NextFin News - Goldman Sachs Asset Management is doubling down on the "picks and shovels" of the artificial intelligence boom, forecasting a 65% surge in AI-related capital expenditure for 2026 as the industry shifts from experimental models to massive infrastructure deployment. The firm’s latest analysis suggests that while the initial wave of software excitement has cooled, the physical backbone of the AI economy—semiconductors, data centers, and power systems—is entering a period of sustained, high-intensity investment.
Brook Dane, co-head of public tech investing at Goldman Sachs Asset Management, argues that the market is currently in the early stages of a "profound and hugely impactful" technology transition. Dane, who has long maintained a constructive stance on the structural growth of the semiconductor sector, believes the current cycle differs from previous tech bubbles because the spending is backed by the largest balance sheets in corporate history. According to Dane, the focus is shifting toward companies that provide the essential hardware and infrastructure required to run increasingly complex large language models.
The projected 65% jump in capex represents a significant acceleration from previous years, reflecting a "build it and they will come" mentality among hyperscalers and enterprise tech giants. Goldman’s internal data indicates that this spending is increasingly concentrated in hardware providers, a trend that Dane suggests will create a "dispersion" in the market where infrastructure winners decouple from speculative software plays. This view, while influential, is not yet a universal consensus on Wall Street; some analysts at rival firms have expressed concern that such aggressive capex could lead to overcapacity if the ultimate return on investment for AI software fails to materialize by late 2026.
The "picks and shovels" strategy specifically highlights opportunities in hardware and select database software companies like Snowflake and MongoDB, which are seen as essential for managing the data deluge that AI requires. Dane’s position is that the volatility seen in early 2026 is a "cleansing process" that separates companies with real utility from those merely riding the AI narrative. He maintains that the state of current models provides enough confidence to justify the massive structural changes in capital allocation we are seeing today.
However, the path forward remains tethered to several critical assumptions, most notably the continued scaling of model performance and the availability of power. Goldman’s report acknowledges that the sheer scale of the $1 trillion generative AI investment cycle faces risks from energy grid constraints and potential regulatory headwinds under the current U.S. administration. If the "edge-use cases"—the specific industry applications of AI—do not begin to show clear revenue generation within the next four to six quarters, the current capex trajectory may face a sharp correction as investors demand more immediate proof of profitability.
Explore more exclusive insights at nextfin.ai.
