Goldman Sachs: AI Shifts Focus to Data Centers

A New Direction for AI Investments

The world of investment in artificial intelligence is currently undergoing a significant transformation. Companies and investors are redirecting their attention towards data center infrastructure, which is essential for the proper functioning of AI systems. This trend is highlighted by a recent analysis from Goldman Sachs, which describes this phenomenon as a "flight to quality." Specifically, this means that investors are focusing more on companies that own and operate large data centers and robust computing infrastructures. In contrast, companies offering more specialized AI tools or software still in experimental phases are attracting less attention.

Goldman Sachs forecasts a rapid increase in spending on AI infrastructure. This rise is driven by the growing need for companies to expand their computing capacity for training and deploying models. Hyperscale cloud giants invest tens of billions of dollars each year in new data centers and hardware. At the same time, networking systems are expanding to support this growth.

The Impact of AI Demand on the Data Center Market

Research conducted by Goldman Sachs estimates that, within the next two years, AI-related workloads could account for approximately 30% of the total capacity of data centers. This increase is due to the rising demand for computing power, both in cloud services and enterprise applications. This shift illustrates how AI tasks are distinct from traditional cloud workloads. Indeed, training large models requires the simultaneous use of thousands of chips over extended periods. Additionally, inference, which involves generating responses or predictions, also requires constant computing power when services are operational.

Cloud providers and AI developers are now increasing data center capacity at an unprecedented pace compared to earlier phases of cloud computing. The expansion of infrastructure is not limited to hardware. Energy supply is becoming a central issue in the race for AI.

According to estimates from Goldman Sachs Research, global energy demand from data centers could grow by about 175% by 2030 compared to 2023 levels. This increase is primarily attributed to AI-related workloads. The firm emphasizes that this growth would be comparable to adding the electrical demand of another country among the top ten energy consumers to the global grid. This rising energy demand is also prompting utilities and governments to consider new investments in energy infrastructure.

Infrastructure Challenges Shape AI Strategy

The growing need for energy and cooling is influencing the location of new data centers for AI. Space requirements also play a role in site selection. Large facilities are often located near stable energy sources and high-capacity fiber networks. Some companies choose to build AI training clusters in remote areas where land and electricity are more accessible. The location of data centers can also impact the environment. Academic research on AI infrastructure shows that cooling systems and geographical location can influence energy and water consumption as much as hardware efficiency.

These constraints are beginning to affect how tech companies plan their AI strategies. Building new models or software is only part of the challenge. Companies must also ensure they have the necessary infrastructure to operate these systems reliably. In many cases, constructing this infrastructure takes years.

Building large data centers involves complex supply chains. Projects often require land acquisition and connections to the power grid. Many also depend on long-term energy agreements. Shortages of electrical equipment and delays in grid expansion can slow down new projects. These constraints help explain why investors are paying increased attention to companies that already control large networks of data centers.

A Selective Phase in the AI Market

During the first wave of generative AI adoption, many companies saw their market value increase simply by associating with AI. This phase is now beginning to change as investors reassess where AI growth will occur.

Investors are examining which companies have the infrastructure and revenue models necessary to support long-term deployment. Data center operators and chip manufacturers are positioned near the base of this ecosystem. Their services are required, regardless of the type of AI application that gains traction.

In previous waves of computing growth, companies that built the underlying infrastructure often captured stable revenues. Software platforms, on the other hand, experienced more rapid fluctuations. A similar dynamic could now be forming in the AI sector.

The expansion of infrastructure also raises new questions. Energy demand and grid capacity are becoming central issues for governments and industrial planners. The environmental impact is also under closer scrutiny.

In the coming years, the AI economy could depend as much on power plants and cooling systems as on algorithms and software. This reality is shaping the next stage of the AI race.