AI advancements create a "gap" between tech companies and solidify the threat to the software era.

The artificial intelligence boom has created an unprecedented disparity in the technology sector, shifting value creation to the physical infrastructure that supports AI models and pressuring software companies—including SaaS companies—to reassess their business models. This gap is reflected in the stock market performance of technology companies operating in these two areas.

This is according to a survey by Morningstar , an online data and investment platform, revealing that shares of major companies that supply the inputs for infrastructure—mainly semiconductors and memory—have seen an average increase of 90% this year. Meanwhile, software companies, whose business models and competitive advantages are seen as threatened by AI, are registering losses of up to 40%.

“Essentially, the market thesis is that artificial intelligence tends to shake up the software business model quite significantly,” says Dave Sekera, chief US market strategist at Morningstar. “In fact, many people fear that artificial intelligence will end up completely replacing many software platforms,” he adds, although he considers the current pessimism exaggerated.

Before the era of generative AI, software dominated technology investment, as companies relied on subscription-based products to manage everything from sales and human resources to forecasting and IT.

These software vendors generated recurring revenue and high profits, making software as a service — or SaaS, a distribution model in which subscription-based applications are made available over the internet — one of the most valued business models on Wall Street.

The advancement of artificial intelligence is reshaping this hierarchy, a trend captured by Morningstar. As of May 13th, the difference between the top 10% and the bottom 10% in technology returns reached 133 percentage points—the largest dispersion in years.

At the top of the ranking are almost exclusively hardware companies. Sandisk has accumulated a 511.4% increase in value this year, followed by Intel, with over 225%. On average, hardware stocks rose 92.5%, while semiconductors advanced 76.6%.

Meanwhile, the worst 10% in the technology sector fell by an average of 39.3%, registering their worst performance since 2022. Of the 10 worst-performing technology stocks in 2026, eight belong to the software sector. Among them is HubSpot, a software marketing and sales company, with a drop of approximately 50% in its returns since the beginning of 2026.

Structural bottleneck

According to experts, the movement reflects a structural bottleneck: generative AI requires massive amounts of processing power, memory, and storage, and the supply of these resources is growing more slowly than the demand.

Value capture, therefore, has shifted to the base of the technological pyramid — chips, servers, memory modules — and not to the software layers that have historically led the sector.

For British fund manager James Anderson – famous for identifying the potential of technology and innovation giants long before Wall Street, leading early investments in companies like Amazon, Tesla, Alibaba, ASML, Spotify, and Nvidia – the current cycle marks the end of the era in which software and internet platforms dominated the market.

According to him, the trillion-dollar investments in AI made by giants like Google, Meta, Amazon, and Microsoft are "imploding" their cash flows and transferring lasting value to a small group of hardware vendors.

“The near certainty of the 20-year dominance of exponential platforms is over,” Anderson wrote in an annual letter to investors at Lingotto Innovation Strategy, warning that there is no longer “an obvious path back to low initial investment and high cash flow.”

Morgan Stanley analysts estimate that data center owners will spend around $2 trillion between 2024 and 2027 to expand their AI-focused infrastructure, a volume that is likely to benefit companies like Nvidia, TSMC, and ASML, all cited by Anderson as structural winners in the new cycle.

It is in this context that software loses ground. The correction stems not only from irrational fear, but from three simultaneous pressures. The first is the compression of competitive advantage: AI models reduce barriers to entry in entire software categories, replacing differentiators previously based on automation, analytics, or user experience with native AI solutions integrated directly into operating systems and cloud platforms.

The second is the pressure on the SaaS model. As AI shifts value to infrastructure, corporate clients have begun to question why maintain multiple subscriptions if generative models can perform tasks previously distributed across various software programs. The third is the cost of adaptation: incorporating AI requires heavy investments in computing and engineering, putting pressure on margins that were historically high in the sector.

Despite this, there are signs that the market may have gone too far in penalizing. Microsoft, even with a 15.5% year-on-year decline, exceeded revenue expectations driven by AI, indicating that software companies capable of natively integrating generative models can preserve—and even expand—their relevance.

According to analysts, the sector is not being replaced, but is undergoing a transition that requires strategic repositioning and new monetization models.

The prevailing view among experts and investors like Anderson is that investors will need to be more selective than in previous cycles. The challenge now is to identify which software companies will be able to reinvent themselves before the new market logic takes hold.