Goldman Sachs: AI investment shifts to factories and energy

AI moves from data centers to factory floors

Artificial intelligence investment is entering a new phase focused on the physical economy—factories, energy facilities, mines, and manufacturing plants—according to an exclusive Goldman Sachs report shared with Axios.

The investment bank projects approximately $7.6 trillion in global AI infrastructure spending between 2026 and 2031, covering compute resources, data centers, and power systems. Yet software represents less than 0.5% of global GDP, leaving what Goldman describes as "the other 99.5%" of the economy as AI's next frontier.

Why it matters

While investor attention remains concentrated on frontier AI labs and large language models, the real economic transformation—and the bulk of capital deployment—will occur as AI spreads into industrial operations that employ millions and generate trillions in economic output. This shift could reshape labor markets and competitive dynamics across traditional industries far more dramatically than software-focused AI applications.

Boardrooms face automation urgency

Conversations with manufacturing executives have fundamentally changed, according to Mark Sorrell, Goldman Sachs' global head of industrials. The question is no longer whether factories will adopt AI, but how rapidly automation will expand across production lines.

"There's definitely a little bit of, 'If I don't move, do I get left behind?'" Sorrell told Axios, describing the sentiment in corporate boardrooms.

Looking forward a decade, many manufacturing leaders expect production facilities to rely far more heavily on robots than human workers, particularly for hazardous tasks.

Tech and industrial boundaries blur

The traditional separation between technology companies and industrial firms is dissolving. AI has made tech companies more collaborative with non-tech enterprises, Jung Min, Goldman's global co-head of tech, media and telecommunications, told Axios.

"We're only just beginning" to observe AI's impact on industrial businesses, Sorrell said.

The Goldman report notes that determining "which sectors thrive and which stall" represents "a challenge of capital architecture as much as engineering."

Deal activity accelerates

Technology mergers and acquisitions have already reached $566 billion in 2026, compared to $334 billion for all of 2025, according to Dealogic data cited in the Goldman analysis.

Geopolitical tensions have not slowed transaction activity in the space, Sorrell noted, as global energy prices have declined.

Min argued that companies tapping new capital sources to meet AI demand across the economy represents a "healthy" development, even as AI capabilities advance faster than deployment capital can materialize—a dynamic some view as evidence of a potential market bubble.

The bottom line

AI's next chapter will increasingly unfold across the physical economy rather than remaining confined to Silicon Valley, Goldman Sachs concludes.

These details were first reported by Axios.