Industrial software names have been swept up in a wider market sell-off fueled by anxiety over AI-led disruption, but a Barclays analyst says investors are drawing the wrong conclusions about how value is created in enterprise software.
Guy Hardwick of Barclays challenges the narrative that AI will erode the economics of software-as-a-service businesses. In his view, the market is overstating the role of code in determining enterprise software value and understating the significance of services, domain knowledge and dependable operations - the capabilities customers actually pay for.
Hardwick writes that the "SaaS AI bear case is overdone," and that AI should be treated "as an incremental opportunity for our Industrial Software names rather than a threat." He adds that the recent AI-driven sell-off "is unfairly penalizing Industrial Software."
Central to his argument is the assertion that the idea AI will commoditize enterprise software rests on a mistaken premise: that customers primarily pay for code. He notes that even if AI makes code generation cheaper or more accessible, it serves to emphasize the value of service rather than eliminate it.
"While AI commoditizes code generation, it highlights the economic value of the service in SaaS. If code is a commodity, the value is the deep domain expertise, reliable service, and hybrid (deterministic + probabilistic) architectures," Hardwick wrote.
He stresses that "code is an input, not the product or output," and points out that established SaaS companies have long moved beyond selling code alone, competing instead on the basis of service delivery and industry-specific expertise.
Barclays estimates that coding costs account for only a small share of spending inside mature SaaS businesses - likely in the range of roughly 4% to 8% of revenue. Under that framework, the direct financial hit from automated coding is limited.
Despite that limited direct exposure, valuations across industrial software have tumbled. Barclays highlights that multiples for companies under its coverage have fallen about 50% over the past six months. Manhattan Associates (NASDAQ:MANH) is singled out, with shares down more than 40% from their 2025 peak.
Another dynamic Hardwick raises is the cost structure of running AI applications. Generative AI may lower development expenses, but operating AI at scale introduces substantial costs - particularly inference expenses tied to GPUs, energy consumption and infrastructure.
"Even if software can be generated cheaply or freely, it cannot be operated cheaply at scale," Hardwick wrote, arguing that these operating costs create "a hard economic price floor" that tends to favor established vendors able to manage those costs efficiently.
Barclays notes industrial software valuations have retraced to levels seen during the COVID era and are now at historical lows relative to the S&P 500. The firm also points out that the sector currently offers stronger free cash flow yields than many industrial tech hardware peers.
Looking ahead, Hardwick identifies several potential catalysts that could prompt a rotation back into industrial software. These include growing investor fatigue with AI hardware trades, the possibility of a reversal in semiconductor outperformance versus software, and elevated short interest across the sector. Each of these factors could support a rebound in sentiment toward industrial software shares.
Bottom line: Barclays argues the market has overshot in pricing AI risk into industrial software stocks by conflating the value of code with the full-service proposition SaaS firms deliver. While short-term valuations have compressed materially, structural elements such as services, domain expertise and the economics of operating AI systems remain key considerations.