Global Artificial Intelligence Spending Set to Surge to $1.4 Trillion

2026-11-12 semicon

Amsterdam, Thursday 12 November 2026
Defying market volatility, annual artificial intelligence investment will near four per cent of the US economy by 2027, driven by massive expansions in data centres and computer hardware.

Unprecedented Capex Expansion and Macroeconomic Drivers

Annual enterprise spending committed to artificial intelligence hardware and intellectual property is projected to climb to $1.4 trillion by the end of 2027, according to a macroeconomic forecast released by Goldman Sachs on Wednesday, 11 November 2026 [1]. This projected capital expenditure represents nearly 4% of the US economy [1]. The investment bank forecasts annualized outlays to eclipse $900 billion before the conclusion of 2026, underlining that artificial intelligence will remain the pre-eminent driver of overall business investment growth into 2027 [1]. The projected trajectory comes as macroeconomic forecasters assess capital allocation following the US midterm elections in early November 2026, where Democratic victories prompted regulatory questions concerning power distribution and data centre approvals [1][2]. Goldman Sachs noted that legislative and regulatory shifts pose only limited downside risks to the broad trajectory of AI investment [1].

Bridging Capital Commitments to GDP Growth

On a quarterly basis, Goldman Sachs projects US business investment growth to hit 8% in 2027, almost double the market consensus of 4.2% [1]. At the macroeconomic level, direct AI-related spending is forecast to lift aggregate capital expenditure growth by 5.3 percentage points and overall US gross domestic product (GDP) expansion by 0.6 percentage points in 2027 [1]. When stripping out semiconductor hardware investments, AI initiatives are calculated to boost capex growth by 4.3 percentage points and contribute 0.2 percentage points to US GDP [1]. The investment bank noted that domestic semiconductor investments remain undercounted in standard business expenditure statistics, while a significant proportion of advanced information processing systems is imported from global supply networks and thus deducted from headline domestic product calculations [1].

Strategic Infrastructure Constraints and Valuation Resets

The sustained momentum in capital investment persists in spite of tightening corporate credit conditions and public market pullbacks [1][2]. Goldman Sachs Group Chief Executive David Solomon remarked on Monday, 9 November 2026, that institutional demand for AI infrastructure financing remains resilient against broad market volatility and corrections across cloud computing and semiconductor equities [2][3]. The semiconductor index (SOX) endured a decadal relative drawdown of 30 points against the S&P 500, surpassing previous historic relative corrections of 20 points by 10 points [4]. Analysts at the firm view the valuation correction as an attractive entry point, while outlining fundamental bottlenecks: the expanding funding gap between hyperscaler operating cash flows and capex, and severe physical constraints in acquiring land, grid connectivity, and high-voltage transmission equipment [4].

Benelux Equipment and Power Capacity Demands

Physical data centre installations dedicated to AI compute are estimated to surge from 20 gigawatts in 2026 to 35 gigawatts in 2027, before reaching 57 gigawatts by 2028 [4]. The projected growth between 2026 and 2027 represents a planned capacity addition of 15 gigawatts [4]. Solomon cautioned that unless the United States rapidly expands its sustainable high-voltage power networks and permitting pipelines, critical computing capacity will disperse to overseas jurisdictions capable of fulfilling escalating operational demand [2]. This capital expenditure supercycle filters directly into specialized European supply chains, notably the Benelux semiconductor nexus spanning front-end lithography manufacturer ASML, atomic layer deposition equipment specialist ASM International, and deep-tech integrated photonic networks coordinated under ecosystems like PhotonDelta [GPT]. Upstream capital deployments by hyperscalers directly fund the extreme ultraviolet (EUV) tools and advanced silicon designs essential to deliver high-throughput, energy-efficient server clusters [GPT].

Productivity Divergence and European Strategic Autonomy

Downstream returns on compute infrastructure reveal sharp sector divergences [5]. Data from Morgan Stanley indicated that US non-farm productivity rose 3.6% annualized in the third quarter of 2026 as unit labour costs contracted by 0.7% [5]. However, approximately 83% of productivity gains—or 2.0 percentage points of a 2.4% baseline through the second quarter of 2026—were concentrated in just 12 of 64 tracked industries, led by a 27.0% efficiency leap in data processing and internet publishing [5]. The remaining 52 tracked sectors contributed a combined 0.4 percentage points [5]. Concurrently, hardware production efficiency showed volatility, with computer manufacturing productivity swinging from a 13.0% expansion in late 2025 to a 1.1% decline in the second quarter of 2026 [5]. As Goldman Sachs chief economist Jan Hatzius cautioned that the macroeconomic growth impulse from initial AI rollouts could crest around late 2026 and early 2027 [5], European industrial planners regard home-grown semiconductor tooling, wafer fabrication, and optical communication integration as vital elements to secure technological resilience and shield core digital infrastructure from international supply chokepoints [GPT].

Sources & Ecosystem Partners

Artificial intelligence Semiconductor investment