Goldman Sachs Global Embodied AI: Charting Humanoid Robot Progress, with a Focus on Logistics
Summaries are public. PDF access requires an active membership; all members have the same access. Sign in
Goldman Sachs' cross-sector research team — spanning autos, tech and industrials — maps embodied AI across hardware, software and semiconductors and raises its humanoid robot forecasts to roughly 890,000 units in 2030 and 6.5 million in 2035, a market opportunity near $138bn. Logistics and warehousing emerge as the leading early adoption scenario, driven by e-commerce and retail giants such as Amazon and Walmart: over 50% of e-commerce fulfillment already runs through automated facilities, and upside analysis suggests automation could deliver Amazon about 5.6% cost-to-serve leverage, roughly $72bn of savings and a ~240bps EBIT margin tailwind by 2030.
In autos, at $20,000-60,000 unit prices and 10-50% adoption, humanoids could gross 1-6% of operating-margin expansion, with unit economics of $45,000 initial investment, $28,300 net annual savings and a ~1.6-year payback. Scale and design optimization should cut industry costs about 7% a year and accelerate the shift to virtual PLCs, growing 20-30% annually, with roughly 25% of new PLC sales virtual or software-based by 2030. Buy-rated beneficiaries include Jabil ($482 target), Flex ($154), TE Connectivity ($284) and Amphenol; Sanhua H/A is Buy/Neutral (HK$40.0/Rmb38.6), Shuanghuan Driveline Buy (Rmb45.5) and Daifuku Buy. Tesla's Optimus line is being installed at Fremont for 2026 production, with external shipments targeted by end-2027 at $20,000-30,000 scale cost.
Key exhibits
Exhibit 1
Exhibit 2
Exhibit 3
Exhibit 4
Exhibit 5
Exhibit 6
Exhibit 7
Exhibit 8
