Executive Summary
Goldman Sachs delivered a striking variant perception that directly contradicts Wall Street consensus: AI infrastructure spending has contributed virtually zero to US GDP growth in 2025, not the 50% that many analysts claim. Jan Hatzius, Goldman's chief economist, methodically dismantled the popular narrative by highlighting that AI goods are largely imported and semiconductors are treated as intermediate inputs, not investment. This accounting reality means the AI boom's economic impact has been 'pretty close to zero' on measured GDP. Meanwhile, Ben Snider revealed that investors have avoided the dot-com trap by focusing on current earnings rather than speculative future productivity gains. The S&P 493 (excluding mega-caps) delivered consistent 15% returns for three consecutive years, suggesting broad-based earnings strength beyond the AI infrastructure trade. Goldman forecasts accelerating productivity growth from 1.5% to 2% annually, with the real AI boost still ahead. This creates a critical inflection point: companies that capture productivity gains rather than those selling infrastructure may emerge as the next winners. The firm's 7,600 S&P target assumes this productivity acceleration continues while unemployment remains flat at 4.5%, a historically unusual combination that signals structural labor market shifts.
Key Insights
what Jan Hatzius and Ben Snider said“When we look at the impact of AI investment on measured GDP growth on the numbers that are actually being printed, we're getting only about 20 basis points of contribution over the last three or four years and pretty close to zero over the last year”
This is a preview. Log in to see the full analysis including investment opportunities, risks, catalysts, and detailed insights.