Opinion: What's Needed Is Not Bigger Data Centers, But Faster On-Site Solutions

Goldman Sachs' two analyses from June and July 2023, seemingly addressing different topics—South Korea's humanoid supply chain and shifts in AI investment landscapes—reveal a fundamental change in AI competition rules. A transformative opportunity is emerging in South Korea as the focus shifts from training-centric AI to inference-driven applications. The key now is real-world execution: humanoids represent the pinnacle of this shift. Goldman Sachs estimates South Korean firms could account for 30% of global humanoid production by 2035, with supply chain-driven output reaching 41,200 units by 2035. However, the real bottleneck remains data. As robots deploy in fields, training data accumulates, enhancing capabilities and expanding deployment scale. China already leads with 10,000-15,000 units deployed by 2025, while South Korea and the U.S. remain in the hundreds. The author calls for shifting national strategy metrics from 'how many units produced' to 'how many units deployed domestically.' While South Korea's 2029 target of 1,000 annual units is directionally correct, the scale remains insufficient to drive the data flywheel.