AI巨頭開始與美國財政部搶錢
Microsoft、Alphabet、Amazon、Meta與Oracle的AI資料中心資本支出快速上升，可能迫使它們透過公司債與結構化融資籌集資金。文章認為，AI債券供給正在擠壓美國國債需求，推高長期利率並對高估值AI公司的估值模型造成壓力。
Search
10 results on this page
AI Data Centers Trigger BBU Cell Rush
Rising AI server power density is turning cabinet-level Battery Backup Units from an optional component into a required part of data-center power architecture. Global BBU cell shipments reportedly reached about 276 million units in the first half of 2026, intensifying shortages and competition among Asian battery suppliers.
🔬
Spectral Neuron Makes ML More Interpretable and Controllable
The Spectral Neuron is a proposed ML primitive based on the largest eigenvalue of an input-conditioned matrix. The preprint develops its mathematical properties, initialization and training methods, and evaluates scalability on synthetic and real-world datasets.
Skanska Wins $1.2B US Data Centre Deal
Skanska has signed a $1.2 billion contract to build four data centres for an unnamed existing customer in the southeastern United States. The facilities will cover approximately 75,000 square metres, or 808,000 square feet.
ByteDance Restructures Seed AI Team
ByteDance’s Seed foundation-model division has reportedly completed another internal restructuring. Its foundation-model organization now includes four first-level departments focused on pretraining data, reinforcement learning, product post-training for work, and product post-training for chat.
⚙️
Stecon Targets $1.5 Billion Thailand Data Center Bids
Stecon Group Pcl plans to bid for three data-center construction projects valued at a combined 49.8 billion baht, or about $1.5 billion. The company expects Thailand’s accelerating technology investment to drive further growth.
Transfer More Knowledge with Less Multilingual Data
Apple Machine Learning presents a lexical-intervention approach for improving cross-lingual knowledge transfer when target-language data is scarce. The work targets downstream capabilities such as scientific reasoning, commonsense inference, and world knowledge without relying heavily on parallel data, translation systems, or auxiliary models.
Scaling Laws for Data-Constrained Mixture Training
Apple Machine Learning studies how to mix scarce target-domain data with abundant generic data during language-model pretraining. Across more than 2,000 training runs, the research examines the trade-off between insufficient target exposure, repeated examples, diminishing returns, and overfitting.
OpenAI Raises the Stakes on Enterprise AI Privacy
OpenAI is seeking to compete more directly with Anthropic by introducing stronger privacy protections for enterprise customers. The two companies are increasingly competing over how securely customer data is handled.
Automate End-to-End Document Intake
AWS demonstrates how a mid-size mortgage lender automates document intake from email through classification, extraction, validation, and final data processing. The workflow combines the AWS GAIIC IDP Accelerator with Amazon Quick Automate.