Eighteen months after DeepSeek-R1 panicked markets by suggesting cheaper training invalidated compute investment, Kimi K3 with 2.8T MoE parameters has been called DeepSeek 2.0 moment. The answer is opposite to 2025 fears: Jevons Paradox holds where efficiency gains drive increased total compute consumption.The scale of capital expenditure defies previous expectations. Google raised its 2026 CapEx guidance by $15 billion to $195-205 billion, with…
This story is only covered by news sources that have yet to be evaluated by the independent media monitoring agencies we use to assess the quality and reliability of news outlets on our platform. Learn more here.