By Lina Irawan 
At the heart of this paradigm shift is the generative AI (GenAI) wave, which, as Lisa Shalett, chief investment officer for Morgan Stanley Wealth Management, notes, is "not obviously consumer-centric yet." Unlike its predecessors that rapidly spawned ubiquitous consumer products and services, the current AI build-out is deeply entrenched in the physical world. It demands an unprecedented scale of computational power, necessitating a vast expansion of foundational infrastructure. This translates into trillions of dollars of investment that will ripple through tangible markets, directly impacting sectors such as real estate, construction, power and electricity generation, and industrial metals. The firm argues that this dynamic is catalyzing a multiyear period where "investment dominates consumption as the growth driver amid economic rebalancing."
The sheer scale of this infrastructure demand is staggering. Morgan Stanley’s team highlights that data center-related investment alone is projected to account for a remarkable 25% of annual GDP growth in 2025. This expansion rate is multiples of forecasted real GDP growth, underscoring the immense capital allocation flowing into foundational AI capabilities. To put this into perspective, previous tech booms saw investment primarily in software, services, and light hardware, with marginal costs often trending towards zero. GenAI, however, requires massive, dedicated physical infrastructure to train increasingly complex large language models (LLMs) and perform inference at scale. This involves building colossal data centers, each consuming as much electricity as a small city, and equipping them with cutting-edge, energy-intensive hardware.
The implications for traditional industrial sectors are profound. The demand for new data centers is driving a boom in specialized real estate development, particularly in regions with ample land, affordable power, and robust fiber optic networks. Areas like Northern Virginia, Arizona, and parts of Texas are experiencing unprecedented growth in data center construction, leading to skyrocketing land values and specialized zoning changes. This, in turn, fuels the construction industry, requiring not just general contractors but highly specialized firms capable of designing and building facilities with advanced cooling systems, redundant power supplies, and stringent security protocols. The sheer volume of materials needed, from concrete and steel for structures to miles of fiber optic cabling and specialized HVAC systems, is immense.
Perhaps the most significant impact is on power and electricity generation. The energy appetite of modern data centers is voracious. Training a single large AI model can consume as much energy as several homes for a year, and the continuous operation of thousands of servers running inference operations consumes power 24/7. This necessitates substantial investment in new power generation capacity—whether through natural gas, nuclear, or renewable sources like solar and wind—as well as significant upgrades to existing electrical grids to ensure stable and reliable delivery. The strain on grid infrastructure is a growing concern, raising questions about energy security, environmental sustainability, and the potential for increased electricity costs for consumers. Furthermore, the demand for industrial metals like copper (for wiring, cooling systems, and electrical components), aluminum (for server racks and structural elements), and even rare earth elements (critical for advanced microprocessors and cooling technologies) is escalating, impacting global supply chains and commodity markets. This massive re-orientation of capital towards physical assets signifies a tangible "reindustrialization" of the American economy, a stark contrast to the largely digital and services-oriented growth of recent decades.
While this infrastructure build-out presents a boon for industrial metrics and a select group of specialized workers, the outlook for the broader human labor force is, as Morgan Stanley warns, "markedly less rosy." The report highlights "transformational risks to the labor market" brought on by the rapid diffusion of GenAI technologies. This isn’t merely about automation replacing manual labor; AI is increasingly capable of performing cognitive tasks previously thought to be exclusive to highly skilled professionals. Roles in coding, customer service, data analysis, content creation, and even some aspects of legal and medical professions are now susceptible to significant disruption. While new jobs will undoubtedly emerge in AI development, maintenance, and data center operations, there is a critical concern about a potential skills mismatch and the speed at which displacement could occur, leaving many workers behind.
Morgan Stanley describes prospects for the U.S. consumer as ultimately "unremarkable," weighed down by a confluence of negative factors. These include "depressed sentiment, job anxiety, a low 3.6% savings rate, and rising indebtedness and credit delinquencies." The pervasive fear of job displacement due to AI, coupled with persistent inflation and stagnant real wages for many, contributes to this widespread anxiety. A low savings rate leaves households vulnerable to economic shocks, while increasing credit card debt and delinquencies signal growing financial stress. The firm predicts that consumption growth, traditionally a pillar of the American economy, will likely stall due to a lackluster job market, the inexorable trend of aging demographics, and slow population growth.
These factors combine to trap the populace within "K-shaped economic dynamics," a meme that has leaped from finance Twitter into a harsh reality over the last five years. Unlike a "V-shaped" or "U-shaped" recovery where most segments of the economy rebound, a K-shaped recovery describes a scenario where different parts of the economy recover, or even thrive, at different rates. In the context of AI, this means the wealthy and highly skilled, who own AI companies, develop the technology, or possess skills complemented by AI, represent the upward-sloping arm of the "K." Their wealth and opportunities expand, driven by investment returns and high demand for specialized expertise. Conversely, the working class and those whose jobs are automated or whose skills become obsolete form the downward-sloping arm. They face increased economic insecurity, wage stagnation, and a diminishing share of the economic pie, exacerbating existing inequalities and potentially leading to social and political unrest. The AI revolution, in its current form, threatens to widen this chasm further, benefiting capital and advanced technology more than broad-based human labor.
Interestingly, this new paradigm is also forcing a harsh reality check on the tech titans themselves. For years, U.S. indexes have been dominated by "asset-lite, recurring-revenue tech business models" that enjoyed near-zero marginal costs and seemingly ever-expanding profit margins. Companies like Google, Meta, and Amazon built empires on platforms and software, where adding a new user or a new service often incurred minimal additional cost, allowing for immense scalability and profitability. However, the GenAI revolution is fundamentally different. It is a "cash-hungry R&D arms race" with distinct marginal-cost economics. As tech companies add subscribers to AI services or deploy more advanced models, they must simultaneously spend vastly more on precious "compute" capacity – meaning more GPUs, more power, more cooling, and more physical infrastructure.
Consequently, these former asset-lite darlings are transforming into "capital-intensive, cash-flow-hungry businesses." Developing, training, and running cutting-edge AI models requires staggering upfront investment in specialized hardware (like NVIDIA’s H100 GPUs), vast energy resources, and highly specialized talent. This shift fundamentally alters their financial profiles. Morgan Stanley bluntly states that for these hyper-scalers, "the era of multiple expansion based on seemingly ever-expanding profit margins is likely over." "Multiple expansion" refers to investors being willing to pay a higher multiple of earnings or revenue for companies with rapid growth and low capital requirements. With the AI era demanding massive capital expenditure, growth becomes more expensive, potentially leading to a re-evaluation of these companies’ valuations and profit expectations.
This sentiment is echoed by other leading financial strategists. Bank of America Research chief equity strategist Savita Subramanian has sounded similar alarms about tech’s move away from an asset-lite model, highlighting the increasing capital intensity. Moreover, even Silicon Valley executives are waking up to this new reality. OpenAI CEO Sam Altman, a central figure in the GenAI boom, has openly spoken about the immense costs associated with developing and running advanced AI models, even expressing personal anxiety about feeling "useless" when confronted with the pace of technological advancement. There’s a growing realization that the "tech industry’s profits gravy train," characterized by easy, high-margin growth, may be ending. The automation of most coding work, a cornerstone skill in the tech industry, further underscores the disruptive potential of AI, even within its own ecosystem, creating an existential challenge for a significant segment of the tech workforce.
Ultimately, Morgan Stanley’s vision of 2026 and beyond is one of profound economic realignment. The GenAI revolution may not be delivering a consumer utopia, at least not in the immediate term. Instead, it is fueling a global, capital expenditure-driven infrastructure boom that re-emphasizes the physical world. It is an era in which heavy machinery, robust power grids, vast data centers, and sophisticated semiconductor manufacturing reign supreme. This fundamentally suggests that, at least for now, the AI boom is far better for computers, their supporting infrastructure, and the industries that build and power them, than it is for the broad swaths of human workers who face job insecurity and stagnating economic prospects. The long-term implications for societal structure, wealth distribution, and the very definition of work remain open questions, but the immediate trajectory points towards an investment-led transformation with significant human costs.