In a recent analysis, analysts at Goldman Sachs have revealed that Wall Street’s debate over the potential slowdown in artificial intelligence (AI) spending among major tech firms—commonly referred to as “hyperscalers”—is likely premature. The firm suggests that these companies may continue their aggressive investments in AI technology, estimating that hyperscaler capital expenditure (capex) could reach approximately $1.1 trillion by 2027. This figure surpasses the Wall Street consensus prediction of about $920 billion, and in a more optimistic scenario, spending could soar to $1.4 trillion.
Goldman’s outlook hinges on the premise that demand for AI computing power is still in its nascent stages. The analysts project that token consumption—critical for AI operations—will expand twenty-fourfold by 2030, largely driven by the burgeoning rise of enterprise agents utilizing AI. This surge in token usage necessitates an increase in computing power, subsequently driving demand for essential infrastructure such as data centers, chips, networking equipment, and energy resources.
However, Goldman Sachs acknowledges some challenges. Rising input costs are expected to exert additional pressure on the total capex required to accommodate growing token consumption. In addition, various companies have raised concerns regarding the expenses tied to AI tools, which raises questions about whether the productivity gains achieved will outweigh the costs of operating these increasingly complex models. This situation reflects a larger discussion across corporate America, where firms are heavily investing in AI but struggle to demonstrate clear returns on those investments.
Support for Goldman’s optimistic projections comes directly from the cloud service sector. Both Google Cloud and Amazon Web Services reported a staggering combined backlog of $832 billion in contracts as of the first quarter, a significant increase from $358 billion just six months prior. Nevertheless, Goldman cautions that a balance between AI supply and demand may not materialize until at least the latter half of 2027, indicating that elevated spending levels could persist longer than investors might anticipate.
Historically, previous investment booms linked to transformative technologies, such as railroads, electrification, and automobiles, peaked at approximately 2% to 3% of GDP, suggesting that the ongoing AI-related investment—projected at about 1.5% of GDP by 2026—could still have room for significant growth.
Despite this bullish outlook, Goldman Sachs emphasizes that financing is not likely to be the primary barrier to achieving future spending targets. Instead, physical constraints may present a more substantial challenge. Analysts pointed to numerous delayed data center projects, along with limitations surrounding memory, power, and labor resources as potential bottlenecks impacting capex growth.
Amidst this anticipated surge in spending, Goldman noted that this would likely bolster earnings for companies involved in AI infrastructure, including semiconductor manufacturers and power suppliers. However, it warned that the investment landscape is becoming increasingly crowded, with valuations in the AI infrastructure segment rising swiftly in recent months. This rapid escalation in share prices has led to a disconnect with earnings projections, creating potential risks for market volatility.
Moreover, while many companies have expressed enthusiasm about the productivity benefits associated with AI—54% of firms reported discussing AI during first-quarter earnings calls—only a small fraction quantified these benefits, and an even tinier percentage linked them to earnings impacts.
This analysis comes during a period of uncertainty in the tech sector, characterized by a recent sell-off attributed to geopolitical tensions and a complex interest rate environment. The Nasdaq 100 index experienced a 2% decline on Wednesday and has decreased by 6% since the beginning of the sell-off, although it remains 13% higher year-to-date. Despite these fluctuations, Nasdaq futures displayed a slight uptick early Thursday, suggesting a potentially cautious rebound.
