Advertisement
“Given the massive investments, efficiency is paramount. Otherwise, it amounts to significant waste,” Liang told employees on Thursday.
“We must continuously pursue cheaper computing solutions,” he said. “This involves optimising efficiency across all levels: data centres, power consumption, GPU/CPU [graphics processing unit/ central processing unit] utilisation, system architecture, algorithms and product engineering.”
Advertisement
ByteDance planned to spend about 100 billion yuan (US$14 billion) on AI chips from Nvidia in 2026, a hefty increase from roughly 85 billion yuan in 2025, the South China Morning Post reported last month, citing people familiar with the matter.
