold than US Treasuries – this is the first time since 1996.
• Wealth concentration. Nearly half of all income earned in the US belongs to the top 10% of earners, the highest since WWII.
• Fewer babies, richer adults? Across countries, a one-percentage-point lower birth rate in 1950 was associated with almost 27% higher GDP per working-age adult over 1970-2020, with no change to aggregate GDP.
• Extreme heat. French utility EDF took almost 10% of its nuclear fleet offline (some 6.2 gigawatts) as river temperatures breached the threshold for cooling.

Thanks for reading!

1

Deployed describes chips that have been installed and are in active use. The state of announced chips, whether delivered or not, is not broken out.

2

The Remote Labor Index aims to measure how well AI agents can complete real freelance projects, spanning video, audio, 3D/CAD design, data analysis, and more, with human evaluators as judges of performance.

3

Using a new horizon-scaling method for AI agents. Horizon scaling refers to supervising agents on long‑context, multi‑step trajectories and training the entire decision process, not only the end answers.

4

This process is important in forming carbon-nitrogen bonds needed in many medicines.