AI FIRMS CHASE DATA OVER MODELS

AI DESK■ 1 MIN READ

FRI, AUG 14, 2026

■ AI-SUMMARIZED FROM 1 SOURCE ▸ TIMELINE

Major AI companies like OpenAI and Anthropic are shifting focus toward acquiring better training data rather than building superior models. Anthropic's reported $6 billion bid for data startup Decart signals this strategic pivot.

According to Rudina Seseri, founder of Glasswing Ventures, leading AI firms face a fundamental constraint: their current models hit efficiency limits despite their scale and resources. Instead of continued model development, companies are now prioritizing data acquisition. Anthropic's reported negotiations to purchase Decart—a startup focused on data collection and curation—exemplifies this trend. Seseri argues that efficiency, not raw model size, separates winning AI companies from the rest. Better training data enables more efficient learning and improved performance without requiring exponentially larger computational resources. This shift reflects a maturing AI industry where data quality has become the competitive battleground. As model architectures stabilize, access to superior datasets offers meaningful differentiation and cost advantages.

■ SOURCES

► Bloomberg Tech

■ SUMMARY WRITTEN BY AI FROM THE LINKS ABOVE

■ MORE FROM THE AI DESK

P403AI LABS FACE CRITICISM FOR OVERCONFIDENCE

A viral essay critiques artificial intelligence research institutions for intellectual arrogance, sparking debate across tech communities. The piece argues that leading AI labs underestimate risks and limitations of their own work.

JUST NOW— AI Desk

P399UK LAUNCHES AI BOOT CAMPS FOR UNEMPLOYED YOUTH

The UK government is launching a pilot scheme offering three-week AI training courses to unemployed young people and those at risk of joblessness, aiming to tackle the growing NEET (Not in Education, Employment, or Training) crisis.

JUST NOW— AI Desk

P397ALIBABA OPENS QWEN3.8 WEIGHTS UNDER APACHE 2.0

Alibaba has released model weights for Qwen3.8, including the 27B variant, under the permissive Apache 2.0 license. The company claims Qwen3.8-27B outperforms its predecessor Qwen3.7-Plus while excelling at coding tasks.

JUST NOW— AI Desk

P394DEEPSEEK INTRODUCES PEAK/OFF-PEAK PRICING

DeepSeek has announced a dynamic pricing model that charges different rates based on usage patterns. The update aims to optimize cost efficiency for users while managing infrastructure demand.

1H AGO— Industry Desk

◄ BACK TO NEWS

AI FIRMS CHASE DATA OVER MODELS

■ MORE FROM THE AI DESK

■ SUBSCRIBE TO THE DAILY BRIEF