Here’s how much the AI boom is contributing to inflation, according to Goldman Sachs calculations

Goldman Sachs estimates the AI boom has added about 45 basis points to inflation, while arguing the effect is peaking and its impact has been overstated. Core PCE inflation was 3% year over year in August; Goldman expects it to move back toward the Federal Reserve’s 2% target in 2027.

Analysis

Market mechanism: The key distinction is whether AI adds a temporary price-level impulse or persistent inflation. Data-center construction and electricity demand can lift near-term goods, utility and infrastructure costs; productivity gains may later reduce unit costs. If the inflation impulse is genuinely fading, the larger market channel is a lower expected path for policy rates—not a direct earnings signal for Goldman Sachs. The estimate is a model attribution, not an independently observable series, so it should not be treated as a precise inflation forecast.

Relative effects: A softer rate path would generally support duration-sensitive equities and Treasury duration. Conversely, power-intensive AI operators could face higher input costs even as their long-run productivity case improves; utilities and power suppliers may benefit from demand, but permitting and grid constraints determine who captures the economics. These are conditional spillovers, not company-specific earnings conclusions.

Timing and risks: Days: limited standalone trading value; the estimate is unlikely to outweigh incoming inflation and labor data. Over 1–3 months, watch core PCE composition, electricity prices, wage growth and Fed communication for confirmation that the impulse is peaking. Over 6–18 months, productivity gains could be disinflationary, but continued data-center buildout or power bottlenecks could prolong cost pressure. A single Goldman estimate does not establish the Fed’s reaction function or Goldman Sachs’ financial outlook.

Key Decisions for Investors

• Do not trade GS on this macro estimate alone; it is an economic view, not evidence of a change in the firm’s earnings or balance-sheet outlook.
• Keep Treasury-duration exposure tactical rather than making a large duration bet now. Consider adding 2-year Treasury duration only if subsequent core PCE and wage data corroborate easing inflation and Fed communication turns less restrictive; reduce it if inflation reaccelerates or power costs broaden into services.
• Watch the relative trade between AI beneficiaries and power-constrained operators: the productivity narrative supports the former, while rising electricity and infrastructure costs could pressure the latter. Require evidence in company guidance or power-price data before positioning.
• Falsification alert: renewed acceleration in core services inflation, wages or electricity prices—or continued upward revisions to data-center power demand—would undermine the ‘peaking impulse’ thesis.