Goldman: AI Spend Hits $1.4T by 2027, Only 0.6pt of GDP

Goldman Sachs expects annualised spending on AI-related hardware and intellectual property to reach $1.4 trillion by the end of 2027, an annual pace equal to nearly 4% of the US economy. The bank's forecast sits well above consensus, and it expects US business investment to grow about 8% in 2027 against a consensus of around 4%. Cisco shares traded at $102.01, up 3.93% on the day, within a range of $99.41 to $103.18 as of 23:32 UTC today.

Context — why the $1.4 trillion AI spending forecast matters now

Goldman's projection compares with its own estimate of more than $900 billion in annualised AI-related hardware and IP spending by the end of 2026. The step from roughly $900 billion to $1.4 trillion over twelve months is the core of the bank's above-consensus capex call, and it is the reason the forecast has become the reference point for the AI infrastructure trade.

The forecast lands in a week when three companies reported results that support the spending path. Cisco raised its full-year outlook on strong AI networking demand from hyperscalers. Applied Materials guided above forecasts on chip equipment orders. Cerebras reported a backlog of around $26 billion and said it sees revenue tripling in 2027.

The macro backdrop matters for how much of that spending reaches US output. Goldman estimates AI investment will add more than 5 percentage points to US capex growth in 2027 once semiconductors are included, but only 0.6 percentage point to GDP growth, and 0.2 point without semiconductors. The gap is a trade story, not a demand story.

What changed to make the forecast actionable is the political read. Goldman said Democrats' victories in this month's midterm elections pose only limited risks to AI investment, easing a concern about tighter data centre regulation that had weighed on the sector.

Data — what the numbers show

The headline figure is the $1.4 trillion annualised pace by the end of 2027. Against it, Goldman's own 2026 endpoint of more than $900 billion frames the trajectory: the bank sees roughly a 55% increase in the annual run-rate over that window.

Metric	Goldman forecast
AI hardware and IP spend, end-2027	$1.4 trillion annualised
AI hardware and IP spend, end-2026	more than $900 billion
Share of US economy	nearly 4%
US business investment growth, 2027	about 8% (consensus around 4%)
AI contribution to 2027 capex growth	more than 5 percentage points
AI contribution to 2027 GDP growth	0.6 point with chips, 0.2 point without

The GDP line is where the forecast splits from the trade. A 0.6-point contribution to growth, and 0.2 point excluding semiconductors, is far smaller than the spending total implies, because most information processing equipment is imported and therefore netted out of GDP.

Cisco's session offers a live read on the networking side of the trade. The stock traded between $99.41 and $103.18, with the $102.01 print up 3.93% and sitting near the upper end of that range.

Analysis — what it means for markets, sectors and tickers

The forecast favours the suppliers that capture AI capital spending rather than the economy that hosts it. Chipmakers, chip equipment makers and networking suppliers get the most from a spending path above consensus, because their revenue is booked where the hardware is built, not where it is installed.

For macro traders, the import share is the operative detail. Strong capex does not translate into equally strong US GDP growth. The difference shows up in a wider trade deficit and in demand for Asian supply chains, which absorb the manufacturing and assembly work behind the $1.4 trillion figure. That split means an AI capex beat can coexist with a soft US growth print.

The regulatory read is a second-order positive. Goldman's view that the midterm outcome poses only limited risk to AI investment removes one overhang from data centre-linked names. It does not remove the funding constraint.

Funding costs remain the main pressure point. Rising borrowing costs and tighter investor scrutiny could slow AI capital spending, particularly as companies lean more on debt to pay for the build-out. With markets pricing a strong chance of another Fed hike in December, the cost of that debt is the variable to watch. Positioning follows the same logic: investors are long the hardware supply chain and cautious on debt-funded developers, and flow continues to concentrate in chip and networking names rather than the broader capex complex.

Outlook — what to watch next

The December Fed meeting is the first catalyst. Markets price a strong chance of another hike, and a higher policy rate lifts the cost of the debt financing part of the AI build-out. Goldman flags that channel as the bigger risk to its own forecast, ahead of regulation.

The second catalyst is the earnings calendar for the suppliers named in the forecast. Cisco, Applied Materials and Cerebras have already reported, and each guided or backlogged in a way that supports the spending path. Follow-through from chip equipment orders and hyperscaler networking demand is what would confirm the $900 billion to $1.4 trillion step.

The third is the trade data. If AI hardware imports keep rising, the gap between capex strength and GDP contribution widens, and the trade deficit reflects it. Watch Cisco's $99.41-$103.18 range as a live gauge of how the networking leg is trading while those prints land.

Frequently Asked Questions

What does Goldman's $1.4 trillion AI spending forecast mean for retail investors?

It describes where corporate capital is going, not what any single stock will do. Goldman expects annualised AI hardware and IP spending to reach $1.4 trillion by end-2027, up from more than $900 billion by end-2026. Chipmakers, chip equipment makers and networking suppliers are the groups the bank says capture most of that spending. The forecast is informational and is not investment advice.

Why does $1.4 trillion in AI spending add only 0.6 point to US GDP growth?

Because most information processing equipment is imported. Goldman estimates AI investment adds 0.6 percentage point to 2027 GDP growth including semiconductors, and 0.2 point without them, while adding more than 5 percentage points to capex growth. Imported hardware is netted out of GDP, so the spending shows up in the trade deficit and in Asian supply chain demand instead.

What is the biggest risk to Goldman's AI capex forecast?

Funding costs. Goldman points to rising borrowing costs and tighter investor scrutiny as the main pressure point, particularly as companies rely more on debt for the build-out. Markets price a strong chance of another Fed hike in December. The bank separately said the midterm election outcome poses only limited regulatory risk to AI investment.

Bottom Line

Goldman's $1.4 trillion AI capex call is a supplier trade, not a US growth trade, because the hardware is imported.

Disclaimer: This article is for informational purposes only and does not constitute investment advice. CFD trading carries high risk of capital loss.