Goldman forecasts AI spending at $1.4trn by end-2027, nearly 4% of US economy

Goldman's outlook backs the artificial intelligence infrastructure trade, particularly chipmakers, chip equipment suppliers and networking companies, which benefit most from a spending trajectory well above Wall Street estimates. Its observation that most AI hardware is imported carries significance for macroeconomic investors. Strong capital expenditure does not necessarily translate into equally robust US gross domestic product expansion, and the difference appears instead as a larger trade shortfall and as demand for Asian supply chains. Playing down the regulatory threat from the midterm elections may remove one concern that has weighed on data centre-related equities. Financing costs remain the main point of pressure, as expectations of Federal Reserve rate increases push borrowing expenses higher.

---

Earlier:

• Cerebras raises 2026 outlook, sees revenue tripling in 2027 as $26bn backlog builds
• Applied Materials beats on Q4 and guides above forecasts as chip equipment demand holds
• Cisco beats and lifts FY27 view as AI revenue from hyperscalers set to more than double

---

Goldman projects AI expenditure will run at an annualised rate of $1.4 trillion by late 2027 and continue to power US capital spending, but because much of the hardware is purchased from abroad, the direct contribution to American GDP is far less than the overall outlay.

Summary:

• Annualised AI-linked hardware and intellectual property spending is forecast at $1.4trn by end-2027, nearly 4% of the US economy
• Spending is projected to surpass $900bn by end-2026
• US business investment is expected to expand roughly 8% in 2027, almost double the consensus estimate of about 4%
• AI is forecast to contribute more than 5 percentage points to 2027 capex growth, but only 0.6 point to GDP when semiconductors are included and 0.2 point without them
• Most information processing equipment is imported, so it is subtracted from GDP
• Democratic midterm victories are seen as posing only modest risks to AI investment; climbing borrowing costs are the larger threat

Artificial intelligence will remain the principal engine of US business investment growth in 2027, in the view of Goldman Sachs. The investment bank expects annualised outlays on AI-related hardware and intellectual property to hit $1.4 trillion by the close of next year, equal to nearly 4% of the domestic economy. That contrasts with its prediction of more than $900 billion by the end of 2026, Reuters reported on Wednesday.

The projection stands well above the market consensus. Goldman forecasts that US business investment will rise roughly 8% in 2027, close to double the consensus expectation of about 4%. It calculates that AI-related spending alone will account for more than 5 percentage points of capex growth next year after semiconductors are factored in.

The more notable detail is how little of that spending flows into US output. Goldman estimates AI investment will add 0.6 percentage point to US GDP growth in 2027 when semiconductors are counted, and only 0.2 point without them. The explanation lies in trade: most information processing equipment comes from overseas, so it is netted out of GDP. The bank also argued that official business investment data undercounts semiconductor expenditure. In short, the AI boom is enormous for capital spending and for the firms that supply it, but its direct contribution to the US economy is far smaller than the total outlay suggests.

On the political front, Goldman minimised the danger from Democratic successes in this month's midterm elections, which had sparked worries about stricter regulation of data centre development. The bank said the outcome poses only limited risks to AI investment.

The forecast arrives in a week when corporate earnings have reinforced the spending narrative. Cisco raised its full-year outlook on robust AI networking demand from hyperscale cloud providers, Applied Materials guided above expectations on chip equipment orders, and Cerebras disclosed a backlog of roughly $26 billion.

The principal risk is financing. Climbing borrowing expenses and more stringent investor oversight could slow AI capital spending, particularly as companies lean more on debt to fund the build-out. With markets pricing a strong chance of another Fed rate increase in December, the cost of that debt is worth monitoring.