On November 11, 2026, Goldman Sachs forecast that annualised spending on AI-related hardware and intellectual property in the United States will climb to $1.4 trillion by the end of 2027, close to 4% of the economy, from more than $900 billion expected by the end of 2026. Reuters reported the forecast on November 12. This page sets out the numbers as reported, as of November 13, 2026, and compares them with other banks. Forecasts are estimates, and the banks do not measure the same thing.
The November 2026 Forecast
| Measure | Goldman Sachs figure |
|---|---|
| AI hardware and intellectual property spending, annual rate, end of 2026 | More than $900 billion |
| Same measure, end of 2027 | $1.4 trillion, nearly 4% of the United States economy |
| Contribution of AI investment to 2027 capital spending growth, with semiconductors | 5.3 percentage points |
| Same, without semiconductors | 4.3 percentage points |
| Contribution to 2027 GDP growth, with semiconductors | 0.6 percentage points |
| Same, without semiconductors | 0.2 percentage points |
| United States business investment growth in 2027 | About 8%, against a consensus of 4.2% |
Goldman's summary line, as quoted by Reuters, is that AI "will remain the dominant driver" of business investment growth in 2027.
The Basis for the Numbers
- "Annualised" is a rate, not a yearly total. $1.4 trillion is the pace of spending Goldman expects in the final part of 2027, multiplied to a full year. Total spending during 2027 would be lower than that if the pace rises through the year.
- Why investment adds little to GDP. Most information processing equipment is imported, so it is subtracted again when GDP is calculated. That is why $1.4 trillion of spending translates into 0.6 points of growth, according to the account of the note published by investingLive.
- Semiconductors are the swing factor. Goldman argues official business investment data undercount chip spending, and reports its figures both ways. Without semiconductors the GDP contribution falls to 0.2 points.
- Risks named. Reuters reported Goldman sees only limited risk to AI investment from tighter data center regulation after the November midterm elections. investingLive's account names rising borrowing costs as the main risk.
Goldman's Separate Hyperscaler Forecast
A different Goldman team, the equity strategists led by Ryan Hammond, published a company-level forecast in September 2026 that Bloomberg reported and PYMNTS summarised. It covers the capital spending of five companies: Amazon, Alphabet, Microsoft, Oracle and Meta.
| Year | Five hyperscalers' AI infrastructure spending | Note |
|---|---|---|
| 2026 | $800 billion | |
| 2027 | $1.2 trillion | Consensus at the time: $1.1 trillion |
| 2028 | $1.4 trillion | A 12% rise |
Two different Goldman figures are therefore both "$1.4 trillion": the economy-wide annual rate at the end of 2027, and five companies' spending in 2028. Secondary coverage often merges them, and some summaries place the hyperscaler figure in 2027. Goldman also has a published article titled "Global AI Investment Is Forecast to Exceed $1 Trillion in 2026". We could not open it and do not quote its figures.
Other Banks' Forecasts
| Bank | Date | Forecast | Scope | Source grade |
|---|---|---|---|---|
| Goldman Sachs, economists | Nov 11, 2026 | $1.4 trillion annual rate by end of 2027 | United States AI hardware and intellectual property | B, Reuters |
| Goldman Sachs, equity strategy | Sept 2026 | $1.2 trillion in 2027, $1.4 trillion in 2028 | Five hyperscalers | B, Bloomberg via PYMNTS |
| Morgan Stanley | Nov 9, 2026 | $1.6 trillion in 2027, $1.8 trillion in 2028, both unchanged | Hyperscaler capital spending. Companies not listed in the account we read | C, one secondary report |
| JPMorgan | June 2026 | $650 billion in 2026, more than $1.1 trillion in 2027 | Hyperscaler capital spending | B, Fortune |
| JPMorgan | June 2026 | $5.5 trillion through 2030, raised from $5.1 trillion | Global AI capital spending | B, Fortune |
| Citigroup | Sept 2025 | More than $2.8 trillion through 2029 | Large technology companies' AI infrastructure | C, as recalled by PYMNTS |
- The 2027 figures range from $1.1 trillion to $1.6 trillion. Most of the gap is scope and date. JPMorgan's figure is from June. Morgan Stanley's, by the account we read, is higher and was left unchanged in a note titled "Buy the Drop" after the midterms.
- Morgan Stanley also flagged delays. The same account says it expects 20% of United States data center projects planned for 2028 to slip to 2029.
- Financing is part of the forecast. Fortune reported JPMorgan expects about $150 billion of United States hyperscaler debt issuance in 2026.
For what venture investors put into AI companies over the same period, see AI funding rounds, November 2026 and OpenAI's November round. This page reports published forecasts. It is not investment advice and makes no forecast of its own.
Methodology
The November forecast is from Reuters' report as carried by The Star on November 12, 2026, with detail on imports and risks from investingLive. The hyperscaler forecast and the Citigroup figure are from PYMNTS' summary of Bloomberg. Morgan Stanley is from investingLive, and JPMorgan from Fortune, as carried by Yahoo Finance. We did not have access to any of the bank research notes themselves. All pages were opened on November 13, 2026.