Goldman Sachs projects that non-hardware investments related to artificial intelligence could surpass $1 trillion worldwide as companies invest heavily in data infrastructure, software and internal reorganization to capture AI-driven productivity improvements.
In a client note, analyst Joseph Briggs lays out current and potential costs tied to the AI transition. Briggs reports that labor costs already associated with AI adoption in the U.S. run at about $150 billion per year. In addition, he says executive time devoted to restructuring and planning equates to roughly $40 billion annually in organizational capital investment.
Briggs wrote: "Extrapolating labor restructuring costs incurred so far suggests that workforce reorganization could cost $800-900bn over the AI adoption cycle." That projection reflects the bank's view that the cumulative expense of adjusting workforces to new AI-enabled processes will be substantial over time.
Goldman Sachs frames the scale of this non-hardware spending as consistent with a broader shift toward intangible capital. The note states that intangible investment has grown to become roughly the same size as traditional capital expenditure across G10 economies. The bank expects this shift to produce a productivity J-curve - an initial period where resources are redirected toward internal changes before broader efficiency gains become evident.
The report also suggests that recent acceleration in U.S. productivity growth may be "likely understated" because measurement or timing issues could obscure early improvements tied to AI deployment.
Another theme in the note is industry dispersion. Goldman Sachs warns of a growing competitive divide between companies, arguing that those which deploy AI agents more effectively are likely to widen their lead over peers and could become the next "superstar" firms. The bank added: "companies focused on data structure and AI deployment companies will be key in unlocking the economic value promised by AI," and it cautioned that greater market concentration could produce outsized valuations for early movers.
Taken together, the note portrays a multi-year, economy-wide investment effort in non-hardware AI components, driven by labor restructuring, executive time commitments and strategic spending on data and software resources. It frames these expenditures as potentially transformative but front-loaded, with the prospect of later productivity payoffs.
Key points
- Goldman Sachs estimates global non-hardware AI investments could exceed $1 trillion as firms spend on data, software and organizational redesign - impacting technology, professional services and corporate operations.
- Current U.S. labor costs tied to AI are estimated at $150 billion per year, with an additional $40 billion annually in executive time allocated to organizational change.
- The bank expects a productivity J-curve as intangible investment rises to parity with traditional capital expenditure across G10 economies, implying short-term resource diversion followed by longer-term gains.
Risks and uncertainties
- Workforce reorganization carries a large cumulative cost - estimated at $800-900 billion over the AI adoption cycle - introducing execution and labor market risks for companies and industries undergoing transformation.
- The shift toward firms that best deploy AI could widen competitive gaps, increasing market concentration and valuation risks, particularly for incumbents and late adopters in affected sectors.
- Near-term productivity gains may be obscured or delayed by the J-curve dynamic, creating uncertainty for investors and policymakers about the timing and magnitude of economic benefits.