AI is driving one of the most significant shifts in capital allocation in decades, according to MFS Investment Management, creating a growing divide between companies supplying the physical infrastructure behind AI and those perceived to be vulnerable to disruption.
In a new strategist’s report, portfolio manager and global investment strategist Robert Almeida said the AI boom is redirecting capital away from financial engineering and towards the real-world assets needed to power increasingly sophisticated computing systems.
While AI is often viewed as a software-driven revolution, Almeida argued that every model, query and data centre ultimately relies on physical inputs including electricity, semiconductors, memory, copper, steel, concrete, cooling systems, skilled labour and logistics networks.
Decades of globalisation and offshoring have left the United States with less spare industrial capacity than in previous eras, he said, creating supply constraints that are beginning to influence corporate earnings and broader economic conditions.
According to MFS, evidence of those constraints can already be seen in the pricing power enjoyed by some hardware businesses and the persistence of inflationary pressures.
“The result is a capital-cycle shift — away from financial engineering and toward the physical economy needed to deliver compute,” Almeida said.
That shift has created clear winners across equity markets.
MFS said companies operating closest to critical infrastructure bottlenecks, including hardware manufacturers, computing providers, memory producers and power equipment suppliers, have outperformed as demand has outstripped available supply.
With customers facing limited alternatives in many areas, those businesses have been able to increase prices and generate stronger profits.
The firm cautioned, however, that investors should be wary of extrapolating current conditions indefinitely.
“High prices attract capital. Capital builds supply. Supply changes pricing power, returns and valuations,” Almeida said. “While today’s winners may keep growing, investors should be careful about assuming scarcity lasts forever.”
MFS argued that the current environment resembles previous capital-intensive investment cycles, where periods of exceptional profitability ultimately encouraged additional investment and increased competition.
While AI infrastructure spending continues to accelerate, the report suggested that elevated returns in some segments may eventually normalise as new capacity comes online.
At the same time, the firm believes markets may be taking an overly broad approach when assessing the impact of AI on software and information-based businesses.
Companies involved in software, data, information services and workflow management have faced significant share price pressure despite continuing to meet earnings expectations and showing little evidence of revenue deterioration.
According to Almeida, investors have increasingly treated AI disruption as an industry-wide threat rather than distinguishing between business models.
“The market has treated AI disruption as if the verdict were simple: software bad, compute good, disruption everywhere. We think that’s too blunt,” he said.
While MFS acknowledged that AI is likely to commoditise some repetitive software functions and low-value information services, it argued that businesses built around proprietary datasets, embedded customer workflows, regulatory compliance systems and mission-critical records could prove far more resilient.
The report suggested that these businesses derive much of their value not from software code alone but from the trust, governance, historical records and accountability embedded within their platforms. In some cases, AI could strengthen rather than weaken those competitive advantages.
MFS also pointed to Jevons’ Paradox, the theory that lower costs can increase overall consumption, as a framework for understanding how AI may affect demand for information and decision-making tools.
Rather than reducing the need for analysis, cheaper AI-generated content could increase demand for trusted answers, workflows and decision support.
As generic information becomes more abundant, the ability to verify accuracy, maintain compliance and provide auditable outcomes may become increasingly valuable, the firm said.
“In an AI-enabled world, the bottleneck may not be producing an answer. The bottleneck may be knowing whether the answer is accurate, compliant, auditable and fit for purpose,” Almeida said.
MFS stated the growing divergence between AI infrastructure beneficiaries and businesses perceived as potential disruptors highlights the importance of distinguishing between genuine threats and market overreactions.
“The AI capital cycle is already splitting the market. Hardware scarcity is real, but capital cycles are finite. AI disruption is real, but it isn’t universal,” Almeida said. “The task is to separate temporary scarcity from durable economics, and true disruption from market overreaction.”
