Goldman Sachs warns that the rapid surge in corporate spending on artificial intelligence is substantial but so far appears to be producing only limited national crowding‑out effects, according to a client note reviewed by Investing.com.
Scale of AI investment
The US investment bank estimates AI capital expenditure in the United States will reach roughly US$600 billion in 2026 — a sum the bank described as "equivalent to nearly 2% of US GDP". Converted at R18.50 to the US dollar, that equates to about R11.1 trillion in 2026.
"equivalent to nearly 2% of US GDP,"
Goldman analysts note that AI-related outlays have recently accounted for more than 10% of business fixed investment in some quarters, reflecting a concentrated wave of purchases of specialised hardware, software and associated services.
Where crowding out shows up — and where it does not
Rather than finding widespread displacement of other corporate investment, Goldman reports specific adjustments in spending patterns:
- Large cloud providers — the hyperscalers — financed much of their AI push by trimming share buybacks and by taking on debt, indicating an internal reallocation of capital rather than an economy‑wide squeeze on investment.
- AI hardware and systems are heavily imported, which reduces domestic supply pressures but raises trade and supply‑chain considerations.
- Data‑centre construction has risen to about 9% of private non‑residential construction, but this increase coincided with a fall in subsidised manufacturing facilities, meaning the overall demand for construction resources shows only modest net upward pressure.
The bank concludes that, at present, there are "only limited signs of crowd‑out nationally" — in other words, the reallocation to AI has not yet produced large, measurable declines in non‑AI investment across the broader economy.
Credit markets and financing
Goldman also examined how AI projects are being financed. The note finds AI‑related financings have grown to roughly a quarter of investment‑grade issuance. Despite that, spillover effects into broader credit conditions appear contained: non‑AI credit spreads remain close to long‑run lows, which suggests lenders are not broadly repricing risk because of AI demand for capital.
The bank highlighted that some companies were prepared to borrow at current interest‑rate levels to fund AI strategies, while others shifted cash away from buybacks to free resources for capital expenditure.
Estimated magnitude of crowding out
Goldman’s back‑of‑the‑envelope estimate of displacement is modest relative to the headline AI figure: roughly US$50 billion of incremental crowding‑out in 2026, which is about R925 billion using an exchange assumption of R18.50/US$1. That represents only a small fraction of the total AI investment the bank projects.
| Measure | US$ | Approx. R (R18.50/US$) |
|---|---|---|
| Projected AI investment (2026) | US$600bn | R11.1 trillion |
| Estimated incremental crowding‑out (2026) | US$50bn | R925 billion |
What this means for South African readers
The global emphasis on AI matters for South Africa through several channels. Import‑heavy AI investment limits direct manufacturing opportunities locally but can lift demand for high‑end services, cloud capacity and specialised skills. The shift away from buybacks towards capex in large multinational firms may affect global equity supply and investor returns, while the sizeable borrowing in the sector could shape global capital costs if it intensifies.
For households, the immediate impact is likely to be indirect: increased productivity from AI could support global growth and job creation in some sectors, but the import profile and concentrated nature of investment mean benefits may be uneven. Policymakers and business leaders here will need to weigh whether to pursue local incentives for manufacturing AI hardware, strengthen skills programmes for AI‑related roles, or focus on services and data infrastructure where South African firms can compete.
Goldman’s headline: AI is a very large capital story, but its drag on other investment and on the broader economy is currently smaller than many commentators fear. That balance — between opportunity and displacement — will depend on how investment, trade and financing patterns evolve in the months ahead.
WE NEWS does not provide financial advice.
